Toward designed singlet fission: Solution photophysics of two indirectly coupled covalent dimers of 1,3-diphenylisobenzofuran

Article

pubs.acs.org/JPCB

Toward Designed Singlet Fission: Solution Photophysics of Two

Indirectly Coupled Covalent Dimers of 1,3-Diphenylisobenzofuran

Justin C. Johnson,^†Akin Akdag,^‡,§Matibur Zamadar,^∥Xudong Chen,^‡Andrew F. Schwerin,^‡Irina Paci,^#

Millicent B. Smith,^‡Zdenek Havlas,^⊥John R. Miller,^∥Mark A. Ratner,^#Arthur J. Nozik,^†

̌

and Josef Michl*^,‡,⊥

^†National Renewable Energy Laboratory, Golden, Colorado 80401, United States

^‡Department of Chemistry and Biochemistry, University of Colorado at Boulder, Boulder, Colorado 80309, United States

^§Department of Chemistry, Middle East Technical University, 06800 Ankara, Turkey

^∥Chemistry Department, Brookhaven National Laboratory, Building 555A, Upton, New York 11973, United States

^⊥Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, 16610 Prague, Czech Republic

^#Department of Chemistry and Materials Research Center, Northwestern University, Evanston, Illinois 60208, United States

S

* Supporting Information

ABSTRACT: In order to identify optimal conditions for

singlet ﬁssion, we are examining the photophysics of 1,3-

diphenylisobenzofuran (1) dimers covalently coupled in

various ways. In the two dimers studied presently, the coupling

is weak. The subunits are linked via the para position of one of

the phenyl substituents, in one case (2) through a CH₂linker

and in the other (3) directly, but with methyl substituents in

ortho positions forcing a nearly perpendicular twist between

the two joint phenyl rings. The measurements are accom-

panied with density functional theory (DFT) and time-

dependent DFT (TD-DFT) calculations. Although in neat

solid state, 1 undergoes singlet ﬁssion with a rate constant higher than 10¹¹s⁻¹; in nonpolar solutions of 2 and 3, the triplet

formation rate constant is less than 10⁶s⁻¹and ﬂuorescence is the only signiﬁcant event following electronic excitation. In polar

solvents, ﬂuorescence is weaker because the initial excited singlet state S₁equilibrates by sub-nanosecond charge transfer with a

nonemissive dipolar species in which a radical cation of 1 is attached to a radical anion of 1. Most of this charge transfer species

decays to S₀, and some is converted into triplet T₁with a rate constant near 10⁸s⁻¹. Experimental uncertainties prevent an

accurate determination of the number of T₁excitations that result when a single S₁excitation changes into triplet excitation. It

would be one if the charge-transfer species undergoes ordinary intersystem crossing and two if it undergoes the second step of

two-step singlet ﬁssion. The triplet yield maximizes below room temperature to a value of roughly 9% for 3 and 4% for 2. Above

∼360 K, some of the S₁molecules of 3 are converted into an isomeric charge-transfer species with a shorter lifetime, possibly

with a twisted intramolecular charge transfer (TICT) structure. This is not observed in 2.

concerns, such as redox potentials, independent charge

separation from the two triplets, and so on.

INTRODUCTION

■

The use of singlet ﬁssion¹represents a promising avenue for

improving the eﬃciency of solar cells.²In this process, an

excited singlet state of a chromophore shares its energy with a

ground state neighbor, producing a triplet excited state of each.

If both triplet excitations behaved independently and could be

used for free charge carrier generation in a device utilizing both

conventional and singlet ﬁssion chromophores, the maximum

theoretical eﬃciency would reach ∼46%³instead of the usual

Shockley−Queisser limit of 32%⁴for a single-junction cell.

For practical applications, it would be very useful to have a

wide selection of eﬃcient systems. As discussed in more detail

elsewhere,^1,5,6for this purpose it is desirable to optimize (i) the

choice of the monomeric chromophore, and (ii) the eﬃciency

of interchromophore coupling, before addressing other

Choice of Chromophore. Although singlet ﬁssion was

observed on a number of crystalline and polymeric

chromophores,¹the measured triplet quantum yields were

mostly very small and of little practical interest. Although it has

been clear for some time that the longer polyacenes and some

carotenes produce triplets eﬃciently, only recently have triplet

yield measurements been made and values well in excess of

100% and mostly near the theoretical limit of 200% were

Special Issue: Paul F. Barbara Memorial Issue

Received: November 6, 2012

Revised: January 31, 2013

Published: February 5, 2013

4680

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

actually found for solid tetracene,⁷pentacene,^8,9and a handful

of their simply substituted derivatives,¹⁰⁻¹³as well as for

aggregated zeaxanthin,¹⁴a carotenoid. Another solid for which

a 200% triplet yield was found is 1,3-diphenylisobenzofuran

(1).¹⁵

coupling matrix elements, but also indirectly, by modifying the

energies of the states of the isolated monomers and thus the

exoergicity of the singlet ﬁssion process. The eﬀect of coupling

on the ﬁnal energy levels in the dimer is thus capable of ruining

the validity of the initially designed relations E(S₁), E(T₂) ≥ 2

E(T₁).

In the ﬁrst two full papers of this series,^5,16we argued that in

order to make singlet ﬁssion itself slightly exothermic and to

make all important modes of T₁−T₁annihilation endothermic,

one should choose chromophores for which not only the

energy of the relaxed S₁and T₁states, but also that of the T₂

state are appropriate and meet the conditions E(S₁), E(T₂) ≥ 2

E(T₁). A high yield of ﬂuorescence was identiﬁed as another

desirable property for the isolated chromophore, since it

indicates the absence of competing decay mechanisms such as

photochemical transformations or intersystem crossing. Several

likely candidates were identiﬁed by theoretical means,⁵and 1,3-

diphenylisobenzofuran (1) was selected as a suitable model for

an initial investigation. Its low-lying singlet, triplet, and ionized

states were characterized thoroughly,¹⁶and satisfy the desired

criteria. In a solid ﬁlm at 77 K, 1 was found to produce a 200%

yield of triplets. This compound is the ﬁrst molecule speciﬁcally

designed for eﬃciency in singlet ﬁssion.

The third full paper in this series²²used simple molecular

orbital (MO) theory to examine the eﬀect of structure on the

strength of interchromophore coupling in a series of dimers of

three chromophores that are possible candidates for singlet

ﬁssion studies, and ordered them from weakly to strongly

coupled. The coupling strength was estimated very roughly

from the splitting of frontier orbital and low-lying state energies

when the two chromophores were combined. A more direct

procedure would be to evaluate the action of the total

electrostatic Hamiltonian on the initial state.¹Because of the

complicated spin dynamics that may occur in singlet ﬁssion, the

ultimate triplet yield in a dimer could depend on other aspects

of its structure as well, such as the relation of the spin−spin

dipolar tensors of the two halves, and the splitting of the ¹(TT)

singlet, ³(TT) triplet, and ⁵(TT) quintet levels that result from

the interaction of two local triplets.

In the present paper we examine two covalent dimers of 1,

coupled only very weakly in order to preserve the identity of

their two chromophores. Covalent dimers can be synthesized

with a high degree of precise and systematic control of

interchromophore coupling, diﬃcult to achieve in crystals of

monomers. They are of interest both for fundamental studies of

coupling optimization and for potential practical implementa-

tions in which they could be adsorbed as individual units on

semiconductor support.

Presently, we have selected linear chromophore coupling,

one of the modes that have received theoretical attention, and

examine the dimers 2 and 3.

Singlet Fission Mechanisms. There are at least two ways

in which singlet ﬁssion mechanism can be classiﬁed, and in

both, simple limiting cases can be envisaged.

(A) Coherent versus incoherent process. (i) If the

chromophores are coupled strongly enough and E(S₁)

approximately equals 2E(T₁), the process could occur

coherently on a femtosecond time scale and be faster than

vibrational relaxation, as appears to be the case in solid

pentacene.²³Some theoretical aspects of this possibility have

been examined.^24,25(ii) If the coupling of the chromophores is

weak and/or E(S₁) lies too far below 2E(T₁), vibrational

relaxation will occur ﬁrst, and singlet ﬁssion will be relatively

slow (ps time scale), as appears to be the case in polycrystalline

1.¹⁵It will then be best understood in terms of motion on the

potential energy hypersurfaces that are encountered on the way

from the geometry of the initial vertically excited singlet state to

that of the ﬁnal double-triplet excited singlet state, with their

minima, barriers, and conical intersections.

Choice of Interchromophore Coupling. Little is known

about the degree of interchromophore coupling needed to

make singlet ﬁssion competitive with ﬂuorescence and yet allow

the two triplets, born coupled to an overall singlet, to behave as

independent excitations, each capable of free charge carrier

generation. It is not even known whether in a covalent dimer

these two requirements are not contradictory. Data for

covalently linked tetracene pairs^17,18and initial results for

covalently linked pairs of 1¹⁹indicate a triplet yield of only a

few percent, although in neat solid tetracene and 1, the yield is

close to 200%. The origin of this diﬀerence is not clear. It is

possible that the crystal environment provides a more favorable

interchromophore coupling, by positioning the molecules

favorably and/or by stabilizing low-energy charge-transfer

states that can serve as virtual states in the coupling process.

The inability of the two triplets in the dimer to diﬀuse apart

may be signiﬁcant as well. It is also possible that the

delocalization of the initial excited singlet state in the crystal

is helpful in itself, even though the matrix elements between the

singlet coupled triplet pair and either the localized or the

delocalized initial singlet wave function are known¹to be nearly

the same. Singlet excitation is often delocalized in crystals and

aggregates (e.g., J-aggregates²⁰), whereas in covalent dimers and

oligomers, excitation localization appears to be the rule, albeit

with very small energy barriers for moving from one to the

other excited minimum (e.g., in triptycene, ∼0.15 kcal/mol²¹),

and with exceptions (e.g., excimers).

