Self-assembly of supramolecular light-harvesting arrays from covalent multi-chromophore perylene-3,4:9,10=bis(dicarboximide) building blocks

Article abstract of DOI:10.1021/ja039820c

We report on two multi-chromophore building blocks that self-assemble in solution and on surfaces into supramolecular light-harvesting arrays. Each building block is based on perylene-3,4:9,10-bis(dicarboximide) (PDI) chromophores. In one building block, N-phenyl PDI chromophores are attached at their para positions to both nitrogens and the 3 and 6 carbons of pyromellitimide to form a cross-shaped molecule (PI-PDI₄). In the second building block, N-phenyl PDI chromophores are attached at their para positions to both nitrogens and the 1 and 7 carbons of a fifth PDI to produce a saddle-shaped molecule (PDI₅). These molecules self-assemble into partially ordered dimeric structures (PI-PDI₄)₂ and (PDI₅)₂ in toluene and 2-methyltetrahydrofuran solutions with the PDI molecules approximately parallel to one another primarily due to π-π interactions between adjacent PDI chromophores. On hydrophobic surfaces, PDI₅ grows into rod-shaped nanostructures of average length 130 nm as revealed by atomic force microscopy. Photoexcitation of these supramolecular dimers in solution gives direct evidence of strong π-π interactions between the excited PDI chromophore and other PDI molecules nearby based on the observed formation of an excimer-like state in <130 fs with a lifetime of about 20 ns. Multiple photoexcitations of the supramolecular dimers lead to fast singlet-singlet annihilation of the excimer-like state, which occurs with exciton hopping times of about 5 ps, which are comparable to those observed in photosynthetic light-harvesting proteins from green plants.

Full text of DOI:10.1021/ja039820c

Published on Web 06/05/2004
Self-Assembly of Supramolecular Light-Harvesting Arrays
from Covalent Multi-Chromophore
Perylene-3,4:9,10-bis(dicarboximide) Building Blocks
Michael J. Ahrens,† Louise E. Sinks,† Boris Rybtchinski,† Wenhao Liu,†
Brooks A. Jones,† Jovan M. Giaimo,† Alexy V. Gusev,† Andrew J. Goshe,‡
David M. Tiede,‡ and Michael R. Wasielewski*,†
Contribution from the Department of Chemistry and Center for Nanofabrication and Molecular
Self-Assembly, Northwestern UniVersity, EVanston, Illinois 60208-3113, and Chemistry DiVision,
Argonne National Laboratory, Argonne, Illinois 60439
Received November 26, 2003; E-mail: wasielew@chem.northwestern.edu
Abstract: We report on two multi-chromophore building blocks that self-assemble in solution and on surfaces
into supramolecular light-harvesting arrays. Each building block is based on perylene-3,4:9,10-bis(di-
carboximide) (PDI) chromophores. In one building block, N-phenyl PDI chromophores are attached at their
para positions to both nitrogens and the 3 and 6 carbons of pyromellitimide to form a cross-shaped molecule
(PI-PDI4). In the second building block, N-phenyl PDI chromophores are attached at their para positions to
both nitrogens and the 1 and 7 carbons of a fifth PDI to produce a saddle-shaped molecule (PDI5). These
molecules self-assemble into partially ordered dimeric structures (PI-PDI4)2 and (PDI5)2 in toluene and
2-methyltetrahydrofuran solutions with the PDI molecules approximately parallel to one another primarily
due to π-π interactions between adjacent PDI chromophores. On hydrophobic surfaces, PDI5 grows into
rod-shaped nanostructures of average length 130 nm as revealed by atomic force microscopy. Photo-
excitation of these supramolecular dimers in solution gives direct evidence of strong π-π interactions
between the excited PDI chromophore and other PDI molecules nearby based on the observed formation
of an excimer-like state in <130 fs with a lifetime of about 20 ns. Multiple photoexcitations of the
supramolecular dimers lead to fast singlet-singlet annihilation of the excimer-like state, which occurs with
exciton hopping times of about 5 ps, which are comparable to those observed in photosynthetic light-
harvesting proteins from green plants.
