You are on page 1of 17

Energy Transfer Mechanisms and the Molecular Exciton Model for Molecular Aggregates

Author(s): Michael Kasha


Source: Radiation Research, Vol. 20, No. 1 (Sep., 1963), pp. 55-70
Published by: Radiation Research Society
Stable URL: http://www.jstor.org/stable/3571331
Accessed: 26-04-2016 08:25 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3571331?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Radiation Research Society is collaborating with JSTOR to digitize, preserve and extend access to
Radiation Research

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

RADIATION RESEARCH 20, 55-71 (1963)

Energy Transfer Mechanisms and the Molecular Exciton

Model for Molecular Aggregates "2


MICHAEL KASHA
Department of Chemistry and the Institute of Molecular Biophysics, Florida State University,
Tallahassee, Florida

I. INTRODUCTION

Energy transfer phenomena continue to fascinate the biologist, chemist, and


physicist, and the literature contains numerous experiments designed to demonstrate the presence of excitation energy transfer. At the present stage of development of the theory of such processes, the goal of these researches might well be not

merely the demonstration of the presence of energy transfer, but a more detailed
investigation designed to establish which available mechanism of energy transfer
is operative.
The field of excitation energy transfer is replete with redundant and overlapping

terms, whose multitude itself lends little order to the subject. Various authors
quote, as energy transfer mechanisms, exciton transfer, resonance transfer, dipoledipole transfer, inductive resonance, resonance force transfer, virtual photons, and
diffuse excitation, among others.

This paper will be devoted to the laying down of a further perspective for the
discrimination and understanding of the main types of excitation energy transfer.
A necessary introduction to such a discussion will be the examination of detailed
differences in types of intermolecular electronic interaction. The present treatment
will be a strictly nonmathematical one; a later paper will serve to develop the mathe-

matical framework at an intermediate level to that available in the literature (1, 2).
1 This paper was presented at the Exciton Symposium held at the 10th Annual Meeting of
the Radiation Research Society, Colorado Springs, May 20-23, 1962. The Society is indebted
to Dr. Michael Kasha, Director of the Institute of Molecular Biophysics, The Florida State
University, Tallahassee, for the organization of the symposium and for assembling the manuscripts for publication.
2 This study was supported in part by a contract between the Division of Biology and
Medicine and the U. S. Atomic Energy Commission and in part by the Chemistry Division,
U. S. Air Force Office of Scientific Research, and the Florida State University.
55

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

56

KASHA

II. THE MOLECULAR EXCITON MODEL: DEFINITIONS

The exciton model may be defined as the treatment of the resonance interaction
between excited states of weakly coupled aggregate systems. The various forms of
the theory depend on whether the system consists of atomic ions or molecular units,
on the nature of the external disturbance, on the strength of intermolecular interaction, on the structure of the aggregate, and on the nature of the interacting excited

states. The goal of the treatment is primarily to give the correct description of the
excited states of the aggregate. Secondarily, the nature of excitation migration may
be described.

To limit the discussion to optically excited aggregate systems, this paper will deal
only with transverse excitons, in which case the perturbation of the aggregate system
by the electric vector of the light wave is perpendicular to the direction of propagation of the photon (rather than longitudinal excitons, excited by momentum transfer

parallel to the direction of impingement of a colliding massive particle). Furthermore, atomic excitons, in which the excited electron distribution function covers
many lattice sites and hence may be described as a loosely bound exciton (Wannier
exciton), will not be considered (3).
Instead, molecular excitons, in which the excited electron is strongly bound to a
specific molecular center in a van der Waals aggregate of molecules (e.g., molecular
crystals), will be described (3). We may then seek to describe the excited states
of the molecular aggregate in terms of the excited states of the component molecules as a starting point. Under certain conditions (see later discussion), a collective excitation (Frenkel exciton) of the system of locally excited states serves as
the best description of the excited molecular aggregate; alternatively, a superposition of these collective excited states may be used, which describes the excitation
in terms of a traveling-wave packet of excitation, the exciton.
In general, the formal treatment may be applied to a crystalline molecular solid,
or to other molecular aggregates such as multilayered laminae, or a monolayer; or
to helical and other polymers. However, the formalism may be used in part to
study the intermolecular electronic excited-state interaction in much simpler

molecular aggregates, such as dimers and trimers.