(B) Direct versus mediated process.¹This mechanistic

distinction is less well deﬁned in that it depends on the degree

of sophistication chosen for the description of the initial and

ﬁnal states. The least elaborate choice is to describe all states in

terms of conﬁgurations based on orbitals localized on one or

the other partner and assumed to be orthogonal. Then, (i) in

the ﬁrst approximation, one describes the initial and ﬁnal states

of the process in terms of excitations strictly localized on a

single partner, and the matrix element between them is due to

the two-electron part of the total Hamiltonian (the direct

contribution to the singlet ﬁssion process). (ii) In a better

approximation, one allows the initial and ﬁnal states to contain

small admixtures of charge-transfer conﬁgurations before

evaluating the coupling matrix element between them. This

introduces additional terms that result in an indirect coupling of

the initial and ﬁnal strictly locally excited states through the

intermediacy of virtual intermediate states of charge-transfer

nature. In these terms, the one-electron part of the Hamiltonian

plays a dominant role (the mediated contribution to singlet

ﬁssion). In one imaginable limit, the direct contribution would

The interchromophore interaction can range from mere

physical contact through indirect covalent coupling via

interposed structures to direct covalent bonding that permits

strong π-electron conjugation. Strong coupling may not only

aﬀect the rate of singlet ﬁssion directly, by modulating the

4681

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

dominate, and in the other, the mediated contribution would

prevail. The relative importance of the mediated contribution is

a function not only of the magnitude of the oﬀ-diagonal

elements in the Hamiltonian matrix written in the strictly

localized basis, but also of the diﬀerences between diagonal

elements (the energies of the charge-transfer states relative to

the initial and ﬁnal states). We have already pointed out above

that environmental stabilization of charge-transfer states may be

the primary factor that makes singlet ﬁssion more eﬃcient in

solids than in dimers. In an extreme case, one can imagine that

in a highly polar solvent or in a highly polarizable crystal, some

of the charge-transfer states are stabilized so much that they

become the lowest singlet state somewhere along the reaction

coordinate and then represent real rather than virtual

intermediates. The singlet ﬁssion process would then occur

not in a single kinetic step, but in two steps, via an observable

charge-transfer intermediate.

EXPERIMENTAL SECTION

■

Synthetic Procedures. The syntheses and puriﬁcation of

the covalent dimers 2 and 3 are described in the Supporting

Information.

Measurements. Solvents (dimethyl sulfoxide, DMSO;

acetonitrile, AN; tetrahydrofuran, THF; cyclohexane, CH;

toluene, TOL; N,N-dimethylformamide, DMF) were purchased

from Aldrich (highest purity) and were used as received. They

were ﬂushed with argon for at least 30 min before transfer to a

glovebox. Sample cells were sealed in the glovebox before being

taken into ambient conditions for use. Since 1 is photolabile in

the presence of oxygen,²⁶2 and 3 were expected to exhibit a

similar instability. The procedures used for measurements

resulted in only minor degradation of solutions over the course

of hours under continuous (20 mW/cm²) or pulsed (1 mJ/cm²,

5 ns pulse) radiation. Signiﬁcant degradation was observed in

minutes under more intense pulsed irradiation, and such

conditions were avoided.

Spectroscopic and photophysical measurements were

performed as described previously,¹⁶except that for probe

wavelengths below 450 nm, a frequency-doubled beam (pulse

width below 50 fs) provided by a noncollinear optical

parametric ampliﬁer (NOPA)²⁷was used. The procedures

followed in the determination of the absorption spectra of the

radical cation and radical anion of 3 by pulse radiolysis were the

same as in previous work with 1.¹⁶

The Present Study. We describe the results of our

investigation of the ﬁrst two of the about 10 covalent dimers

of the model chromophore 1 that we have synthesized (2 and

3, Chart 1, blue = 1, red = 2, green = 3; this color coding is

Chart 1. Structures 1−3 and Their Ground State

Conformers 2syn, 2anti, and 3

Triplet Yields. Measurements of triplet yields depended on

the use of anthracene as sensitizer, and are described in detail in

the Supporting Information.

Modeling of Spectrotemporal Data. Global target

analysis was performed for 2 and 3 in DMSO and CH, and

the details are given in the Supporting Information. The three-

dimensional data sets (time delay, wavelength, change in optical

density) were used as input into the global analysis program

after correcting for the chirp of the probe beam. We used the

built-in global analysis in the software package Igor,²⁸with a

custom ﬁt function. The ﬁt was performed using compartmen-

tal analysis in matrix form in order to extract true rate constants

for the processes.²⁹

Calculations. The procedures are described in detail in the

Supporting Information. All calculations were of the density

functional theory (DFT) type, and their interpretation needs to

take into account the presence of artifacts due to under-

estimation of the energies of charge-transfer excitations.

RESULTS

■

Synthesis of 2 and 3. The preparation of the dimers 2 and

3 was analogous to the literature procedure for the preparation

of 1 by reaction of phenylmagnesium bromide with 3-

phenylphthalide (4).^30,31Use of the double Grignard reagents

obtained from bis(p-bromophenyl)methane³²and 4,4′-dibro-

mo-2,2′,6,6′-tetramethylbiphenyl³³yielded 2 and 3 in 91% yield

for both compounds (Scheme 1). Alternatively, the dibromo

compounds can be converted to organolithium compounds and

treated with 4.³⁴Although the syntheses were facile, the

puriﬁcation was not. The main diﬃculty was the air sensitivity

of these compounds in light. The puriﬁcation was carried out

under argon atmosphere by recrystallization until no traces of

impurities were detectable by NMR. The products were stored

in the dark under inert atmosphere.

used throughout). In these two dimers, the coupling of the two

halves is weak, and we ﬁnd negligible triplet yields in nonpolar

environments. As described below, they rise up to almost 10%

at low temperature in highly polar environments where a real

charge-transfer intermediate state is observed. These may be

the ﬁrst observations of the stepwise singlet ﬁssion discussed

just above, but the errors in our quantum yields unfortunately

are on the order of a few percent and they do not allow us to

distinguish whether the step that leads from a charge-transfer

excited molecule to the ﬁnal product yields one (ordinary

intersystem crossing) or two (singlet ﬁssion) triplet excitations

in the dimer.

Calculated Geometries of 1−3 (Charts 1 and S1, Table

S1 in Supporting Information). The Ground State.

4682

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

Their values optimized in various electronic states are given in

Table S1 (Supporting Information).

Scheme 1. Synthetic Routes to 2 and 3

In addition to the presumably inconsequential conforma-

tional isomerism associated with mild twisting of the rings

attached to the furan moieties, we need to consider the possibly

more important isomerism due to diﬀerent possible angles of

rotation of the two approximately planar π-electron systems 1

around the central bonds in 2 or bond in 3 (see Chart 2). Two

major conformers of 2 were subjected to geometry

optimization. The 2anti conformer is more stable by 0.47

kcal/mol. It has C₂symmetry, ω₃is −89.3°, and the

isobenzofuran moieties are oriented anti to each other.

Rotation of one of the monomers until ω₃reached nearly 0°,

followed by geometry optimization, converted the 2anti to the

2syn conformer, in which the mutual disposition of the

isobenzofuran moieties is syn.

Geometry optimization of 3 revealed an almost exactly

orthogonal twist around the central C−C bond. The two

enantiomers of this chiral C₂structure 3 are interconverted by a

180° internal rotation about the central C−C bond. The

geometry of each half of the molecule is nearly equal to that of

1, except that the phenyl twist angles deviate slightly.

Monomer. The ground-state geometry of one of the two

conformers of 1 (C₂symmetry) is known from single crystal X-

ray diﬀraction, and the bond lengths and angles are in very

good agreement with ab initio and density functional theory

calculations.¹⁶The other conformer (C_s) is predicted to have

nearly identical bond lengths and angles, but conrotatory rather

than disrotatory relative angles of ∼18° twist between the

planes of the central isobenzofuran moiety and the two phenyl

substituents. All calculated properties of the two conformers are

very similar, and it was actually diﬃcult to prove their

independent existence experimentally. Diﬀerent shapes of the

Franck−Condon envelopes of the ﬁrst transition are the main

diﬀerentiating feature.¹⁶

The C₂symmetric ground state geometries of 1, 2anti, and 3

were used as initial points for the optimization of the ﬁrst

singlet, triplet, and quintet excited state geometries.

The First Excited Singlet State (S₁). In 2 and 3, the

calculations clearly localize the excitation in one of the

chromophores, whose geometry is nearly planar (in 2, ω₁= 5

^oand ω₂= 2 ^o) and very close to that of the S₁geometry of 1,

whereas the other is somewhat twisted (in 2, ω₄= 15 ^oand ω₅

In the S₁and T₁excited states of 1, the calculated twist angle

is only about 5 and 0°, respectively. The molecule is essentially

planar and, eﬀectively, only one conformer exists. The strong

bond length alternation in the ground state, characteristic of its

o-quinoid structure, is reduced in the S₁state, and even more so

in the T₁state.¹⁶

o

= 12 ) and close to the S₀geometry of 1. In agreement with

these observations, the adiabatic S₀−S₁excitation energies

calculated for 2 and 3 (58.1 and 59.2 kcal/mol, respectively) are

only a little diﬀerent from that calculated for 1 (61.8 kcal/mol).

The solvent eﬀect on the energy of the S₁state is computed

(COSMO, DMSO, ε = 46.7) to be similar to that in 1 (Table

1).

The First Triplet State (T₁). In the dimers 2 and 3, triplet

excitation is also calculated to be localized on one of the

chromophores, whose geometry resembles that of the T₁state

of 1, while the geometry of the other is very close to that of the

S₀state of 1. In agreement with this observation, the adiabatic

S₀−T₁excitation energies for 1−3 are calculated to be nearly

identical (33.2, 32.9, and 33.0 kcal/mol, respectively). The

triplet excitation energies are not aﬀected by inclusion of

COSMO solvation (DMSO, ε = 46.7) in the calculation (Table

1).

Dimers. Experimental structural information is not available

for 2 and 3, but their formulas suggest that their ground state

bond lengths and angles will be very similar to those of the

conformers of 1, since only a weak interaction can be expected

between the two halves. The number of conformers will

however be larger, since the two aromatic substituents at

positions 1 and 3 of each isobenzofuran are inequivalent and

each can be attached with a positive or negative helical sense.

We assume that all other ground-state properties of the various

conformers of 2 and 3 are also very similar and that it is

adequate to calculate the properties of one that is

representative. This expectation has been conﬁrmed by

calculations for a few conformers, and we have arbitrarily

chosen one of the C₂symmetry minima in the ground state

potential energy surface for further work. Chart 2 indicates the

labels used to refer to the various angles and bonds in 1−3.

Chart 2. Deﬁnition of Geometrical Parameters in 1−3

4683

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

upon inclusion of the DMSO solvent (COSMO, ε = 46.7) in

the calculation (Table 1), and they become competitive with

the double triplet state.

The charge distribution is almost the same when calculated

in the absence or presence of the solvent. It is symmetric in that

the absolute total charges on the terminal phenyl groups are

similar, but both the positive and the negative charges are

polarized toward the center of the molecules in comparison

with their distribution in the radical cation and radical anion of

1 (Figure S1).

Table 1. Optimized States of 1−3: DFT Energies (B3LYP/

SVP without and with COSMO, DMSO, ε = 46.7) Relative

to the Ground State; in Square Brackets, the Solvent

a,b

Stabilization Energy (kcal/mol)

a

species

state

gas phase

DMSO solution

1

S₀

0

0 [−6.1]

54.7 [−13.2]

32.9 [−6.4]

0 [−12.1]

S₁

61.8

33.2

0

T₁

2

S₀

S₁

58.1

32.9

54.3 [−16.2]

32.7 [−12.3]

In the discussion of the geometry of the dipolar CT states,

we assumed that the separation of the dimer molecules into the

cationic and anionic halves occurs symmetrically, by a twist at

the center of the dimer. A justiﬁcation for this is provided by

the nearly perfect agreement of the absorption spectra of the

CT states of both 2 and 3 with the sum of the spectra of the

radical cation and the radical anion of 1, discussed below. The

agreement with the time-dependent DFT (TD-DFT) calcu-

lated sum of the spectra is less satisfactory but acceptable.