Introduction
another. The protein structure provides specific distances and
orientations between the chromophores as well as an electronic
The preparation of molecular photonic devices such as solar
cells, light-emitting diodes, organic electronic components, and
sensors frequently requires the assembly of well-ordered materi-
als in which energy and charge transport are rapid and
directional.1-7 For many years photosynthetic antenna and
reaction center proteins have served as important models for
designing chemical systems optimized for energy and electron
transport.1,2 An important feature of these proteins that differs
from most model systems is the fact that none of their energy
or electron transport chromophores are covalently linked to one
environment that is critical to proper functioning of the system.
Over the past few years we have developed several approaches
to controlling charge transport in organic nanostructures that
take advantage of the speed and efficiency of ultrafast photo-
induced electron-transfer reactions.8-11 However, as the size and
complexity of these systems have grown, it has become apparent
that a new strategy based on a combination of synthetic covalent
building blocks and self-assembly has significant advantages
for producing photofunctional materials. We are currently
exploring an approach using functional covalent building blocks
having structures designed to elicit particular self-assembly
motifs to produce supramolecular photofunctional materials. In
this way, we can optimize their energy and charge transport
properties without resynthesizing the complete system. This
approach also offers the possibility that materials can be made
† Northwestern University.
‡ Argonne National Laboratory.
(1) Wasielewski, M. R. Chem. ReV. 1992, 92, 435-461.
(2) Gust, D.; Moore, T. A.; Moore, A. L. Acc. Chem. Res. 2001, 34, 40-48.
(3) Wagner, R. W.; Lindsey, J. S.; Seth, J.; Palaniappan, V.; Bocian, D. F. J.
Am. Chem. Soc. 1996, 118, 3996-3997.
(4) Wu¨rthner, F.; Thalacker, C.; Diele, S.; Tschierske, C. Chem.sEur. J. 2001,
7, 2245-2253.
(8) Gosztola, D.; Niemczyk, M. P.; Wasielewski, M. R. J. Am. Chem. Soc.
1998, 120, 5118-5119.
(5) Aratani, N.; Osuka, A.; Cho, H. S.; Kim, D. J. Photochem. Photobiol., C.
2002, 3, 25-52.
(9) Hayes, R. T.; Wasielewski, M. R.; Gosztola, D. J. Am. Chem. Soc. 2000,
122, 5563-5567.
(6) Jordens, S.; De Belder, G.; Lor, M.; Schweitzer, G.; Van der Auweraer,
M.; Weil, T.; Herrmann, A.; Wiesler, U.-M.; Mu¨llen, K.; De Schryver, F.
C. Photochem. Photobiol. Sci. 2003, 2, 1118-1124.
(7) Furuta, P.; Brooks, J.; Thompson, M. E.; Frechet, J. M. J. J. Am. Chem.
Soc. 2003, 125, 13165-13172.
(10) Lukas, A. S.; Bushard, P. J.; Wasielewski, M. R. J. Am. Chem. Soc. 2001,
123, 2440-2441.
(11) Andersson, M.; Sinks, L. E.; Hayes, R. T.; Zhao, Y.; Wasielewski, M. R.
Angew. Chem., Int. Ed. 2003, 42, 3139-3143.