III. COMPARISON OF ENERGY TRANSFER MECHANISMS

The application of the Frenkel exciton treatment to geometry-determined


molecular crystal problems was made by Davydov (1, 4). A characteristic result
is the generation of an exciton band, whose width is dependent on the oscillator
strength of the corresponding electronic transition in the unit molecule, inversely
proportional to the cube of the intermolecular center distance, and a function of
the geometrical relationship between transition dipoles in the unit molecules for the
optical transition in question. For an aggregate of N molecular units, the exciton

band will consist of N discrete exciton states; again characteristically, only one or

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

57

ENERGY TRANSFER MECHANISMS

ENERGY TRANSFER MECHANISMS

FREE

COLLECTIVE EXCITATION

EXCITON

Band shift, splittings


transfer rate > 10 secl

C I/r3

EXCITON >> I

MODEL 2 << I
LOCALIZED

Davydov E

EXCITON

LOCALIZED EXCITATION
STEPWISE TRANSFER

(Identical units)

12 -I

transfer rate 10 sec

C: I /r3

INTERACTION OF
CONTINUA

MODEL
Forster

VERY>

WEAK

U?I

A"E7

VIBRATIONAL-

STEPWISE TRANSFER

RELAXATION

(Dissimilar units)

RESONANCE

transfer rote < 10 sec

TRANSFER

Theoretical

Energy

P henomenological

Model

Coupling

Model

OC I/r6

Characteristics

Criteria

FIG. 1. Energy transfer mechanisms

a very few (equal in number to the number of molecules per unit cell) of these
exciton states for the aggregate can be reached by allowed electric-dipole transitions.

Leaving the question of the detailed exciton band structure and the consequent
observable spectral effects to Sections IV and V, we turn first to the consideration
of the various limits for intermolecular coupling cases (Fig. 1) as proposed by Simpson and Peterson, by F6rster, and by Davydov.
Simpson and Peterson (5) proposed what seem to be the most operationally practicable criteria for the strength of intermolecular electronic interaction between
excited states relative to the strength of intramolecular vibrational-electronic
coupling. For their strong-coupling exciton case they propose the criterion 2U/AI >>

1, where 2U is the exciton band width, and AE is the Franck-Condon band width
of the corresponding molecular electronic transition in the individual molecular

unit. For the weak-coupling criterion, Simpson and Peterson propose the converse
inequality, 2U/Ae << 1. Thus, if a band shift for the exciton states of a molecular

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

,58

KASHA

aggregate is observed which is greater than the total electronic band width of the
corresponding transition in the individual molecule, the strong-coupling case prevails. We shall favor the Simpson and Peterson strong-coupling and weak-coupling
classification and criteria and discuss these and the case intermediate to those in
Section IV.

Forster (6), in one of the keynote presentations of energy transfer theory, proposed a slightly modified designation of these two criteria. In his attempt to bring
unity to the subject, F6rster proposed that Simpson and Peterson's criteria (aside

from a factor of 2) be reclassified as "strong-coupling" and "medium-coupling"


cases, reserving "weak coupling" as a term for his own vibrational-relaxation resonance transfer case. This suggestion does not seem satisfactory for two reasons. Fore-

most is the fact that Forster's proposed "weak-coupling" criterion (labeled "very
weak" in Fig. 1), U/Ae' << 1, where U is the interaction energy for excited states,
and AE' is the individual vibronic band width, although understandable in principle,
is not operationally accessible. Individual vibronic bands are seldom resolved in
most solution spectra, and these are the spectra of primary interest. (Experimentally

one could not distinguish Simpson and Peterson's original weak-coupling case and
Forster's weak-coupling case by direct spectral criteria.) Secondly, the exciton model
is based essentially on a first-order perturbation theory treatment, with an inverse-

cube dependence on the distance between interacting excited-state transition dipoles. There is a theoretical discontinuity in the transition to the Forster model.
Forster's vibrational-relaxation resonance transfer model has been derived by him
both by a phenomenological path (7) as well as by the quantum-mechanical interaction of continuous configurations (6, 8), which yields results essentially of the
form familiar in second-order perturbation theory and results in an interaction
energy which depends on the inverse sixth power of the distance between interacting excited transition dipoles. Further discussion of this mechanism will be given
in Section VI.