There are problems with this comparison in that each of the

two radical ion spectra is taken in a diﬀerent solvent, and

moreover, it is already known¹⁶that the transition energies of

both radical ions of 1 calculated at this level of approximation

are too high relative to those observed.

The Twisted Intramolecular Charge-Transfer (TICT) State.

Twisting around other single bonds in the dimers could occur

as well. Derivatives of monomeric 1 in which one of the phenyl

substituents carries electron-withdrawing substituents are

known to form TICT states upon excitation in polar solvents.³⁵

In the TICT state, the plane of the modiﬁed phenyl substituent

of 1 is believed to be approximately orthogonal to the furan

ring, based both on calculations and in analogy to the TICT

states of other molecules of the donor−acceptor type,³⁶in at

least one of which transient orthogonal twisting was proven

directly.³⁷

Therefore, it appeared possible that at least in 3, in which the

internal phenyl substituents on the isobenzofuran moieties are

modiﬁed considerably, a similar TICT species, isomeric with

the symmetrically twisted CT species described above, might

exist. Unfortunately, a CDFT/B3LYP/SVP (COSMO, ε =

46.7) geometry optimization in which either a positive or a

negative charge was constrained to the 1-phenylisobenzofuran

fragment and an opposite charge was constrained to the 1-

(2,6,2′,6′-tetramethylbiphenyl-4-yl)-3-phenylisobenzofuran

fragment did not produce satisfactory results. More than one

minimum was found, but none had a near orthogonal twist

angle at the charge separation point in the dimer. Since this is a

side issue in our project, we did not pursue the search further.

The Intermolecular Charge-Transfer State: An Ion Pair

(IP). Electron transfer from one molecule of a dimer 2 or 3 to

another will generate a radical cation 2^{+ •}or 3^{+ •}and a radical

anion 2^−•or 3^{− •}of the dimer. The optimized geometries of

these ions (Table S1) and the charge distribution reveal that

the radical ion character is localized in one-half of the molecule

while the other half is analogous to the ground state of 1. Since

the conditions for solvation of the separated ion pair are

optimal, its stabilization by the DMSO solvent is huge (Table

1), and its energy is computed to be signiﬁcantly below that of

the intramolecular charge transfer state.

T₁

b

¹(TT)

⁵(TT) = Q₁

66.0

65.3 [−12.8]

b

65.9

65.6 [−12.4]

c

b

(CA)

96.1

65.7

C

A

136.1

−26.3

109.8

0

111.2 [−36.9]

−54.7 [−40.5]

56.5 [−77.4]

0 [−11.6]

d

C and A

3

S₀

S₁

59.2

33.0

54.2 [−16.6]

32.8 [−11.8]

T₁

b

¹(TT)

⁵(TT) = Q₁

66.1

65.5 [−12.2]

b

66.0

65.6 [−12.0]

c

b

(CA)

85.7

61.1

C

A

136.0

−25.6

110.4

111.2 [−36.4]

−54.3 [−40.3]

56.9 [−76.7]

d

C and A

a

b

Calculated using the Gaussian program. Calculated using the

NWChem program. Intramolecular charge-transfer state, C stands for

radical cation and A for radical anion; constrained DFT. IP, separated

c

d

ion pair.

Double Triplet States: Quintet ⁵(TT) and Singlet ¹(TT). The

computer programs available to us permitted calculations for

the quintet ⁵(TT) and singlet ¹(TT) levels that result from the

coupling of two T₁states, one on each half of the covalent

dimer, but not for the triplet ³(TT). Since the results for ⁵(TT)

1

and (TT) were nearly identical (Table 1), it is safe to assume

3

that they would be no diﬀerent for (TT).

The optimized geometries of the double triplet states of the

dimers 2 and 3 have C₂symmetry. In each half, the geometry is

essentially indistinguishable from that of triplet 1. The central

dihedral angle ω₃is similar to that in the S₀state. In line with

the description of the quintet state as carrying a triplet

excitation on each of the two chromophores 1, the computed

adiabatic S₀−Q₁excitation energies of 2 and 3 are 65.9 and 66.0

kcal/mol, respectively, essentially exactly twice their triplet

excitation energy, and only 0.4 kcal/mol less than twice the

triplet excitation energy of the monomer 1. Explicit

consideration of the DMSO solvent has little eﬀect (Table 1).

The Intramolecular Charge-Transfer State (CT). The

CDFT/B3LYP/SVP gas-phase optimized geometries of the

charge-transfer states of 2 and 3 (Chart S1 and Table S1)

consist of radical cation and radical anion halves that are very

close to the published¹⁶calculated geometries of the

monomeric radical cation and radical anion of 1, respectively.

In 2, the 104° valence angle between the chromophores at the

central methylene group is much smaller than the 115°

computed for the ground state. In 3, the calculated dihedral

angle between the mean planes of the two halves is 78°,

signiﬁcantly lower than the 90° calculated for the ground state.

The very high CDFT B3LYP/SVP gas phase energy of the

charge-transfer states of 2 and 3 is reduced by 25−30 kcal/mol

Calculated Electronic Transitions in 1−3. B3LYP/SVP

vertical excitation energies, oscillator strengths if spin-allowed,

and polarization directions for low-energy S₀−S_n(Table S2,

Figure 1, parts A, B, and D), S₀−T_n(Table S3), and T₁−T_n

4684

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

(Table S4, Figure 2) transitions were calculated at the

optimized geometries (Table S1) of the initial states, using

the C₂symmetric conformers of 1, 2anti, and 3. Computational

results have also been incorporated in many of the other ﬁgures

(Figures 3−11). Speciﬁcally, Figure 11C shows the calculated

spectra of the radical ions of the molecular fragments present in

the TICT structure of 3 (Tables S5 and S6 give data for both

choices of polarity). Figures 7 and 11B display the calculated

absorption spectra of the radical anions and radical cations of 2

and 3 (Table S7). Their sum represents the calculated

absorption of the separated ion pair state IP, and can be used

to approximate the absorption spectra of the intramolecular CT

states of these dimers (Figure 7).

Because of the known inadequacy of the TD-DFT procedure

employed here,³⁸⁻⁴⁰artiﬁcial low-energy charge-transfer states

are calculated in weakly coupled dimers, and the very weak

computed transitions into these states need to be ignored. For

instance, four instead of two nearly degenerate low-energy

states of the dimer result from the S₀to S₁excitation in the

monomer. The two that are of locally excited nature are

represented correctly, but the two that are of charge-transfer

nature are artifacts. Since they have almost no intensity, their

presence has virtually no eﬀect on the predicted absorption

spectra. In reality, such states occur at much higher energies, as

can be seen in Table 1.

The present results for 1 are very similar to those found

earlier at a higher level of calculation,¹⁶and all spectral

interpretations remain unchanged. In particular, the S₁and T₁

states correspond to highest occupied molecular orbital to

lowest unoccupied molecular orbital (HOMO−LUMO)

promotions.

Absorption from S₀and Steady-State Emission from

S₁. The S₀−S_nabsorption and S₁−S₀ﬂuorescence spectra of 1−

3 in DMSO are displayed color coded in Figure 1C, and the

spectral characteristics measured in four solvents are listed in

Table 2. The peak absorption coeﬃcient of 1 in DMSO¹⁶is 2.5

× 10³M⁻¹cm⁻¹, and those of the dimers 2 and 3 are about

twice higher, 5.7 × 10³and 4.8 × 10³M⁻¹cm⁻¹, respectively.

The broad lowest absorption band of 2 and 3 peaks near 23 ×

10³cm⁻¹, only very little below that of 1. The relative strength

of vibronic features changes systematically from 1 to 3.

Adjusting solvent polarity causes a solvatochromic shift of the

absorbance maximum for 1−3 of less than 250 cm⁻¹, nearly the

same in each compound. The ﬁrst absorption band and the

ﬂuorescence of 1−3 can be ﬁtted to three Gaussian bands. The

relative amplitude of the 0−0 vibronic band is largest in

nonpolar solvents and is larger in 2 and 3 than in 1. Two higher

energy absorption bands are observable in the ultraviolet for

each compound, near 31 × 10³and 38.5 × 10³cm⁻¹.

In nonpolar or weakly polar solvents, the ﬂuorescence

quantum yield Φ_Fof 1−3 lies between 0.92 and 1.0. In contrast

to 1, as the solvent polarity is increased, 2 and 3 exhibit a large

decrease in Φ_F. However, the shape of the ﬂuorescence

spectrum does not change substantially.

Long-Lived Transients. Absorption from T₁. No

phosphorescence was observed, even at liquid nitrogen

temperature, and the triplet excitation energies are unknown.

For 2 and 3, the same transient T₁−T_nspectra were obtained

by anthracene sensitization and upon direct excitation in polar

solvent (Figure 2). No T−T absorption is observed upon direct

excitation in nonpolar solvents. The spectra were measured

both by steady-state photomodulation and by ﬂash photolysis.

The directly excited spectra were used to evaluate triplet

Figure 1. UV−vis absorption (solid lines) and ﬂuorescence (dashed

lines, normalized to absorption) of 1−3 in AN. Color-coded calculated

spectra are displayed above (2 and 3) and below (1).

quantum yields Φ_T(Table 2). In any solvent, the spectrum of

1,¹⁶also shown in Figure 2, is only observable upon

sensitization. The spectra of 2 and 3 contain a Gaussian band

near 21.5 × 10³cm⁻¹and a weak broad feature peaking near 14

× 10³cm⁻¹. From ground state depletion, the peak molar

extinction coeﬃcient for the former is 34 × 10³and 37 × 10³

M⁻¹cm⁻¹for 2 and 3, respectively, and can be compared with

the value of 32.1 × 10³M⁻¹cm⁻¹reported¹⁶for 1. Negative

peaks near 24 × 10³cm⁻¹are due to ground state depletion.

The lifetimes of the triplets of 2 and 3 in various solvents are

reported in Table 2. These lifetimes were determined at low

concentration from a single exponential ﬁt to the data. At

higher concentration, the decay of the triplet becomes

nonexponential. The lifetimes of both 2 and 3 at concentrations

4685

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

Figure 2. Direct (color) and sensitized (dotted) photoinduced T₁−T_n

absorption spectra of 1−3 in DMSO at delay times longer than 100 μs.

Color-coded calculated spectra are shown above.

below 10 μM are 220 20 μs, while at concentrations above 50

μM, a faster component with a concentration-dependent decay

time of less than 50 μs is apparent, presumably due to triplet−

triplet annihilation.