9
8284
J. AM. CHEM. SOC. 2004, 126, 8284-8294
10.1021/ja039820c CCC: $27.50 © 2004 American Chemical Society

Supramolecular Light-Harvesting Arrays
A R T I C L E S
in which defects can be removed by disassembly-assembly
cycles.
linkages between the porphyrins and the electronic structure of
the two nearly degenerate highest occupied molecular orbitals
of the porphyrin play important roles in determining rates of
energy transfer.27 Dendrimeric systems having many aromatic
molecules, such as polyphenylbenzenes and perylene-3,4-
dicarboximides, have been used to elucidate the details of energy
transport within complex covalent structures as well as the
photophysics of single molecules having multiple chromo-
phores.6,7,31-37 Thus far, studies of energy transport in arrays
consisting of many chromophores have been focused largely
on polymers,38-45 self-assembled structures made from single
chromophores4,46-48 and dendrimeric metal complexes.49-51 We
now report on two antenna arrays based on PDI, a chromophore
that has been well characterized previously along with other
closely related derivatives.47,48,52-58 Unlike our earlier work with
(ZnTPP-PDI4)n arrays, which was focused on efficient electron
transfer, the new systems presented here are designed to explore
molecular architectures to create an efficient energy transport
material. These covalent multi-chromophore arrays, PI-PDI4 and
PDI5, Figure 1, self-assemble into partially ordered photofunc-
tional supramolecular structures using π-π interactions among
the PDI chromophores. Photophysical studies show that energy
transport within these supramolecular antenna structures is both
rapid and efficient.
Covalent syntheses of large, ordered molecular arrays are
usually inefficient and costly, thus making self-assembly the
method of choice to achieve ordered architectures from func-
tional building blocks. Self-assembly can be based on a variety
of weak interactions such as hydrogen bonding12-16 or π-π
interactions.17-19 We recently reported on a photofunctional self-
assembled array consisting of a zinc meso-tetraphenylporphyrin
(ZnTPP) with four perylene-3,4:9,10-bis(dicarboximide) (PDI)
chromophores covalently attached to the para positions of the
phenyl groups (ZnTPP-PDI4).20 This covalent antenna/reaction
center molecule self-assembles into supramolecular structures
having significant order among the ZnTPP-PDI4 building
blocks. Photoinduced electron transfer from the ZnTPP to one
of its attached PDI acceptors occurs in 3 ps and is followed by
electron transport among several neighboring noncovalently
attached PDI molecules within the assembly. The ability to
photogenerate charges with nearly 100% quantum yield and
have the resultant charges readily migrate within a material is
important for its potential utilization in solar cell applications.
Another desirable characteristic of chromophore arrays for
use in solar cells is efficient energy transport. Good multi-
chromophore antenna arrays not only have excellent cross
sections for light absorption but also in addition carry out
internal energy transport rapidly and efficiently to deliver singlet
excitons to sites at which charge separation can occur. Antenna
proteins in photosynthetic organisms fulfill these requirements
using many chlorophyll molecules to transport singlet excitons
to the primary electron donor chlorophyll within a nearby
reaction center protein leading to charge separation in a few
picoseconds.21 Several elegant multi-porphyrin covalent arrays
have been prepared and used to study energy transport.5,22-30
Recent examples involving both planar and linear porphyrin
arrays show that the electronic properties of the covalent
(31) Cotlet, M.; Gronheid, R.; Habuchi, S.; Stefan, A.; Barbafina, A.; Mu¨llen,
K.; Hofkens, J.; De Schryver, F. C. J. Am. Chem. Soc. 2003, 125, 13609-
13617.
(32) Hofkens, J.; Maus, M.; Gensch, T.; Vosch, T.; Cotlet, M.; Koehn, F.;
Herrmann, A.; Mu¨llen, K.; De Schryver, F. J. Am. Chem. Soc. 2000, 122,
9278-9288.
(33) Vosch, T.; Cotlet, M.; Hofkens, J.; Van Biest, K.; Lor, M.; Weston, K.;
Tinnefeld, P.; Sauer, M.; Latterini, L.; Mu¨llen, K.; De Schryver, F. C. J.
Phys. Chem. A 2003, 107, 6920-6931.
(34) He, G. S.; Lin, T.-C.; Cui, Y.; Prasad, P. N.; Brousmiche, D. W.; Serin, J.
M.; Frechet, J. M. J. Opt. Lett. 2003, 28, 768-770.