Davydov (1) gave a theoretical classification of exciton cases (Fig. 1) based on


quite different premises compared with the spectroscopic criteria just discussed. He
classified free excitons as those for which the resonance lifetime of a locally excited

state (t = h/2AU, where A U is the exciton band width) is much shorter than the
lattice relaxation time (time for the lattice to re-equilibrate as a result of changes in

intermolecular forces caused by excitation). Davydov's localied excitons correspond


to the converse inequality. In the latter case, the excitation transfer rate is limited
by the lattice relaxation, and the exciton may be identified with a slowly traveling
localized lattice deformation. In a rough way these two cases have a correspondence

to the strong-coupling and weak-coupling cases defined by Simpson and Peterson.


but their different origins must be understood in examining Fig. 1.

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

59

ENERGY TRANSFER MECHANISMS

IV. VIBRATIONAL-ELECTRONIC COUPLING CASES: SPECTRAL EFFECTS

In this section we shall examine schematically the types of spectral effect which
can be observed for the strong-coupling, intermediate-coupling, and weak-coupling
exciton cases of Simpson and Peterson (5).
Strong coupling requires the Born-Oppenheimer separability of intramolecular

electronic and vibrational wave functions for a molecule (5, 9). Thus, the allowed
excited electronic states of identical molecules in a molecular aggregate will interact,

and the vibrational envelope will follow along. Figure 2 schematically illustrates
the spectral effects (molar absorption coefficient versus frequency in cm-1) for
such a case. The upper curve shows a Franck-Condon band envelope for a strong
electric-dipole molecular electronic transition in a unit molecule. In a dimer (middle curves), a splitting occurs in principle, although one of the components may be
forbidden (see Section V) and not observable. However, in the dimer, the excitation
is now distributed over both molecules, so the vibrational frequencies in the excited
state will change (exaggerated in figure). In the infinite polymer (bottom curve),
STRONG COUPLING

MONOMER

DIMER

INFINITE
POLYMER

I,

FIG. 2. Spectral effects of the strong-coupling exciton model

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

60

KASHA

a much larger exciton splitting will be observed, but the band width of the allowed
components now becomes characteristically very narrow. The latter effect is caused
by the fact that the excitation is now spread over many molecules, so that each

molecule has approximately the same electronic structure as a ground-state molecule. In spectroscopic language, only the 0,0 or vibrationless electronic transition
is now allowed by the Franck-Condon principle.

In strong coupling, the splitting of the spectral band would be exactly double for
the infinite polymer compared with the dimer, in the approximation of nearestneighbor interaction (since in the former case each molecule has two neighbors).
Actually, the splitting in the infinite polymer would be somewhat greater than
double that for the dimer, because the net interaction would be obtained by summing the interaction over all molecule neighbors to a given molecule. Since the
dipole-dipole interaction falls off as the inverse cube of distance, most of the interaction would come from the first six neighbors on each side. In our schematic illustration, a topologically linear polymer is considered. The actual geometry for a real
case would modify the observed spectral components of the exciton band. Furthermore, the splitting would be symmetrical about the origin only for a molecule with
no permanent dipole moment.
Weak coupling (Fig. 3, right side) corresponds to the failure of Born-Oppenheimer
separability of intramolecular electronic and vibrational wave functions for a mole-

cule (5, 9). Application of the Simpson and Petersen criteria to the whole electronic

band having determined that the weak-coupling case prevails, one now calculates
the splitting of each vibronic sub-band separately. In the limit of weak coupling,
no spectral band shape change is observed, but instead second-order effects such as
hypochromism or hyperchromism (band intensity decreases or increases in aggre-

INTERMEDIATE COUPLING WEAK COUPLING

",
FIG. 3. Spectral effects of the intermediate-coupling and weak-coupling cases in the exciton

model.

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

ENERGY TRANSFER MECHANISMS

61

gate) may be observed. In practice, this case may apply roughly when the molar
absorption coefficient of an electronic absorption band falls below 1000.
Intermediate coupling (Fig. 3, left side) constitutes one of the difficult unsolved
problems of exciton theory (10, 11). The electronic and vibrational wave functions
are not separable, so that individual vibronic sub-bands interact according to their
individual oscillator strengths (intensities). However, here the splitting of vibronic
sub-bands degenerate in the molecular aggregate is comparable with the vibronic

spacings in the original electronic transition for the individual molecule. Consequently, a general confusion of the band contour may result, as schematically in-

dicated in the lower curve (Fig. 3, left side). The exact spectral contours for various
coupling strengths are the subject of current investigation. Intermediate coupling
probably prevails for molecular electronic bands of normal band width and of molar

absorption coefficients in the range -1000 to 10,000.