Triplet quantum yield was determined in a series of solvents

of varying polarity. Without sensitization, Φ_Tof 1 is at or below

our instrumental sensitivity (∼0.5%) in all solvents,¹⁶whereas

for 2 and 3 this is only true in nonpolar solvents. In polar

solvents, Φ_Tof 2 and 3 exceeds 1%, and at 300 K the value for

3 is roughly twice that for 2. The highest yield was observed for

3 in DMF at 230 K and equaled ∼9%. The triplet

photoproduction action spectra of 2 and 3 in DMSO are

shown in Figure 3 and follow the absorption spectra faithfully.

Short-Lived Transients. Whereas in 1 the excited singlet

S₁is the only short-lived transient in solvents of any polarity,

the behavior of the dimers 2 and 3 is similarly simple only in

nonpolar solvents, where their ﬂuorescence decays are

monoexponential. Their 4−5 ns lifetimes in such solvents

(Figure 4A) are slightly shorter than the monoexponential 6−7

ns lifetimes of the monomer 1 in solvents of any polarity.¹⁶

In polar solvents, we ﬁnd an additional short-lived transient

from both 2 and 3. We describe ﬁrst the room-temperature

results and then examine their temperature dependence.

Time-Resolved Emission from S₁. In polar solvents, the

ﬂuorescence decays of 2 and 3 cannot be ﬁtted with a single

exponential rate constant (Figure 4B), instead requiring two

exponential components for a satisfactory ﬁt (Tables 2 and 3).

The components vary in a systematic fashion with solvent, with

the faster decays decreasing in amplitude as solvent polarity

decreases.

Time-Resolved Absorption. The photoinduced absorp-

tion and bleaching of 2 and 3 were measured as a function of

pump−probe delay from 100 fs to 8 ns. The raw kinetic data

were ﬁtted globally to determine rate constants for formation

4686

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

Table 3. Rate Constants (ns⁻¹) Derived from

Compartmental Analysis of Spectrotemporal Data

a

solution

T

k_F

k_ETk_BET

k_NR

k_T

k_Y

k_YX

2 DMSO

3 DMSO

2 CH

RT

0.23

0.25

0.20

0.22

1.8

3.8

14

2.8

4.9

16

1.1

0.04

0.08

n/a

∼0

16

n/a

0.84

425 K

RT

13

3 CH

RT

a

RT = room temperature; n/a = not applicable. Error limits: ∼10%

except for k_NRand k_F, which show some covariance and an error limit

of ∼20%.

The ﬁrst in a series of S₁−S_nabsorption peaks appears at 21 ×

10³and 20.5 × 10³cm⁻¹in 2 and 3, respectively, close to the

location of the strongest S₁−S_npeak in 1 (Figure 5). Its

Figure 3. Direct excitation triplet production action spectra in DMSO

compared with the absorption spectra (black) for 2 and 3 in DMSO.

Dashed line is T₁−T_nsignal divided by the S₀−S₁absorption,

normalized such that the value at 24 000 cm⁻¹equals the measured

triplet quantum yield.

Figure 5. S₁−S_nabsorption spectra for 1−3 in DMSO extracted from

global ﬁt.

absorption coeﬃcient is 30 (2) and 34 × 10³M⁻¹cm⁻¹(3),

comparable to the value reported for 1. S₁−S_nabsorption bands

at higher (above 22 000 cm⁻¹) and lower (below 15 000 cm⁻¹)

energies are also observed, but the peak positions cannot be

accurately measured since they lie outside the range of

detection. In polar solvents, the transient absorption spectra

are complicated by the appearance of a short-lived nonemissive

intermediate X that appears to form from the initially excited S₁

state on a sub-nanosecond time scale and to be transformed

subsequently into the triplet T₁on a nanosecond scale. At

elevated temperatures, an additional nonemissive intermediate

Y appears in 3 (but not in 2), as discussed further below.

The results of time-resolved absorption measurements were

ﬁtted to a kinetic scheme depicted in Figure 6 using global

target analysis. The processes involved in this scheme after

photoexcitation are ﬂuorescence from the nonpolar monomer-

like singlet S₁to the ground state, transfer of population back

and forth between S₁and the intermediate state X, nonradiative

decay of X to S₀, and the formation of the triplet T₁from X,

where the disappearance of a single molecule of X could cause

the appearance of one (intersystem crossing) or two (singlet

ﬁssion) triplet excitations T₁in the same dimer molecule. In the

case of 3, at elevated temperatures, the conversion of a fraction

of S₁into Y and the subsequent decay of Y into X are also

included. Values obtained for the rate constants, deﬁned in

Figure 6, are displayed in Table 3.

Figure 4. Fluorescence decay traces of 1−3 in (A) cyclohexane and

(B) DMSO.

and decay of various species, as well as the species associated

diﬀerence spectra. As reported previously,¹⁶in the probe

wavelength range 22 000−14 000 cm⁻¹the transient absorption

spectrum of 1 in all solvents is characterized fully by stimulated

S₁−S₀emission and S₁−S_nsinglet absorption with a peak at

14.5 × 10³cm⁻¹with an absorption coeﬃcient of very roughly

20 × 10³M⁻¹cm⁻¹. This behavior is mimicked faithfully by 2

and 3 in nonpolar solvents, where 2 and 3 exhibit an

instrument-limited rise of stimulated emission and excited

singlet absorption, the decays of which can be ﬁtted with an

exponential function with rate constants displayed in Table 3.

The resulting species associated diﬀerence absorption spectra

of 2 and 3 obtained at room temperature, where Y is not

4687

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

Figure 6. Kinetic scheme used for global target analysis of transient

absorption of 2 and 3 in polar solvents.

observed, are plotted in Figure 7, and the associated decay

kinetics are shown in Figure 8. The S₁−S_nabsorption spectra

are also shown separately in Figure 5. The intermediate species

X is associated with transient absorption bands near 18.5 × 10³

and 14.8 × 10³cm⁻¹. These absorptions rise with a time

constant of ∼250 ps (2 in DMSO) or ∼150 ps (3 in DMSO),

which correlates with the decay of excited state absorption due

to S₁, strongly suggesting that X is formed from S₁. The

absorption by X decays with 0.9 ns (2) or 1.2 ns (3) time

constants. The decay of this signal is correlated with the rise of

T₁absorption features, leaving little doubt that T₁is formed

from X.

Temperature Dependence of the Fluorescence

Quantum Yield. The ﬂuorescence yield Φ_Fin DMF was

examined as a function of temperature. The compounds were

not readily soluble in glass-forming solvents, and the lower end

of the temperature range was limited by the solvent freezing

point. The data points displayed were obtained by integrating a

band near the peak of the ﬂuorescence spectrum. The

ﬂuorescence line shapes broaden slightly with increasing

temperature, but there is no noticeable peak shift. Going

from high to low temperatures, Φ_Fof 2 and 3 decreases

strongly to a minimum near 300 K, before increasing again

between room temperature and the solvent freezing point.

The Φ_Fvalues obtained above 300 K were converted to

values of the equilibrium constant K_eqfor the interconversion of

S₁and X, assuming that only S₁and X excited states are

populated. As at these temperatures, the interconversion is fast

compared with the radiative decay rate, K_eqis equal to the ratio

(1 − [S₁])/[S₁], determined from Φ_Fof ﬂuorescence from S₁,

measured in the steady state (Figure 9A). A plot of ln K_eq

against 1/T is linear in the high temperature region, and its

slope yields ΔG⁰values of 1.7 kcal/mol (2 in DMSO), 1.9 kcal/

mol (3 in DMSO), 2.4 kcal/mol (2 in DMF), and 4.6 kcal/mol

(3 in DMF) with uncertainties on the order of 5%, judging by

ﬁt errors. At lower temperatures, the interconversion of S₁and

X slows. The S₁to X transition becomes irreversible, and

equilibrium is not reached. The excited state reaction then

enters a kinetic regime, in which an Arrhenius-like dependence

of the rate k_ETon temperature is expected.⁴¹Fitting the data

below 250 K produced activation energy values in DMF of E_a=

6.1 kcal/mol for 2 and 4.9 kcal/mol for 3. The error estimated

for these steady-state values is on the order of 25% because the

low-temperature end of the curve is not quite linear. The values

thus are in reasonable agreement with those derived from

Figure 7. Species associated decay spectra resulting from global target

analysis for 2 (red) and 3 (green) in DMSO at room temperature (A).

The black dashed line is the sum of the absorption spectra of the

radical cation and radical anion of 1.¹⁶Calculated transitions for the

radical cation (orange) and anion (light blue) of 2 and 3 are shown in

B and C, respectively.

transient absorption measurements described in the next

section. It is expected that the thermally activated behavior is

complicated at low temperatures because the solvent viscosity

becomes nearly inﬁnite, causing charge transfer to be severely

hindered. The intersection of the ﬁt lines for low and high

temperature produces a crossing temperature T_c, at which k_ET

_NR. It is about 260 K for both 2 and 3.

Temperature Dependence of Transient Absorption.

=

k

Ultrafast transient absorption of 2 and 3 was measured in

DMSO and DMF. Results for 3 in DMSO probing exclusively

the kinetics of X are shown as function of temperature in Figure

10A. The logarithm of the rate constant k_ETfor 2 and 3 in

DMF is plotted against 1/T in Figure 10B. For 2, the activation

energy for the formation of X from S₁is 2.7 kcal/mol and log A

is 11.4, and for 3, the values are 3.0 kcal/mol and 11.8 (A is the

frequency factor in S⁻¹), with an estimated 10% error. These

4688

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

Figure 9. Inverse temperature dependence for 2 (red) and 3 (green)

in DMF. (A) (1 − [S₁])/[S₁]; solid lines are linear ﬁts in the low and

high temperature regimes (see text). For 2, squares, and for 3, circles.

(B) Φ_T; dashed lines show η(1 − [S₁]), where η is the eﬃciency of the

conversion of X into T₁, estimated from time-resolved spectroscopy.

For 2, circles, and for 3, squares.

Figure 8. Concentration vs time traces for 2 (A) and 3 (B) in DMSO

from the rate constants deduced from global ﬁtting of room-

temperature results, assuming initial population only in the S₁state.

Light green: S₁; orange: X; magenta: T₁, light blue: Y. The dashed

green line is S₁decay in CH.

analysis of the spectrotemporal data shows that at these

temperatures two species with diﬀerent but strongly over-

lapping spectra (Figure 11), diﬀerent decay times, and

indistinguishable rise times, equal to the decay time of S₁,

contribute to the absorption in the 14000−19000 cm⁻¹region.

The ﬁt was based on an expanded version of the kinetic

equation system S2 (Supporting Information), to which the

formation of an additional species Y from S₁and its decay to T₁

have been added. Y decays in less than 500 ps, leaving the

remaining transient population X to decay on a 2−3 ns time

scale. Due to severe spectral overlap and the small amount of

T₁present, the concentration of T₁cannot be monitored until

the other transients have decayed and we cannot tell whether

its rise is biexponential. The primary fate of Y appears to be

conversion into X, since its decay does not produce a clear

recovery of S₁or S₀. The other contributing species is the

already known X, based on its spectrum and rise and decay

times. Clearly, although X and Y are formed from S₁in

competition with each other, they are not in fast equilibrium.