(35) Brousmiche, D. W.; Serin, J. M.; Frechet, J. M. J.; He, G. S.; Lin, T.-C.;
Chung, S. J.; Prasad, P. N. J. Am. Chem. Soc. 2003, 125, 1448-1449.
(36) Serin, J. M.; Brousmiche, D. W.; Frechet, J. M. J. Chem. Commun. 2002,
2605-2607.
(37) Frechet, J. M. J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 4782-4787.
(38) Das, P. K.; Encinas, M. V.; Scaiano, J. C. J. Photochem. 1980, 12, 357-
361.
(12) Kotera, M.; Lehn, J.-M.; Vigneron, J.-P. Tetrahedron 1995, 51, 1953-
1972.
(13) Sessler, J. L.; Wang, B.; Harriman, A. J. Am. Chem. Soc. 1995, 117, 704-
(39) Tamai, N.; Masuhara, H.; Mataga, N. J. Phys. Chem. 1983, 87, 4461-
714.
4467.
(14) Springs, S. L.; Andrievsky, A.; Kral, V.; Sessler, J. L. J. Porphyrins
Phthalocyanines 1998, 2, 315-325.
(40) Holden, D. A.; Guillet, J. E. Macromolecules 1980, 13, 289-295.
(41) Kamioka, K.; Webber, S. E. Chem. Phys. Lett. 1987, 133, 353-358.
(42) Fox, H. H.; Fox, M. A. Macromolecules 1995, 28, 4570-4576.
(43) Stewart, G. M.; Fox, M. A. J. Am. Chem. Soc. 1996, 118, 4354-4360.
(44) Kiserow, D. J.; Itoh, Y.; Webber, S. E. Macromolecules 1997, 30, 2934-
2940.
(45) Ghaddar, T. H.; Whitesell, J. K.; Fox, M. A. J. Phys. Chem. B 2001, 105,
8729-8731.
(46) Wu¨rthner, F.; Thalacker, C.; Sautter, A. AdV. Mater. 1999, 11, 754-
758.
(15) Arimura, T.; Ide, S.; Sugihara, H.; Murata, S.; Sessler, J. L. New J. Chem.
1999, 23, 977-979.
(16) Berg, A.; Shuali, Z.; Asano-Someda, M.; Levanon, H.; Fuhs, M.; Mo¨bius,
K.; Wang, R.; Brown, C.; Sessler, J. L. J. Am. Chem. Soc. 1999, 121, 7433-
7434.
(17) Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990, 112, 5525-
5534.
(18) Shimomura, M.; Olaf, K.; Ijiro, K. Synth. Met. 1996, 81, 251-257.
(19) Zangmeister, R. A. P.; Smolenyak, P. E.; Drager, A. S.; O’Brien, D. F.;
Armstrong, N. R. Langmuir 2001, 17, 7071-7078.
(20) van der Boom, T.; Hayes, R. T.; Zhao, Y.; Bushard, P. J.; Weiss, E. A.;
Wasielewski, M. R. J. Am. Chem. Soc. 2002, 124, 9582-9590.
(21) Diner, B. A.; Babcock, G. T. AdV. Photosynth. 1996, 4, 213-247.
(22) Ikonen, M.; Guez, D.; Marvaud, V.; Markovitsi, D. Chem. Phys. Lett. 1994,
231, 93-97.
(47) Wu¨rthner, F.; Sautter, A. Chem. Commun. 2000, 445-446.
(48) Wu¨rthner, F.; Thalacker, C.; Sautter, A.; Schartl, W.; Ibach, W.; Hollricher,
O. Chem.sEur. J. 2000, 6, 3871-3886.
(49) Baudin, H. B.; Davidsson, J.; Serroni, S.; Juris, A.; Balzani, V.; Campagna,
S.; Hammarstro¨m, L. J. Phys. Chem. A 2002, 106, 4312-4319.