V. EXCITON BAND STRUCTURE FOR STRONG COUPLING

In the previous section it was seen that the most dramatic spectral effects for
exciton states of molecular aggregates are seen for strong-coupling cases. It is quite
rewarding to translate the quantum-mechanical results for these cases into a quasiclassical vector model for the electrostatic interaction of dipoles.
The intermolecular interaction in molecular aggregates should be expressed accurately by potential terms of coulomb type between the electrons and nuclei of

molecules of an aggregate. Since this accurate formulation leads to insurmountable


mathematical difficulties, the intermolecular coulomb potential is replaced by a

multipole expansion, which summed over all multipoles (dipole-dipole, quadrupole-quadrupole, octupole-octupole, . . ., dipole-quadrupole, . . ., etc.) becomes
equivalent to the coulomb potential. However, since the electronically excited states
are usually of (allowed) electric-dipole type, only the dipole-dipole intermolecular
potential term is used, as an approximation, in the molecular exciton model. Then,
one may easily factor the quantum-mechanical intermolecular interaction integrals;
and also in this approximation one may understand the nature of the exciton model
in terms of a quasi-classical electrostatic vector model.

Dimers of dye molecules have been studied by the exciton formalism by Levinson
et al. (12). Figure 4 shows the possible exciton band structure's and selection rules
(after Kasha et al., 13). Two separate questions may be answered by such vector
diagrams: (1) What is the energy of the resulting exciton states, and (2) what are
the electric dipole selection rules for optical transitions to these states?

In the diagrams of Fig. 4, it is assumed that the polarization vector for light absorption is parallel to the long axis of a long planar molecule, whose skeleton is
schematically indicated by a flat rectangular plate. Other electronic transitions may
of course be considered, polarized along the short axis of the plate, or out-of-plane;
or oblique to the skeletal axes for an "unsymmetrical" molecule. No new selection
rules arise for these additional cases.

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

62

KASHA

PARALLEL HEAD-TO-TAIL OBLIQUE

-G

-E'

AX

E'EAE

MONOMER DIMER MONOMER DIMER MONOMER DIMER

BLUE-SHIFT RED-SHIFT BAND-SPLITTING

(Hypsochromic) (Bathochromic)

FIG. 4. Diagrams for exciton band structure in molecular dimers with various geometrical
arrangements of transition dipoles. After Kasha et al. (13).

To deduce the energy of the exciton state of the dimer relative to the monomer

excited-state energy, one must ask whether the dipole-dipole interaction will be
attractive or repulsive. To deduce the selection rule for light absorption, one must
take the vector sum of the transition dipoles for the given exciton state. It naturally

turns out that only in-phase arrangements of dipoles give allowed exciton states,
which is necessary, since the wavelength of the light used for electronic excitation
of molecules is much greater than molecular dimensions of ordinary molecules, and
thus molecules of a dimer should be simultaneously under an in-phase perturbation.
The oblique dimer yields a splitting into two allowed components, but with mutually

perpendicular polarization for light absorption to the exciton states of the dimer.
Numerous actual examples of molecular dimers corresponding to several different
geometrical arrangements have been studied recently (13, 14).
Linear polymer exciton models are of considerable value to consider, since they

afford the simplest example of full exciton band development in a geometry-determined molecular aggregate system, even though actual linear polymers (as dis-

tinguished from helical ones) may be rare or unstable in nature. Selection rules in
the linear polymer exciton model were developed by McRae and Kasha (15) together with interaction energy expressions in the nearest-neighbor approximation
for the cases illustrated in Fig. 5.
In each energy-level diagram of Fig. 5, the polymer is assumed to consist of N
molecular identical units or chromophores. Before excited-state interaction, an

N-fold excitation degeneracy exists, depicted schematically on the left side of each
energy-level diagram. After the electrostatic perturbation caused by the intermolecular dipole-dipole interaction of transition dipoles, an exciton band of N

discrete exciton states is produced, as shown on the right side of each energy-level
This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

ENERGY TRANSFER MECHANISMS 63


HEAD-TO-TAIL CARD- PACK ALTERNATE TRANSLATIONAL

7_

__

_A__

__^

EXCITED

STATE

-__

GROUND STATE

(.. ^1 -^ ......) (V4.tt 4l..tt ..4) (t^ .-tt).