Like the absorption spectrum of X, that of Y also contains a

peak attributable to the radical cation and a peak attributable to

a radical anion of the chromophore 1, but they are somewhat

shifted and present in a diﬀerent intensity ratio. The observed

ratio [Y]/[X] increases with temperature, and just below the

DMSO boiling point the absorbance due to Y is responsible for

roughly 40% of the total. The ratio of Y to X and the rise times

of both are independent of the concentration of 3 in the

numbers agree within error limits with those derived above

from steady state measurements. The rate constants of

ﬂuorescence and nonradiative decay are independent of

temperature within the experimental accuracy of about 20%.

Temperature Dependence of the Triplet Yield. Figure

9B shows Φ_Tas a function of temperature for 2 and 3 in DMF.

Since T₁is formed from X, Φ_Tis a function of [X], which itself

changes with temperature; the temperature dependence of the

scaled S₁− X equilibrium constant, approximated by η(1 −

[S₁])/[S₁], is also shown for comparison (dashed lines). Here,

η is the eﬃciency of the conversion of X into T₁, which is

estimated below to be 0.04 for 2 and 0.09 for 3. The agreement

between Φ_Tand the scaled equilibrium constant is quite good

except for temperatures below about 275 K, where Φ_T

continues to rise while the equilibrium constant for the

conversion of S₁into X reaches a plateau and subsequently

decreases. This deviation is more pronounced for 3 than for 2.

All results described up to now were qualitatively similar for

2 and 3. Next, we describe the only instance where the two

dimers diﬀer strikingly. For 2 in DMSO at temperatures above

∼360 K, the decay of X remains monoexponential, and the

system still follows the scheme described in Figure 6 and in eq.

S2 without the intervention of Y. However, under these

conditions, for 3 the transient absorption so far attributed solely

to X develops a new faster decay component and becomes

biexponential (cf. the light blue trace in Figure 8B). Global

4689

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

Figure 10. Temperature dependence of transient decay rate constants

for 2 (A) and 3 (B) in DMF. Color code: red circles, k_ET; black

squares, k_BET; green diamonds, k_NR; blue triangles, k_F.

solution in the 5−75 μM range, limited by the absorbance

range appropriate for achieving reasonable signal in transient

absorption experiments. DOSY NMR measurements provided

no indication that 3 is aggregated in the solution.

Pulse Radiolysis of 3. The absorption spectra of the radical

cation and radical anion of 3 were obtained by pulse radiolysis

in 1,2-dichloroethane and THF, respectively, and are shown in

Figure 11A.

Figure 11. (A) Absorption spectra of the radical cation (3C, purple,

squares) and anion (3A, dark yellow, circles) of 3 in 1,2-

dichloroethane and THF, respectively, and their sum (dashed line).

(B) Calculated transitions for 3C (purple) and 3A (dark yellow). (C)

Calculated transitions in the charged fragments of 3 at a geometry

twisted next to the furan ring, the radical cation of 1-phenyl-

isobenzofuran (purple) and radical anion of 1-(tetramethyl-p-

biphenylyl)-3-phenylisobenzofuran radical anion (dark yellow), at

their optimized geometries. (D) Transient absorption spectra of X

(orange) and Y (cyan), formed from 3 in DMSO.

DISCUSSION

■

Electronic States. Given the prior knowledge of the

electronic states of the monomer 1,¹⁶the locally excited states

of the covalent dimers obtained from experiment or calculation

contain no surprises. There is every reason to believe that the

excitation, or the charge and spin, are well localized on one side

of the molecule, and that the potential energy barriers for their

transfer to the other half are very small.

There is nothing unusual about the observed or the TD-DFT

calculated absorption spectra of the various states, and the

comparison of the two is very satisfactory for the neutral states

as long as the artifactual charge-transfer states are dismissed, as

discussed above. Since the interaction between the two

chromophores in the dimers is feeble, allowed vertical

transitions from the symmetric structures, allowed for S₀−S₁

and Q₁−Q_nand forbidden for S₀−T₁, are expected to be nearly

the same as in the S₀−S₁, T₁−T_n, and S₀−T₁spectra of 1,

respectively, but with twice the molar intensity. Electronic

transitions occur in nearly degenerate pairs of in-phase and out-

of-phase excitation in the two chromophores. The in-phase

transitions should be polarized in the 2-fold symmetry axis and

the out-of-phase transitions perpendicular to it.

The main conclusion from Table 1 is that in a nonpolar

medium, the calculated excitation energy of the S₁state is only

slightly lower than twice that of the T₁state, which in turn is

essentially identical with that of the double triplet states (TT),

while charge-transfer states (CA) lie far higher. In DMSO, the

situation is similar, except that now the excitation energy of the

CA states is similar to those of the S₁and TT states. Although

the DFT calculated T₁excitation energy of 1 is a little higher

than half the S₁energy, experimentally the two are

approximately equal,¹⁶and this is likely to be true for 2 and

3 as well. As far as state energies go, the conditions for eﬃcient

singlet ﬁssion in these covalent dimers are met, and it would be

approximately thermoneutral.

The T₁−T_nand S₁−S_nspectra of the dimers should resemble

the sums of the T₁−T_nor S₁−S_nspectrum of the monomer 1

4690

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

with its S₀−S_nspectrum, and they do. The transitions that

correspond to T₁−T_nor S₁−S_nexcitations should be polarized

approximately along the local short (z′) or long (y′) in-plane

axis of the excited chromophore, and those that correspond to

S₀−S_nexcitations should be polarized along the local short (z″)

or long (y″) in-plane axis of the ground-state chromophore. In

the UV region where the S₀−S_nspectrum of 1 is intense,

observation of T₁−T_nand S₁−S_nspectra was not possible, but

the observed parts of the T₁−T_nand S₁−S_nspectra look nearly

the same in 1−3, as expected. The minor diﬀerences in the

intensities of the three vibrational components of the ﬁrst

transition are probably related to the presence of more than

one conformation, due to the ambiguity in the sense of twisting

of the aryl substituents on the isobenzofuran rings.

Nonpolar Solvents. In nonpolar solvents, the photo-

physical behavior of the dimers 2 and 3 is extremely simple and

essentially identical to that of the monomer 1 in any solvent.

Fluorescence is virtually the only fate of the initially excited S₁

state. There is no indication that S₁is converted to T₁, either by

singlet ﬁssion or by intersystem crossing. The latter would not

be expected, since it does not take place in 1. We are forced to

conclude that the degree of coupling between the two

chromophores 1 in the dimers 2 and 3 is insuﬃcient to permit

singlet ﬁssion to compete successfully with emission, whose

rate constant is 0.22 (2) or 0.25 (3) ns⁻¹. Since a 1% triplet

yield would have been detected, we conclude that in a nonpolar

environment, and starting with a localized singlet excitation, the

rate constant for singlet ﬁssion in the dimers 2 and 3 is less

than 2.2 and 2.5 μs⁻¹, respectively. This contrasts starkly with

the ∼40 ns⁻¹rate constant observed in neat solid 1, and is

reminiscent of the observations that were made for covalent

dimers of tetracene.^17,18

Here, (l_Al_B)||(l_Ah_B) stands for the two-electron repulsion

integral between the overlap density formed by the multi-

plication of the LUMO on partner A (l_A) with the LUMO on

partner B (l_B) and the overlap density formed by the

multiplication of l_Awith the HOMO on partner B (h_B), and

the deﬁnition of (h_Ah_B)||(h_Al_B) is similar. Even in the best of

cases, this term is generally small because the partners A and B

are separated in space, their molecular orbitals hardly overlap at

all, the overlap densities are minute, and their mutual repulsion

usually negligible. In 2 and 3, the overlap of molecular orbitals

located on one and the other half of the molecule, the partners

A and B, is especially small, and the contribution from the

direct term must be almost exactly zero.

The second “mediated” term originates in mixing with the

charge-transfer states ¹(CA) and ¹(AC), and we believe it to be

dominant in most cases of singlet ﬁssion:⁴²

1

− (TT)|H|CA [ CA|H|S₁S₀+ AC|H|S₁S₀]/ΔE

(2)

1 1

where ΔE is the energy diﬀerence between the degenerate

charge-transfer states and the nearly degenerate S₀S₁and ¹(TT)

states. Now, both the one-electron and the two-electron parts

of H contribute

¹TT|H|CA = (3/2)^1/2[ l_A|F|h_B+(l_Ah_B)||l_Bl_B)

− (l_Ah_B)||(h_Ah_A)]

(3)

¹CA|H|S₁S₀= l_A|F|l_B+2(h_Al_A)||(h_Al_B) − (h_Ah_A)||(l_Al_B)

(4)

1

AC|H|S₁S₀= − h_A|F|h_B+2(h_Bl_B)||(h_Al_B) − (l_Bl_B)||(h_Ah_B)

(5)

where F is the Fock operator.

There could be various reasons for the much smaller rate

constant for singlet ﬁssion in the dimer than in the solid. For

instance, the higher dielectric constant in the solid stabilizes

virtual charge-transfer states, the ability of the triplets to diﬀuse

apart in the solid provides an entropic driving force for their

formation, and the likely delocalization of the initial excited

state over several chromophores might be somehow favorable

even though this does not follow from inspection of the

relevant matrix elements.¹However, the most likely source of

the small rate constant for singlet ﬁssion in the covalent dimers

is poor coupling between the individual chromophores 1 due to

inappropriate geometrical relation between them. Sensitivity to

this factor has been noted in the work with thin polycrystalline

ﬁlms of neat solid 1, in that triplet yields observed for diﬀerent

crystal modiﬁcations can diﬀer by an order of magnitude.¹⁵

In a crystal of 1, the neighboring molecules are slip-stacked,

and inspection of the expressions for the matrix elements for

both the direct and the mediated contribution to singlet ﬁssion

reveals this to be a highly desirable arrangement^1,42(other

relations between neighbors are also present and are

presumably less eﬀective). The arrangement in 2 and 3, and

also in the previously examined covalent dimers of

tetracene,^17,18is very diﬀerent, and is easily seen to be

unfavorable.

To obtain qualitative insight, we neglect the two-electron

part, in which case the mediated term can be written as

−(3/2)^1/2l_A|F|h_B[ l_A|F|l_B− h_A|F|h_B]/ΔE

(6)

Upon expansion of the orbitals l and h in atomic orbitals,

recognition of the chemical identity of partners A and B, and

introduction of the tight-binding approximation, this simpliﬁes

to

2

(3/2)^1/2c_hc_l(c_h− c_l²)β ²/ΔE

(7)

CC

where β_CCis the resonance (hopping) integral across the

junction between the two halves of the dimer, i.e., between the

carbon 2p_zatomic orbitals of the two partners at the junction

that participate in their frontier orbitals, and c_hand c_lare the

amplitudes of HOMO and LUMO at these atomic orbitals.