(50) Balzani, V.; Campagna, S.; Denti, G.; Juris, A.; Serroni, S.; Venturi, M.
Acc. Chem. Res. 1998, 31, 26-34.
(23) Wagner, R. W.; Johnson, T. E.; Lindsey, J. S. J. Am. Chem. Soc. 1996,
118, 11166-11180.
(51) Denti, G.; Campagna, S.; Serroni, S.; Ciano, M.; Balzani, V. J. Am. Chem.
Soc. 1992, 114, 2944-2950.
(24) Seth, J.; Palaniappan, V.; Wagner, R. W.; Johnson, T. E.; Lindsey, J. S.;
Bocian, D. F. J. Am. Chem. Soc. 1996, 118, 11194-11207.
(25) Hsiao, J.-S.; Krueger, B. P.; Wagner, R. W.; Johnson, T. E.; Delaney, J.
K.; Mauzerall, D. C.; Fleming, G. R.; Lindsey, J. S.; Bocian, D. F.;
Donohoe, R. J. J. Am. Chem. Soc. 1996, 118, 11181-11193.
(26) Strachan, J.-P.; Gentemann, S.; Seth, J.; Kalsbeck, W. A.; Lindsey, J. S.;
Holten, D.; Bocian, D. F. J. Am. Chem. Soc. 1997, 119, 11191-11201.
(27) Yang, S. I.; Seth, J.; Balasubramanian, T.; Kim, D.; Lindsey, J. S.; Holten,
D.; Bocian, D. F. J. Am. Chem. Soc. 1999, 121, 4008-4018.
(28) Tsuda, A.; Osuka, A. Science (Washington, DC, United States) 2001, 293,
79-82.
(52) Kirmaier, C.; Hindin, E.; Schwartz, J. K.; Sazanovich, I. V.; Diers, J. R.;
Muthukumaran, K.; Taniguchi, M.; Bocian, D. F.; Lindsey, J. S.; Holten,
D. J. Phys. Chem. B 2003, 107, 3443-3454.
(53) Zhao, Y.; Wasielewski, M. R. Tetrahedron 1999, 40, 7047-7050.
(54) van der Boom, T.; Hayes, R. T.; Zhao, Y.; Bushard, P. J.; Weiss, E. A.;
Wasielewski, M. R. J. Am. Chem. Soc. 2002, 124, 9582-9590.
(55) van der Boom, T.; Evmenenko, G.; Dutta, P.; Wasielewski, M. R. Chem.
Mater. 2003, 15, 4068-4074.
(56) Lukas, A. S.; Zhao, Y.; Miller, S. E.; Wasielewski, M. R. J. Phys. Chem.
B 2002, 106, 1299-1306.
(57) Gosztola, D.; Niemczyk, M. P.; Svec, W.; Lukas, A. S.; Wasielewski, M.
R. J. Phys. Chem. A 2000, 104, 6545-6551.
(29) Tsuda, A.; Osuka, A. J. Inclusion Phenom. Macrocyclic Chem. 2001, 41,
77-81.
(30) Tsuda, A.; Osuka, A. AdV. Mater. (Weinheim, Germany) 2002, 14, 75-
79.
(58) Ego, C.; Marsitzky, D.; Becker, S.; Zhang, J.; Grimsdale, A. C.; Mullen,
K.; MacKenzie, J. D.; Silva, C.; Friend, R. H. J. Am. Chem. Soc. 2003,
125, 437-443.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8285

A R T I C L E S
Ahrens et al.
Figure 1. Structures of PI-PDI4 and PDI5 and reference molecule PDI2, where R ) 3,5-di-tert-butylphenyl and R′ ) 2-ethylhexyl.
filtered through a 0.45 µm PTFE filter prior to measurement, and care
was taken to exclude any dust from the experimental setup. The angle
of detection varied between 90° and 30°, and each sample was averaged
using several hundred scans.