.- = ;, = (-1 H-Hi H-H-H-t (_..m


FIG. 5. Diagrams for exciton band structure in linear molecular polymers with various
geometrical arrangements of transition dipoles. After McRae and Kasha (15).

diagram. The solid line indicates the allowed exciton state of the polymer, which
can be reached by electric-dipole radiation; the dashed lines correspond to forbidden
exciton states.

Below the energy-level diagrams for each case are shown vector diagrams for a
few of the exciton states of the polymer. In the head-to-tail case, the nodeless exciton

wave function (bottom vector array) corresponds to all transition dipoles in phase,
yielding the lowest-energy exciton state for this exciton band, and also an allowed
exciton state, with a finite net transition moment. The exciton wave function with a

single excitation node (vector diagram just above bottom array) will correspond to
the exciton state next higher in energy, but now with a transition dipole vector sum

equal to zero, corresponding to a forbidden exciton state. The number of excitation


nodes increases as the exciton-state energy increases, until there are excitation nodes

between each molecular unit (top vector diagram for the head-to-tail case). This
last gives the highest-energy exciton band component, with repulsive electrostatic
interaction between each transition dipole, and with a net transition moment of zero,

corresponding to a forbidden exciton state.


The diagrams for the other two geometries can be interpreted analogously, but
with differing net exciton band structure. The first two cases from the left in Fig. 5

correspond to one molecule per unit cell, and the case on the right to two molecules
per unit cell in the space group for the system, and lead consequently to one, one,
and two allowed levels in the exciton band system, respectively. Thus, the linear
polymer exciton models demonstrate very simply the relation of exciton wave funcThis content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

64

KASHA

tion nodal properties and exciton-state energy and selection rules for electric-dipole
transitions.

The forbidden exciton states, although not observed directly by light absorption
in most cases, nevertheless play an extremely important role in excitation and light
emission from molecular aggregates (12, 15). A forbidden lowest exciton level in an
exciton band corresponds to a metastable singlet level. If a triplet level lies below
such a level, greatly enhanced triplet-state excitation may result, accompanied by a
total quenching of normal singlet-state emission of the unit molecule.
Helical polymer exciton models have special interest in connection with spectra
and energy transfer in polypeptides and proteins, polynucleotides, and deoxyribonucleic acids. Moffitt (16) has developed the formalism for the exciton model
applied to polypeptides.
We may develop a more complete exciton band diagram to describe the exciton
states for a helical polymer with four molecular chromophore units (not electroni-

cally conjugated) per turn of the helix (Fig. 6).


The operational resolution of interest for light absorption in a helical polymer is
for exciton states yielding light absorption polarized along the helix axis and perpendicular to the helix axis. Taking components of the peptide chromophore transition dipoles parallel to the helix axis leads to the linear head-to-tail exciton band
structure previously discussed (Fig. 5) and depicted in the lower right section of
Fig. 6.

++perpendicular++ +to ax+- +is--+ +parallel to axis -+-

perpendicular to axis parallel to axis

x,y

\--- Iz
FIG. 6. Diagrams for exciton band structure in helical molecular polymers with four nonconjugated molecular chromophore units per turn.

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

ENERGY TRANSFER MECHANISMS

65

Components of the peptide chromophore transition dipoles perpendicular to the


helix axis may be most simply discussed for one complete turn of the polymer (if
one assumes here four chromophores per turn). The four vector diagrams of Fig. 6
correspond to the four exciton states with excitation nodes indicated by changes in
algebraic signs below the vector diagrams. For example, the left diagram is for a
nodeless exciton wave function, with lowest binding energy. The second from the
left is a vector diagram corresponding to one node between the W and N transition
dipole and the E and S transition dipole. The exciton wave function would be, for
example, of the form
"2 = ~~7 (sW*oPNPE(PS + <P W<PNEP*S - f WJPNPE *0S - TO W WNPE1S*)

There are evidently two one-noded exciton wave functions, giving an exciton-state

degeneracy. The dipole-dipole interactions may be regarded as consisting of two


attractive and two repulsive interactions, leading to zero shift (relative to unit
chromophore transition) as far as the transition dipole interaction is concerned. The
single two-noded exciton is indicated at the right side of Fig. 6, with the transition
dipole phase or direction changing at each chromophore, yielding four repulsive
interactions, or the exciton state of highest energy. In the present discussion nearest neighbor interactions, only, are considered.