This quantity must be clearly small, for several reasons: (i) the

large energy diﬀerence ΔE between the S₁and the charge

transfer states (Table 1), (ii) the relatively small amplitude c_hof

the HOMO and c_lof the LUMO at the junction points, (iii) the

2

cancellation of the squares of the expansion coeﬃcients c_h

from the HOMO and c_l²from the LUMO (because of the

pairing theorem for alternant hydrocarbons, these squares are

equal in tetracene; they will be close to equal even in 1), and

last but not least, (iv) the small magnitude of β_CC². In 2, where

the carbon atoms at the junction are connected through a CH₂

group, β_CConly reﬂects the eﬀects of hyperconjugation through

this group. In 3, the junction carbons are connected directly,

but the two π systems are twisted nearly orthogonal and β_CC

must again be small. Clearly, linking the two halves of a

covalent dimer symmetrically end-to-end, as has been done

In the ﬁrst approximation, the electronic matrix element

whose square enters into the rate expression contains a sum of

two terms. The ﬁrst “direct” term contains only contributions

from the two-electron part of the Hamiltonian H,

¹(TT)|H|S₁S₀= (3/2)^1/2[(l_Al_B)||(l_Ah_B) − (h_Ah_B)||(h_Al_B)]

(1)

4691

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

here and in the earlier work with tetracene,^17,18is far from

optimal if fast singlet ﬁssion is the objective. However, at least

the choice of the linking point is somewhat favorable, since for

most other choices the coeﬃcients c_hand c_lare even smaller.

Polar Solvents. The novelty oﬀered by the dimers 2 and 3

is in their photophysics in polar solvents, where in both cases S₁

is in rapid equilibrium with an intermediate X. An intriguing

puzzle is provided by the observation that at elevated

temperatures the S₁state of the dimer 3 also produces an

additional intermediate Y, while that of the otherwise extremely

similar dimer 2 does not. In the following, we focus on these

aspects of the results.

From the temperature dependence of the equilibrium

constant for the interconversion of X with S₁, the former is

1.5−4.5 kcal/mol below the latter. Since X and Y only appear in

polar solvents, an immediate suspicion is that they are dipolar

species produced by electron transfer from one to the other

chromophore within the dimer, stabilized by twisting⁴³and

asymmetric solvation. Intramolecular charge transfer in weakly

coupled (usually twisted) dimers is well precedented, for

instance in the classical case of 9,9′-bianthryl.⁴⁴

singlet X and triplet X are expected to be very similar. The

absorption spectrum attributed to X might actually be due to a

1

3

1

3

mixture of X and X, or exclusively to one or the other. Since

the intermediate is nonemissive as nearly as we can tell, we

cannot use ﬂuorescence for a diﬀerentiation. However, the

normal values of the frequency factor for the rates of the

formation of X from S₁in 2 and 3, about 10¹², leave no doubt

that when ﬁrst formed, X is in its singlet state ¹X. The

frequency factor for the reverse reaction is also roughly 10¹²

and the rapid equilibration between S₁and X is certainly

compatible with the notion that the bulk of the dipolar species

is present as ¹X throughout. Its conversion to T₁thus requires a

multiplicity change. This could occur by the usual spin−orbit

coupling-induced intersystem crossing or by going through the

second phase of two-step singlet ﬁssion.

It is clear from the experimental results and also from

calculated energies (Table 1) that opportunities for singlet

ﬁsion are improved in polar solvents. The improvement is not

large for the one-step process, since the value of ΔE in the

denominator in expression 7 refers to virtual charge-transfer

states, which are stabilized by a solvent that is more polarizable

but oﬀer no time for solvent reorientation and therefore are not

stabilized by solvent polarity. The relevant ΔE is therefore

nowhere near zero as Table 1 might otherwise suggest.

However, the two-step singlet ﬁssion process, which proceeds

through a real intermediate that can take full advantage of

solvent polarity, now becomes feasible. The observation that

the dipolar intermediate is indeed formed rapidly makes it clear

that the electronic matrix elements for the initial charge

transfer, ⟨¹CA|H|S₁S₀⟩ or ⟨¹AC|H|S₁S₀⟩ (in our approximation,

The assignment of X as a charge-transfer species, in which

the donor and the acceptor are of equal size, separated by the

central covalent junction in the molecule, and hence both equal

to 1, is conﬁrmed beyond reasonable doubt by comparison of

peak positions and intensities in its absorption spectrum with

those of the radical cation and the radical anion of the

monomer 1, which were reported previously¹⁶(Figures 7 and

11). Within the observed region, the radical cation of 1 has a

single strong peak at 18.5 × 10³cm⁻¹(ε_max= ∼17.8 × 10³

M⁻¹cm⁻¹) and its radical anion has an even stronger single peak

at 15.2 × 10³cm⁻¹(ε_max= ∼23.5 × 10³M⁻¹cm⁻¹). The spectra

of the radical cation and radical anion of 3, reported presently,

are very similar to those of the radical ions of 1, but are more

intense. The radical cation of 3 peaks at ∼17.9 × 10³cm⁻¹(ε_max

= ∼44.5 × 10³) and the radical anion of 3 at ∼15.2 × 10³cm⁻¹

(ε_max= ∼33.5 × 10³).

2

−(3/2)^1/2c_l²β_CCor −(3/2)^1/2c_hβ_CC²), have a suﬃcient

magnitude. The question is whether the matrix element for

the step in which two triplets are generated from the charge-

transfer state, ⟨¹TT|H|CA⟩ (in our approximation, −(3/2)^1/2c_hc_l

β_CC²), is suﬃciently large for this process to be competitive.

2

Since c_l², c_h, and c_lc_hhave the same order of magnitude, this

appears to be possible.

In a covalent dimer such as 2 or 3, where triplets formed by

singlet ﬁssion cannot diﬀuse apart, they most likely remain

correlated and need to be thought of as superpositions of nine

sublevels. If the spin part of the Hamiltonian is neglected, they

form the ¹(TT), ³(TT), and ⁵(TT) states. The ﬁrst two are not

the lowest states of their multiplicity and can perhaps decay

rapidly by internal conversion, the former into S₀and possibly

also S₁, and the latter to T₁. Thus, even if singlet ﬁssion works

well and the ¹(TT) state is initially produced eﬃciently, there is

no guarantee that much T₁will be observed in the experiments.

The spin part of the Hamiltonian is capable of mixing the

singlet, triplet, and quintet sublevels of the double triplet state

and may protect some of the triplets formed from decay. These

considerations suggest that it would be interesting to examine

the eﬀect of outside magnetic ﬁeld on the kinetics of the X to

T₁conversion.

The spectra of X contain two intense peaks in the same

region, and their positions and intensities agree very well with

those expected for the sum of the spectra of the radical cation

and the radical anion. They occur at 15.5 × 10³cm⁻¹(ε_max= 30

× 10³M⁻¹cm⁻¹) and 19.2 × 10³cm⁻¹(ε_max= 20 × 10³

M⁻¹cm⁻¹), respectively, for 2, and at 15.0 × 10³cm⁻¹(ε_max

=

34 × 10³M⁻¹cm⁻¹) and 18.4 × 10³cm⁻¹(ε_max= 17 × 10³

M⁻¹cm⁻¹), respectively, for 3. The spectra of X formed from 2

and 3 thus diﬀer only by minor shifts, and their sum is nearly

exactly equal to the sum of the spectra of the radical ions for

both 2 and 3 (Figure 7A).

In contrast to the spectra calculated by the TDDFT method

for the uncharged states, those calculated for the radical ions of

1−3 agree with observations only qualitatively, in that a single

intense peak is predicted in each, but its excitation energy is too

high by 2−3 × 10³cm⁻¹in the case of the radical cation and by

4−5 × 10³cm⁻¹in the case of the radical anion. A similar

discrepancy was noted earlier for the radical ions derived from

1.¹⁶If one approximates the calculated absorption spectrum of

X by the sum of the spectra calculated for its radical cation

constituents, an agreement with experiment is reached only if

one assumes that similar empirical corrections need to be made

as for the monomeric radical ions, and this is entirely

reasonable.

The rate constant of formation of X from S₁, k_ET, for 2 in

DMF and DMSO is considerably smaller than in AN (and to a

lesser extent for 3 in AN), a less viscous solvent. In the limit of

nonadiabatic charge transfer, a relationship between ΔG⁰, ΔG^‡

and the reorganization energy λ can be formulated:

ΔG^‡= ΔG⁰/2 + λ/4 + (ΔG⁰)²/4λ

(8)

where ΔG⁰represents the driving force, and ΔG^‡is the

eﬀective barrier for the transition. From known values of ΔG⁰

and ΔG^‡, values of λ can be derived. For 2 and 3 in DMF, the

The spin state of the dipolar species X is not apparent from

its visible spectrum, since the absorption spectra of the dipolar

4692

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

reorganization energies are 7.5 and 4.8 kcal/mol, respectively.

Although the charge-transfer process may not be in the fully

nonadiabatic regime, values of λ are comparable to those

measured for other charge-transfer transitions in covalently

bound dimers,⁴⁵suggesting that this zero-order approximation

captures the essential reaction dynamics. The larger reorganiza-

tion energy for 2 may be an indication that the geometry (and

possibly dipole moment) of 2 changes more dramatically

between S₁and ¹X than it does for 3, necessitating a

concomitant change in the solvent conﬁguration. This is

consistent with the fact that 2 has more numerous low energy

conformers than 3. An additional contribution may arise from

inner-sphere eﬀects, including intramolecular reorganization,

although speciﬁc vibronic contributions to λ are not included in

the previous analysis, and a more extensive study would need to

be performed in order to test the full inﬂuence that torsional

motion might have on the transition rate, as has been

concluded in other cases.⁴⁶

The dipolar species X, whose bulk is present in the singlet

state, is in a thermal equilibrium with S₁. Since its relative

amount in the equilibrium increases at low temperatures, it is

not surprising that the yield of T₁increases as temperature is

reduced. In the low-temperature limit, the return of X to S₁

becomes negligibly slow, and the supply of X is dictated by the

rate of its formation from S₁. Most of the species X returns

nonradiatively to the ground state by diabatic back electron

transfer, and only a small fraction crosses to the relaxed triplet

state T₁. The complex nature of the disappearance of S₁is

reﬂected in the multiexponential decay of its ﬂuorescence. The

three unimolecular processes that depopulate X (going to S₁, to

S₀, or T₁) compete, but the return to S₀is the fastest and

determines the lifetime. Since neither the rate constant for the

disappearance of X nor that for the appearance of T₁shows a

temperature dependence, the internal conversion to S₀is not

thermally activated, and this is not surprising, given its

exothermic nature that places it in the Marcus inverted region.

The triplet yield exhibits a temperature dependence that

mimics that of the S₁−X equilibrium above about 275 K.