X-ray Scattering Measurements. X-ray scattering measurements
were carried out using the undulator beam line 12-ID at the Advanced
Photon Source (APS), Argonne National Laboratory. The X-ray
scattering instrument utilized a double-crystal Si(111) monochromator
and a two-dimensional mosaic CCD detector.59 The X-ray wavelength
was set at λ ) 1.0 Å, and the sample to detector distances were adjusted
to achieve scattering measured across two different q regions: 0.02
Å-1 < q < 0.3 Å-1 and 0.1 Å-1 < q < 1 Å-1, where q ) (4π/λ)sinθ,
λ is the X-ray wavelength, and 2θ is the scattering angle.
Solid Film Preparation and Characterization. Quartz plates
(Chemglass) were cleaned by immersion in a freshly prepared “piranha”
solution (concentrated H2SO4/H2O2 30% ) 7:3 v/v) at 80 °C for at
least 45 min. Caution! This solution is an extremely strong oxidizing
agent. After cooling to room temperature, the slides were rinsed
repeatedly with deionized (DI) water and subjected to an RCA-type
cleaning procedure (H2O/H2O2 30%/NH4OH ) 5:1:1 v/v/v), sonicating
at room temperature for 45 min. The substrates were then rinsed with
Experimental Section
General Information. The syntheses of PI-PDI4 and PDI5 as well
as the intermediates leading to them are detailed in the Supporting
Information. All solvents were spectrophotometric grade unless other-
wise noted. Toluene was purified by passing it through series of CuO
and alumina columns (GlassContour), while 2-methyltetrahydrofuran
(Aldrich) was purified by passing it through a basic alumina column
immediately prior to use. The sizes of the aggregates in solution were
determined by gel permeation chromatography on a Hewlett-Packard
1100 HPLC system with a 4.6 × 300 mm2 Waters Styragel HR-4E
column. This column covers a molecular weight range from 50 to
100 000. The column was calibrated using polystyrene standards
covering a molecular weight range of 2430-29 300. The sample diluent
and mobile phase were 100% tetrahydrofuran (HPLC grade), and the
standards and samples were passed through a 0.45 µm PTFE syringe
filter prior to analysis. Both sample and standard solutions were
prepared at a concentration of 0.1-0.2% (w/v). A 20 µL injection
volume was used for both standard and sample solutions. An isocratic
separation was performed with a run time of 25 min and a flow rate of
0.3 mL/min. The column temperature was controlled and kept at 40
°C, and the column output was monitored using a diode-array detector.
Dynamic light scattering (DLS) experiments were performed with a
Coulter N4 Plus instrument (632.8 nm light source). All solutions were
(59) Seifert, S.; Winans, R. E.; Tiede, D. M.; Thiyagarajan, P. J. Appl.
Crystallogr. 2000, 782-784.
9
8286 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Supramolecular Light-Harvesting Arrays
A R T I C L E S
DI water and dried in an oven overnight at 115 °C. The freshly cleaned
substrates were placed in an oven dried Schlenk-type reactor. A 50
mM solution of CH3SiCl3 in dry toluene solution was added and reacted
for 5 h at 25 °C. The solution was then removed with a canulla. The
functionalized substrates were washed twice with toluene, acetone, and
methanol and dried under vacuum. Highly ordered pyrolytic graphite
(HOPG) (SPI Supplies, SPI-2 grade) was freshly cleaved using Scotch
tape immediately prior to use. A 10-4 M toluene solution of PDI5 was
either spun cast (Laurell WS-400) at 2000 rpm for 30 s or drop cast
and dried at room temperature and stored in the dark under nitrogen
until examination. The PDI5 films were examined by tapping-mode
atomic force microscopy using a Digital Instruments Multimode
Nanoscope IIIa. The tips used were Nanoprobe TESP-70 with a
cantilever length of 125 µm and a resonance frequency of 278-338
kHz. Optical absorption spectra of the film were recorded on the
instrument described below.
times, and the resulting time constants were averaged. Deviation from
the mean was generally no worse than (5%.