If the interactions are summed over the entire helix relative to the perpendicular
projection, additional contributions to the splitting will occur, but the polarization
results will be unchanged. The exciton band structure thus shows two allowed bands

at the same energy and with polarization vectors for light absorption which are
mutually perpendicular and perpendicular to the helix axis, and one allowed exciton
state at the bottom of the exciton band polarized parallel to the helix axis.
The spectroscopic observations on the polarized ultraviolet absorption of oriented

films of poly-L-alanine and poly-y-methyl-L-glutamate by Gratzer et al. (17) confirm the pattern and polarizations of exciton states for the helical polymer exciton

model and confirm Moffitt's calculated value (16) of the exciton splitting. The spectroscopic studies of the helix-random coil transition in poly-L-lysine and poly-Lglutamic acid by Rosenheck and Doty (18) lend further support to the justification
of the treatment of the polypeptide a-helix by the helical polymer exciton model.
In polynucleotides the near-ultraviolet absorption band at about 2600 A is of

such an intensity as to make the strong-coupling case inapplicable. Intermediate


coupling in the scheme of Simpson and Peterson probably prevails in this case.

Extensive studies on the optical properties of synthetic and natural polynucleotides


are under way in several laboratories. Rhodes (19) has recently investigated the

electronic states of double-stranded polynucleotides by space-group theory.


Molecular crystal exciton band structure may become very complex compared to
the cases discussed above. Figure 7 shows the Davydov splitting, as the exciton
splitting is usually called in molecular crystals, schematically for the case of anThis content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

66

KASHA

- *.mA+ mB- ma

/ A IA Bm
o s (^ . . _ - C _ _ _ _

,2A \M ?B ,_ _

/ ( \ .... _.
/D

--

I ,- I W _ _ _ _ _

Florida, January 1961.__

vapor

crystal

FIG. 7. Exciton band structure (Davydov splitting) for the anthracene crystal (schematic).
From M. Ashraf El-Bayoumi's Doctoral Dissertation, Florida State University, Tallahassee,
Florida, January 1961.

thracene (20). This crystal contains two molecules per unit cell. The displacements
of the exciton manifold of states by energy 2A and 2B correspond to excitation
exchange between translationally equivalent molecules along the two axes of the
unit cell, and the splitting by energy 8C corresponds to the excitation exchange between the translationally nonequivalent molecules of the unit cell. Two allowed
components for the exciton states of the crystal are thus possible. However, the
exciton-state pattern is not simple, and a correlation between exciton-state wave
functions and exact location of a particular exciton state within the band is not
available.

In the simple discussion presented in this section, several features of interest were
lost owing to the simplifications or the approximations used. Thus, the symmetrical

splitting of an exciton band is merely a consequence of the nearest neighbor interaction approximation; in polymeric aggregates the splitting would be unsymmetrical

in general. Also, the displacement of the whole exciton band origin, owing to the
changes in static intermolecular forces between an excited molecule and its neighbors, was also omitted.

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

ENERGY TRANSFER MECHANISMS

67

VI. RATES OF ENERGY TRANSFER FOR VARIOUS MECHANISMS

Rates of energy transfer for various theoretical excitation energy transfer models

were considered by F6rster in his elegant recent comparative study (6). Here a few
words of commentary will be added which may focus attention on new experimental
approaches to the problem.

It should be clear from the present paper and from Forster's that several energy
transfer mechanisms are available for the exchange of excitation energy between
similar and between dissimilar molecules. The interest should lie in distinctions

which are possible between these mechanisms.


Let us compare the rates of excitation transfer by rewriting Forster's expressions

(6) as a function of experimental parameters (Table I). In these expressions n is the


transfer rate in number of transfers per second; in the first two cases it is calculated

as the reciprocal of the resonance lifetime (t = h/2AU) of the oscillatory exciton


state, whereas in the third case it is for the nonoscillatory transition.
The quantity in parentheses in each expression includes the oscillator strength f
for the electronic transition in the unit molecule, a geometric function 0(0) expressing the angular dependence of the dipole-dipole interaction, and the distance
between interacting centers r.