However, as the solvent approaches the freezing point, the rate

constant for the formation of X decreases, presumably due to

the slowing rate of solvent motion that must accompany the

charge transfer event.

known molar absorption coeﬃcient of T₁to ﬁnd for 2 in

DMSO that each molecule of X that disappears produces 0.04

0.02 molecules of T₁. For 3, the value is 0.09 0.03. The

values of the T₁−T_nextinction coeﬃcient and the transient

absorption amplitudes (especially of the triplet, with its small

relative yield) are the largest source of uncertainty. These

values are clearly far below unity and thus do not provide direct

information about the possible role of singlet ﬁssion in the

formation of triplets. Modeling the bleach, decay of X, and rise

of T₁versus time assuming either the production of two triplets

from a single X (singlet ﬁssion) or one triplet from a single X

(ordinary intersystem crossing) does not produce results

diﬀerent enough to decide with certainty which process occurs.

Although we ﬁnd that 2 and 3 are good candidates for the type

of two-step singlet ﬁssion discussed in the literature,¹we do not

have strong enough evidence for this process to be able to

deﬁnitely assign any of the triplet yield to singlet ﬁssion. Even if

the observed triplets originate in singlet ﬁssion, their yield is

disappointingly small, due to the eﬃcient decay of the dipolar

state X to S₀.

The transient spectrum of the species Y at higher

temperatures in 3 (cyan curve in Figure 11D) also contains

two strong peaks in the visible region but is quite diﬀerent from

the spectrum of X obtained from 3 (dark yellow curve in Figure

11D), most notably by a reversal of the relative strengths of

radical cation and radical anion peaks, a red shift of the cation

peak, and doubling of the anion peak. It is, however, suﬃciently

similar to the sum of spectra obtained for the radical cation and

anion of 3 by pulse radiolysis (Figure 11A) that an assignment

to a free radical ion pair 3C + 3A had to be considered. This

would have to be formed in an encounter of the dipolar species

X with a ground state 3. A diﬀusive encounter is ruled out by

the fast (<100 ps) formation of Y even at relatively low

concentrations and the concentration independence of its

formation rate. A search for ground-state aggregation by

diﬀusion-ordered spectroscopy (DOSY) NMR provided no

support for this possibility (Supporting Information).

Somewhat reluctantly, we must conclude that the conversion

of S₁of 3 to Y is monomolecular and that Y is a conformational

isomer of the intramolecular charge-transfer state X, whose

formation is favored at higher temperatures. The structure of 3

certainly contains a suﬃcient number of single bonds around

which rotation is possible. The absence of Y among the

transients formed from 2 would then be attributed to an

absence of an analogous low-energy conrformational minimum

or a higher barrier for its formation from S₁. The rapid decay of

Y shows that it is not in fast equilibrium with X. The decay does

not appear to lead to the ground state of 3 since no

correspondingly fast component of ground state bleach

recovery is seen. The observations are compatible with a

decay of Y to X or possibly T₁(by singlet ﬁssion or ordinary

intersystem crossing).

A likely structure of Y is suggested by a recent report of

twisted intramolecular charge-transfer states of derivatives of 1

substituted in the phenyl ring in polar solvents, in which one of

the phenyl rings is twisted orthogonal to the furan ring to

which it is attached.³⁵TD-DFT calculations of the absorption

spectra of the radical ions of the fragments that would result

from cutting the conjugated system of 3 between a furan ring

and a tetramethylbiphenyl moiety (Tables S5 and S6) showed

that the sum of the spectra of a radical cation on the smaller

fragment and a radical anion on the larger fragment is

compatible with the spectrum observed for Y if one makes a

If ordinary intersystem crossing is largely responsible for the

triplet formation, the additional temperature dependence of Φ_T

not contained in the S₁−X equilibrium could be explained by

1

3

the existence of an equilibrium between X and X with only a

3

small fraction of the species in the X state. The X species

present may have to overcome a thermal barrier to proceed to

T₁. The temperature dependence of the T₁formation rate will

1

3

then reﬂect both the diﬀerence in the energies of X and X,

which aﬀects the equilibrium constant, and the activation

energy for the ³X to T₁step. There is precedent for intersystem

crossing in a similar charge-transfer state.^47,48

Is any of the triplet produced by singlet ﬁssion? In principle,

the results of the global analysis can be used to tell whether the

decay of one molecule of X produces one or two molecules of

T₁, if the true yields of X and T₁can be accurately measured

versus time. This is diﬃcult because most of the decay of X

produces S₀. Assuming that S₁is the only source of X, and that

ﬂuorescence is the only other decay mode of S₁, the quantum

yield of formation of X is 1 − Φ_F. Assuming that below 360 K,

X is the only source of T₁, and recognizing that the eﬃciency η

with which T₁is formed from X is Φ_T/(1 − Φ_F), we use the

4693

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

similar empirical shift to lower energies as was necessary to

reach agreement with the spectrum of X (Figure 11). The

appearance of two calculated transitions of lower intensity

instead of a single intense transition in the radical anion appears

to ﬁt the observations and argues against the opposite

assignment of charges. However, we make no claim that the

structure of Y has been established with certainty.

at Northwestern University was supported by the MRSEC

program and the Northwestern Materials Research Science and

Engineering Center (DMR-1121262). Theory work in Prague

was supported by the Institute of Organic Chemistry and

Biochemistry (RVO:61388963) and the Czech Science

Foundation (P208/12/G016).

REFERENCES

■

CONCLUSIONS

■

(1) Smith, M. B.; Michl, J. Singlet Fission. Chem. Rev. 2010, 110,

6891−6936.

(2) Nozik, A. J.; Ellingson, R. J.; Micic, O. I.; Blackburn, J. L.; Yu, P.;

́ ́

Triplet formation and decay were observed in dimers of 1

designed to test the coupling strength and energy-level

dependence of singlet ﬁssion. Triplet quantum yields in the

linear dimers 2 and 3 reach ∼9% in polar solutions, much in

excess of those observed in monomeric 1 in solution, but far

below those observed in neat solid 1. They depend systemati-

cally on solvent and temperature. Clear evidence was obtained

for a charge-transfer intermediate in their formation, but it is

not certain whether two-step singlet ﬁssion or ordinary

intersystem crossing is involved when the charge-transfer

intermediate is converted to T₁.

We conclude that at least for this chromophore, triplet yield

in the weakly coupled linear dimers 2 and 3 is nowhere near

200%, and singlet ﬁssion is far too slow to observe at all in a

nonpolar solvent and maybe not in a polar solvent either. We

have examined the electronic matrix elements involved and

concluded that the direct mechanism to singlet ﬁssion plays an

entirely negligible role in 2 and 3, and that even the normally

dominant mediated mechanism cannot contribute much,

because of the way in which the two halves of the dimers are

connected. The argument applies even more strongly to the

covalent dimers of tetracene examined by others earlier.^17,18

Even if two-step singlet ﬁssion in our covalent dimers 2 and 3

in polar solvents does take place via the real polar intermediates

observed, and triplet formation is not due to ordinary

intersystem crossing, it does not compete well with loss

channels. At this point, it raises interesting mechanistic issues

but does not appear to oﬀer an advantageous route to

eﬃciency.

Murphy, J. E.; Beard, M. C.; Rumbles, G. Unique Approaches to Solar

Photon Conversion Based on Semiconductor Nanostructures and

Novel Molecular Chromophores; Dynamics of Electron Relaxation,

Interfacial Charge Transfer, and Carrier Multiplication. Proceedings of

the 27th DOE Solar Photochemistry Research Conference, June 6−9,

2004, Airline Conference Center, Warrenton, VA, pp 63−66. Available

online: http://science.energy.gov/∼/media/bes/csgb/pdf/docs/

solar_photochemistry_2004.pdf.

(3) Hanna, M.; Nozik, A. J. Solar Conversion Efficiency of

Photovoltaic and Photoelectrolysis Cells with Carrier Multiplication

Absorbers. J. Appl. Phys. 2006, 100, 074510/1−074510/8.

(4) Shockley, W.; Queisser, H. J. Detailed Balance Limit of Efficiency

of p−n Junction Solar Cells. J. Appl. Phys. 1961, 32, 510−519.

́

(5) Paci, I.; Johnson, J. C.; Chen, X.; Rana, G.; Popovic, D.; David, D.

E.; Nozik, A. J.; Ratner, M. A.; Michl, J. Singlet Fission for Dye-

Sensitized Solar Cells: Can a Suitable Sensitizer Be Found? J. Am.

Chem. Soc. 2006, 128, 16546−16553.

(6) Johnson, J. C.; Nozik, A. J.; Michl, J. The Role of Chromophore

Coupling in Singlet Fission. Acc. Chem. Res. 2013, DOI: 10.1021/

ar300193r.

(7) Burdett, J.; Muller, A. M.; Gosztola, D.; Bardeen, C. J. Excited

̈

State Dynamics in Solid and Monomeric Tetracene: The Roles of

Superradiance and Exciton Fission. J. Chem. Phys. 2010, 133, 144506−

144517.

(8) Lee, J.; Jadhav, P.; Baldo, M. A. High Efficiency Organic

Multilayer Photodetectors Based on Singlet Exciton Fission. Appl.

Phys. Lett. 2009, 95, 033301−033303.

(9) Rao, A.; Wilson, M. W. B.; Hodgkiss, J. M.; Albert-Seifried, S.;

Bassler, H.; Friend, R. H. Exciton Fission and Charge Generation via

Triplet Excitons in Pentacene/C₆₀Bilayers. J. Am. Chem. Soc. 2010,

132, 12698−12703.

(10) Roberts, S. T.; McAnally, R. E.; Mastron, J. N.; Webber, D. H.;

Whited, M. T.; Brutchey, R. L.; Thompson, M. E.; Bradforth, S. E.

Efficient Singlet Fission Found in a Disordered Acene Film. J. Am.

Chem. Soc. 2012, 134, 6388−6400.

(11) Ramanan, C.; Smeigh, A. L.; Anthony, J. E.; Marks, T. J.;

Wasielewski, M. R. Competition between Singlet Fission and Charge

Separation in Solution-Processed Blend Films of 6,13-Bis-

(triisopropylsilylethynyl)pentacene with Sterically-Encumbered Per-

ylene-3,4:9,10-bis(dicarboximide)s. J. Am. Chem. Soc. 2012, 134, 386−

397.

̈

ASSOCIATED CONTENT

* Supporting Information

■

S

Supporting Information available: synthetic procedures, triplet

yield measurements, modeling of spectrotemporal data, DOSY

NMR, methods of calculation, and computational results:

molecular geometries (Chart S1 and Table S1), charge

distribution in the charge-transfer state (Figure S1), and

electronic transitions (Tables S2−S7). This material is available

free of charge via the Internet at http://pubs.acs.org.

(12) Chen, Y.; Lee, B.; Fu, D.; Podzorov, V. The Origin of a 650 nm

Photoluminescence Band in Rubrene. Adv. Mater. 2011, 23, 5370−

5375.