Results and Discussion
Synthesis. The synthesis of each multi-chromophoric array
was carried out in two parts. First, a core molecule was
synthesized; second, peripheral PDI chromophores were at-
tached. This stepwise approach allows for the production of
asymmetric systems that may prove useful in future studies of
energy and/or electron transfer; see Figure 2. The first molecule
contains a modified pyromellitimide (PI) core with four
peripheral PDI chromophores held planar to one another to give
PI-PDI4. The PI core serves as a convenient 4-fold point of
attachment and does not participate in either electron or energy
transfer processes involving the attached PDI molecules. The
second core molecule is itself a PDI modified at the 1 and 7
positions by the direct attachment of p-nitrophenyl groups. The
1,7-diphenyl-PDI core molecule is also surrounded by four PDI
molecules to give PDI5, in which all five PDI molecules lie in
the same plane with respect to the core and each other in a
rhombic structure. Substitution of 3,5-di-tert-butylphenoxy
groups at the 1,7 positions of the peripheral PDI chromophores
increases their solubility significantly. The ground-state absorp-
tion maxima of both 1,7-diphenyl and 1,7-diphenoxy substituted
PDI derivatives shift from 520 nm in the parent molecule to
550 nm.62
The pyromellitimide core was prepared using a reaction
sequence analogous to the known route to 3,6-diphenylpyro-
mellitic dianhydride.63 Treatment of durene with Br2 and a
catalytic amount of iodine in refluxing dichloromethane affords
1,4-dibromodurene, 1, in 78% yield. Palladium-catalyzed cross
coupling of 4-iodonitrobenzene and pinacolborane under basic
conditions yielded nitroboronate ester 2 in 62% yield. This
molecule is used directly in a Suzuki coupling with dibromo-
durene to give the dinitrotetramethylterphenyl 3 in 41% yield.
Complete oxidation of the methyl groups of 3 to carboxylic
acids is done in two stages using KMnO4 in refluxing pyridine
followed by KMnO4 in refluxing aqueous NaOH to give 4 in
84% yield. The final dianhydride is obtained by treating 4 with
glacial acetic acid and acetic anhydride at reflux temperature
to give 5 in 37% yield. The second PDI core molecule is
prepared by Stille coupling of 1,7-dibromoperylene-3,4:9,10-
tetracarboxydianhydride64 with p-(tri-n-butylstannyl)nitro-
benzene in toluene catalyzed by Pd(PPh3)4 to give 13 in 66%
yield.
Optical Spectroscopy. Steady-state absorption and emission spectra
were performed on a Shimadzu 1601 UV/vis spectrophotometer and
PTI single photon counting spectrofluorimeter in a right angle config-
uration, respectively. A 10 mm quartz cuvette was used for both the
absorption and fluorescence measurements, and the optical density at
λmax for the fluorescence measurements was maintained at 0.1 ( 0.05
to avoid reabsorption artifacts.
Femtosecond transient absorption measurements were made using
a regeneratively amplified titanium sapphire laser system operating at
a 1 kHz repetition rate outfitted with a CCD array detector (Ocean
Optics PC2000) for simultaneous collection of spectral and kinetic
data.60 The frequency-doubled output from the laser provides 400 nm,
80 fs pulses for excitation. Focusing a few microjoules of the 800 nm
fundamental into a 1 mm sapphire disk generated a white light
continuum probe pulse. All-reflective optics were used to focus the
800 nm pulse into the sapphire and recollimate the white light output,
thus limiting the chirp on the white light pulse to < 200 fs from 450
to 800 nm. Cuvettes with a 2 mm path length were used, and the
samples were irradiated with 0.5-1.0 µJ per pulse focused to a 200
µm spot. The optical density at λmax was typically 0.4-0.8. The total
instrument response function (IRF) for the pump-probe experiments
was 130 fs. Kinetic analyses were performed at several wavelengths
using a nonlinear least-squares fit to a general sum-of-exponentials using
the Levenberg-Marquardt algorithm while accounting for the presence
of the finite instrument response.