For the strong-coupling case, the oscillator strength is integrated over the entire
electronic band. For weak coupling an additional factor is present, which consists
in summing over, for the vibrational levels v' and v of the nonexcited and the excited molecules, respectively, the populations of these levels, gv' and gv*, and the
vibronic overlap integrals. Effectively, in this case, individual vibronic band intensities are involved.
TABLE I
RATE OF EXCITATION TRANSFER

Strong-coupling exciton (Simpson and Peterson):

Weak-coupling exciton (Simpson and Peterson):

n V I( 36() ) g,9v* gV 'vv

Continuum interaction model (F6rster):

n(f @(a)) V,V'gv* gv Svv


(~~~3 \/ ~ ~ IE

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

68

KASHA

Let us now compare the distinctive features of the three mechanisms. It is clear

that the two exciton cases yield an inverse cube of the intermolecular distance dependence as a distinctive feature (see also Fig. 1). This thus becomes a very longrange effect as an extremely characteristic feature. With large spectral effects (e.g.
SA - 1000 cm-1) in the strong-coupling case, resonance rates may be as high as 1015
sec- . However, with much smaller exciton band widths (e.g., AP - 10 to 100 cm-1),
the resonance lifetimes may still be in the range 1012 to 1013 sec-1. However, now in-

termediate-coupling or weak-coupling cases may prevail (Simpson and Peterson


criteria), so the Franck-Condon integral squared comes in as a factor and slows
down the rate over what one would calculate from the f value for the whole elec-

tronic transition. It should thus be noted that, although the strong-coupling exciton
model is often thought of as an excitation energy exchange mechanism for identical,
strongly absorbing molecules in the solid state, nevertheless it also may apply in the

intermediate-coupling or weak-coupling case to dissimilar molecules with fairly


close lowest excited energy levels, or to molecules dispersed in a medium with inter-

molecular distances to 100 A. The exciton model may yield an important competing
excitation energy transfer process in many chemical and biological cases.
The mechanism introduced by F6rster (7, 6, 21), which is termed here vibrational-

relaxation resonance transfer (cf. Table 1 and Fig. 1), differs fundamentally from
the exciton model mechanism. It is widely thought to be the only long-range energy

transfer mechanism available from theory. It is of interest to compare it carefully


with the exciton model. As F6rster has shown, it originates quantum mechanically
from the interaction-of-continua model (8), in which the resonance transfer rate
is now proportional to the square of the interaction energy, so that now an inverse

sixth-power dependence on distance between interacting molecules becomes the


characteristic feature. Thus, although this still provides for a "long-range" intermolecular interaction, it is evidently not so long-range an effect as the exciton models
provide.
The F6rster vibrational-relaxation resonance transfer mechanism has a rate of

transfer which is severely cut down over that of the weak-coupling exciton model
by the fact that the fourth power of the Franck-Condon integral for vibronic transi-

tions now multiplies the f value; Ae', the individual vibronic band width (in principle), appears in the denominator as a compensating factor. The F6rster mechanism
may be the dominant one at large distances in comparison with the weak coupling
exciton case.

Because of its detailed nature, the vibrational-relaxation resonance transfer


mechanism is especially applicable to interaction between appropriate dissimilar
molecules. Nevertheless, it may in some situations apply to identical-molecule interaction in a limited way (6).
It is of extreme importance to seek experimental discrimination between these two
classes of mechanism, especially in regard to distance dependence. Forster (6)

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

ENERGY TRANSFER MECHANISMS

69

considers a continuous transition between the two classes of mechanism (see especially his Figs. 3 and 4). It is possible that in some experiments the direct competition between the weak-coupling exciton mechanism and the vibrational-relaxation
resonance transfer mechanism may be detected, or in other words the weak-coupling
exciton mechanism might extend far into the range of the vibrational-relaxation
resonance transfer rates (106 to 1011 sec-1).
One final point which remains on the vibrational-relaxation resonance transfer
mechanism concerns the exact validity of the continuum interaction theory to the
problem. The origin of the continuous feature of spectra in solution is of course not

the same as, e.g., the predissociation continua and other cases to which it is usually
applied.
The topic of intermolecular excitation energy provides fertile territory for much
new investigation.
VII. SUMMARY

The molecular exciton model is defined as the resonance interaction between the

excited states of loosely bound molecular aggregates (molecular crystals, nonconjugated polymers, molecular laminae, molecular dimers and trimers). The Simpson
and Peterson vibrational-electronic coupling cases are applied to spectral characteristics of molecular aggregates.
The quasi-classical electrostatic vector model is described for simple molecular
aggregates as an approach to understanding the formation of the exciton-state
formation. Selection rules and exciton band structures are described for several

molecular aggregate structures in the strong-coupling exciton model.