(13) Ryasnyanskiy, A.; Biaggio, I. Triplet Exciton Dynamics in

Rubrene Single Crystals. Phys. Rev. B: Condens. Matter Mater. Phys.

2011, 84, 193203−193206.

AUTHOR INFORMATION

Notes

The authors declare no competing ﬁnancial interest.

■

ACKNOWLEDGMENTS

■

(14) Wang, C.; Tauber, M. J. High-Yield Singlet Fission in a

Zeaxanthin Aggregate Observed by Picosecond Resonance Raman

Spectroscopy. J. Am. Chem. Soc. 2010, 132, 13988−13991.

(15) Johnson, J. C.; Nozik, A. J.; Michl, J. High Triplet Yield from

Singlet Fission in a Thin Film of 1,3-Diphenylisobenzofuran. J. Am.

Chem. Soc. 2010, 132, 16302−16303.

This material is based upon work supported by the U.S.

Department of Energy, Oﬃce of Basic Energy Sciences,

Division of Chemical and Biosciences, under Contract DE-

AC36-08GO28308 with NREL for optical spectroscopy of

dimers, DE-SC0007004 at CU-Boulder for dimer synthesis and

calculations, and DE-AC02-98-CH10886 with BNL for pulsed

radiolysis experiments including use of the LEAF Facility of the

BNL Accelerator Center for Energy Research. Theoretical work

(16) Schwerin, A. F.; Johnson, J. C.; Smith, M. B.; Sreearunothai, P.;

́

Popovic,

́

D.; Cerny, J.; Havlas, Z.; Paci, I.; Akdag, A.; MacLeod, M. K.;

́

et al. Toward Designed Singlet Fission: Electronic States and

4694

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

The Journal of Physical Chemistry B

Article

Photophysics of 1,3-Diphenylisobenzofuran. J. Phys. Chem. A 2010,

114, 1457−1473.

Intramolecular Charge-Transfer States and Structures. Chem. Rev.

2003, 103, 3899−4032.

(17) Muller, A. M.; Avlasevich, Y. A.; Mullen, K.; Bardeen, C. J.

́

́ ́

(37) Dobkowski, J.; Wojcik, J.; Kozminski, W.; Kołos, R.; Waluk, J.;

̈

Michl, J. An Experimental Test of C−N Bond Twisting in the TICT

State: Syn−Anti Photoisomerization in 2-(N-Methyl-N-isopropylami-

no)-5-cyanopyridine. J. Am. Chem. Soc. 2002, 124, 2406−2407.

(38) Dreuw, A.; Weisman, J. L.; Head-Gordon, M. Long-Range

Charge-Transfer Excited States in Time-Dependent Density Func-

tional Theory Require Non-local Exchange. J. Chem. Phys. 2003, 119,

2943−2946.

(39) Gritsenko, O.; Baerends, E. J. Asymptotic Correction of the

Exchange-Correlation Kernel of Time-Dependent Density Functional

Theory for Long-Range Charge-Transfer Excitations. J. Chem. Phys.

2004, 121, 655−670.

Evidence for Exciton Fission and Fusion in a Covalently Linked

Tetracene Dimer. Chem. Phys. Lett. 2006, 421, 518−522.

(18) Muller, A. M.; Avlasevich, Y. A.; Schuller, W. W.; Mullen, K.;

̈

Bardeen, C. J. Exciton Fission and Fusion in Bis(tetracene) Molecules

with Different Covalent Linker Structures. J. Am. Chem. Soc. 2006, 129,

14240−14250.

(19) Michl, J.; Nozik, A. J.; Chen, X.; Johnson, J. C.; Rana, G.; Akdag,

A.; Schwerin, A. F. Toward Singlet Fission for Excitonic Solar Cells. In

Organic Photovoltaics VIII; Kafaﬁ, Z. H., Lane, P. A., Eds.; Proceedings

of SPIE; SPIE: Bellingham, WA, 2007; Vol. 6656, p 66560E1.

(20) Wurthner, F.; Kaiser, T. E.; Saha-Moller, C. R. J-Aggregates:

̈

(40) Autschbach, J. Charge-Transfer Excitations and Time-Depend-

ent Density Functional Theory: Problems and Some Proposed

Solutions. ChemPhysChem 2009, 10, 1757−1760.

From Serendipitous Discovery to Supramolecular Engineering of

Functional Dye Materials. Angew. Chem., Int. Ed. 2011, 50, 3376−

3410.

(41) Figure 66 and associated text on p. 3954 of ref 36.

(42) Smith, M. B.; Michl, J. Recent Advances in Singlet Fission. Annu.

Rev. Phys. Chem. 2013, 64, 361−386.

(21) Furlan, A.; Riley, M. J.; Leutwyler, S. The Jahn−Teller Effect in

Triptycene. J. Chem. Phys. 1992, 96, 7306−7320.

(22) Greyson, E. C.; Stepp, B. R.; Chen, X.; Schwerin, A. F.; Paci, I.;

Smith, M. B.; Akdag, A.; Johnson, J. C.; Nozik, A. J.; Michl, J.; et al.

Singlet Exciton Fission for Solar Cell Applications: Energy Aspects of

Interchromophore Coupling. J. Phys. Chem. B 2010, 114, 14223−

14232.

(23) Chan, W. L.; Ligges, M.; Jailaubekov, A.; Kaake, L.; Miaja-Avila,

L.; Zhu, X. Y. Observing the Multiexciton State in Singlet Fission and

Ensuing Ultrafast Multielectron Transfer. Science 2011, 334, 1541−

1545.

(24) Greyson, E. C.; Vura-Weis, J.; Michl, J.; Ratner, M. A.

Maximizing Singlet Fission in Organic Dimers: Theoretical Inves-

tigation of Triplet Yield in the Regime of Localized Excitation and Fast

Coherent Electron Transfer. J. Phys. Chem. B 2010, 114, 14168−

14177.

̌ ́

́

(43) Bonacic-Koutecky, V.; Michl, J.; Kohler, J. Prediction of

̈

Structural and Environmental Effects on the S₁−S₀Energy Gap and

Jump Probability in Double-Bond cis−trans Photoisomerization. A

General Rule. Chem. Phys. Lett. 1984, 104, 440−443.

(44) Schneider, F.; Lippert, E. Elektronenspektren und Elektronen-

struktur von 9,9′-Dianthryl. Ber. Bunsenges. Phys. Chem. 1968, 72,

1155−1160.

(45) Kang, T. J.; Kahlow, M. A.; Giser, D.; Swallen, S.; Nagarajan, V.;

Jarzeba, W.; Barbara, P. F. Dynamic Solvent Effects in the Electron-

Transfer Kinetics of S₁Bianthryls. J. Phys. Chem. 1988, 92, 6800−6807.

(46) Elich, K.; Kitazawa, M.; Okada, T.; Wortmann, R. Effect of S₁

Torsional Dynamics on the Time-Resolved Fluorescence Spectra of

9,9′-Bianthryl in Solution. J. Phys. Chem. A 1997, 101, 2010−2015.

(47) Mac, M.; Kwiatkowski, P.; Pischel, U. Temperature Dependence

of Bianthryl Dual Fluorescence. Chem. Phys. Lett. 2002, 357, 440−449.

(48) Dance, Z. E. X.; Mickley, S. M.; Wilson, T. M.; Ricks, A. B.;

Scott, A. M.; Ratner, M. A.; Wasielewski, M. R. Intersystem Crossing

Mediated by Photoinduced Intramolecular Charge Transfer: Juloli-

dine−Anthracene Molecules with Perpendicular p Systems. J. Phys.

Chem. A 2008, 112, 4194−4201.

(25) Teichen, P. E.; Eaves, J. D. A Microscopic Model of Singlet

Fission. J. Phys. Chem. B 2012, 116, 11473−11481.

(26) Stevens, B.; Marsh, K. L; Baltrop, J. A. Photoperoxidation of

Unsaturated Organic Molecules. 21. Sensitizer Yields of O₂¹Δ_g. J.

Phys. Chem. 1981, 85, 3079−3082.

(27) Johnson, J. C.; Gerth, K. A.; Song, Q.; Murphy, J. E.; Scholes, G.

D.; Nozik, A. J. Ultrafast Exciton Fine Structure Relaxation Dynamics

in Lead Chalcogenide Nanocrystals. Nano Lett. 2008, 8, 1374−1381.

(28) http://www.wavemetrics.com/products/igorpro/igorpro.htm.

(29) Stokkum, I. H. M.; Larsen, D. S.; van Grondelle, R. Global and

Target Analysis of Time-Resolved Spectra. Biochim. Biophys. Acta

2004, 1657, 82−104.

(30) Hauser, C. R.; Tetenbaum, M. T.; Hoffenberg, D. S.

Condensations Involving the Metalation of the 3-Position of 3-

Phenylphthalide by Means of Alkali Amides. Carbonation of Phthalide.

J. Org. Chem. 1958, 23, 861−865.

(31) Newman, M. S. Evidence Favoring a Two-Step Mechanism for

the Diels−Alder Reaction. J. Org. Chem. 1961, 26, 2630−2633.

(32) Streitwieser, A.; Vorpagel, E. R.; Chen, C.-C. Carbon Acidity.

66. Equilibrium Ion Pair Acidities of Substituted Diphenylmethanes in

Cyclohexylamine. J. Am. Chem. Soc. 1985, 107, 6970−6975.

(33) Helms, A.; Heiler, D.; McLendon, D. Electron Transfer in Bis-

porphyrin Donor−Acceptor Compounds with Polyphenylene Spacers

Shows a Weak Distance Dependence. J. Am. Chem. Soc. 1992, 114,

6227−6238.

(34) Dodge, J. A.; Bain, J. D.; Chamberlin, A. R. Regioselective

Synthesis of Substituted Rubrenes. J. Org. Chem. 1990, 55, 4190−

4198.

(35) Jacq, J.; Tsekhanovich, S.; Orio, M.; Einhorn, C.; Einhorn, J.;

Bessieres, B.; Chauvin, J.; Jouvenot, D.; Loiseau, F. Structure and

̀

Dynamics of the Excited States of 1,3-diarylisobenzofurans: An

Experimental and Theoretical Study. Photochem. Photobiol. 2012, 88,

633−638.

(36) Grabowski, Z. R.; Rotkiewicz, K.; Rettig, W. Structural Changes

Accompanying Intramolecular Electron Transfer: Focus on Twisted

4695

dx.doi.org/10.1021/jp310979q | J. Phys. Chem. B 2013, 117, 4680−4695

Article Doi

Toward designed singlet fission: Solution photophysics of two indirectly coupled covalent dimers of 1,3-diphenylisobenzofuran

DOI: 10.1021/jp310979q

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jp310979q

Full text of DOI:10.1021/jp310979q

Products guided by the article

R&D Labs maybe for 1204678-87-4

Relevant to this article

Hot Product