In strong-coupling exciton cases, pronounced spectral effects of molecular aggregates compared with individual molecules are predicted, with very fast resonance
transfer rates ( 1015 sec-1) depending on the inverse cube of intermolecular dis-

tance. In weak-coupling exciton cases, no conspicuous spectral changes are predicted, but the transfer rates are still high (--1012 to 1013 sec-1) and preserve the
inverse-cube intermolecular distance dependence.
Forster's vibrational-relaxation resonance transfer mechanism is emphasized as a

contrasting mechanism based on a different theoretical origin (interaction of continua model), with inverse sixth-power dependence on intermolecular distance and
much slower transfer rates (106 to 1011 sec-1).
Emphasis is placed on the need for new discriminative experimental work.
REFERENCES

1. A. S. DAVYDOV, Theory of Molecular Excitons (translated by. M Kasha and M. Oppenheimer, Jr.), 174 pp., McGraw-Hill Book Company, New York, 1962.
2. D. S. McCLURE, Electronic spectra of molecules and ions in crystals. Solid State Phys. 8,
1-47 (1960).
3. General discussion and references will be found in M. KASHA, Relation between exciton

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

70

KASHA

bands and conduction bands in molecular lamellar systems. Rev. Mod. Phys. 31, 162-169

(1959); also in Biophysical Science (Oncley, Schmitt, Williams, Rosenberg, and Bolt,
eds.), pp. 162-169, John Wiley & Sons, New York, 1959.
4. A. S. DAVYDOV, Theory of absorption spectra of molecular crystals. Zhur. Eksptl. Teoret.
Fiz. 18, 210-218 (1948).

5. W. T. SIMPSON and D. L. PETERSON, Coupling strength for resonance force transfer of


electronic energy in van der Waals solids. J. Chem. Phys. 26, 588-593 (1957).
6. TH. FORSTER, Excitation transfer. In Comparative Effects of Radiation (M. Burton, J. S.
Kirby-Smith, and J. L. Magee, eds.), pp. 300-319, John Wiley & Sons, New York, 1960.

7. TH. FORSTER, Zwischenmolekulare Energiewanderung und Fluoreszenz. Ann. Phys. 2, 55


(1948).

8. CF. H. A. KRAMERS, Quantum Mechanics, pp. 218 ff, North-Holland Publishing Co., Amsterdam, 1958.

9. CF. D. S. MCCLURE, Energy transfer in crystals and double molecules. Can. J. Chem. 36,

59 (1958).
10. E. G. McRAE, Molecular vibrations in the exciton theory for molecular aggregates. Australian J. Chem. 14, 329, 344, 354 (1961).
11. CF. R. E. MERRIFIELD, Vibronic states of dimers. Radiation Res. 20, 154-158 (1963).
12. G. L. LEVINSON, W. T. SIMPSON, AND W. CURTIS, Electronic spectra of pyridocyanine dyes
with assignment of transitions. J. Am. Chem. Soc. 79, 4314-4320 (1957).
13. M. KASHA, M. ASHRAF EL-BAYOUMI, and W. RHODES, Excited states of nitrogen base-

pairs and polynucleotides. J. Chim. Phys. 58, 916 (1961).


14. M. KASHA and M. ASHRAF EL-BAYOUMI, in preparation for publication.
15. E. G. McRAE and M. KASHA, Enhancement of phosphorescence ability upon aggregation
of dye molecules. J. Chem. Phys. 28, 721-722 (1958).
16. W. MOFFITT, The optical rotatory dispersion of simple polypeptides. Proc. Natl. Acad.
Sci. U. S. 42, 736 (1956).
17. W. B. GRATZER, G. M. HOLZWARTH, and P. DOTY, Polarization of the ultraviolet absorp-

tion bands in alpha-helical polypeptides. Proc. Natl. Acad. Sci. U. S. 47, 1785 (1961).
18. K. ROSENHEOK and P. DOTY, The far ultraviolet spectra of polypeptide and protein solutions and their dependence on concentration. Proc. Natl. Acad. Sci. U. S. 47, 1775 (1961).
19. W. RHODES, Symmetry of helical polymers. Electronic states of double-stranded polynucleotides. J. Chem. Phys. 37, 2433-2439 (1962).

20. D. P. CRAIG and P. C. HOBBINS, The polarized spectrum of anthracene. J. Chem. Soc.
1955, 539, 2302, 2309.

21. TH. F6RSTER, Transfer mechanisms of electronic excitation. Discussions Faraday Soc.
27, 717 (1959).

This content downloaded from 129.125.32.217 on Tue, 26 Apr 2016 08:25:49 UTC
All use subject to http://about.jstor.org/terms

You might also like