Fluorescence lifetime measurements were made using a Hamamatsu
C4780 picosecond fluorescence lifetime measurement system, consisting
of a C4334 Streakcope and a C4792-01 synchronous delay generator.
The excitation light source was supplied by a home-built cavity-dumped
Ti:Sapphire laser61 with a NEOS N13389 3 mm fused-silica acousto-
optic modulator (AOM). The AOM was driven by a NEOS Technolo-
gies N64389-SYN 10 W driver to deliver 38 nJ, sub-50 fs pulses at an
820 kHz repetition rate. The laser pulses were frequency doubled to
400 nm by focusing the 800 nm fundamental into a 1 mm Type I BBO
crystal. The energy of the resulting blue pulses was attenuated to
approximately 1.0 nJ/pulse for all fluorescence lifetime experiments.
The total IRF of the streak camera system was 25 ps. The samples
were prepared in glass cuvettes, and the optical density at the excitation
wavelength was typically 0.020-0.035. At λmax of the samples, the
optical density was typically 0.080-0.105. All fluorescence data were
acquired in single photon counting mode using the Hamamatsu HPD-
TA software. The data were fit using the Hamamatsu fitting module
and deconvoluted using the laser pulse profile. Two fits were generated
for each experiment, one for the emission intensity integrated between
530 and 640 nm and another for the emission intensity integrated
between 640 and 740 nm. Measurements were generally repeated 3
The synthetic schemes used to attach the peripheral PDI
derivatives to the core molecules differ slightly for each array
because two different functionalized PDI derivatives are needed
for the stepwise synthesis of PI-PDI4, one with an anhydride, 8
to condense with the amine already present on the PI core, and
one with an aniline group already incorporated into PDI, 9 to
condense with the core dianhydride. In the case of PDI5 only
one type of functionalized PDI, 8, is required. 1,7-Dibromo-
perylene-3,4:9,10-tetracarboxydianhydride is reacted with 3,5-
di-tert-butylphenol in refluxing DMF in the presence of Cs2CO3
followed by precipitation of the product from glacial acetic acid
(62) Holtrup, F. O.; Muller, G. R. J.; Quante, H.; Feyter, S. D.; Schryver, F. C.
D.; Mu¨llen, K. Chem.sEur. J. 1997, 3, 219-225.
(63) Schmitz, L.; Rehahn, M.; Balauff, M. Polymer 1993, 34, 646-649.
(64) Bohm, A.; Arms, H.; Henning, G.; Blaschka, P. BASF: German Patent
No. DE 19547209A1, 1997.
(60) Giaimo, J. M.; Gusev, A. V.; Wasielewski, M. R. J. Am. Chem. Soc. 2002,
124, 8530-8531.
(61) Pshenichnikov, M. S.; de Boerij, W. P.; Wiersma, D. A. Opt. Lett. 1994,
19, 572-575.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8287

A R T I C L E S
Ahrens et al.
Figure 2. Structures of intermediates used to prepare PI-PDI4 and PDI5.
to give 7 in 90% yield. The dianhydride is converted to the
monoanhydride-monoimide by reacting 7 with one molar
equivalent of 2-ethylhexylamine in refluxing pyridine. Although
a mixture of monoimide 8, the diimide, and the dianhydride
are obtained, the latter two can be separate easily and are
recycled. Refluxing 8 in pyridine/imidazole with 1,4-diamino-
benzene gives the target chromophore 9 in 83% yield. Reaction
of dinitro compound 5 with amine 9 in the presence of imidazole
in refluxing pyridine for 24 h produces 10 in 27% yield.
Reduction of the nitro groups using SnCl2 in THF under acidic
conditions affords the diamine product 11 in quantitative yield.
Finally, coupling of 11 with 8 in quinoline/zinc acetate at 180
°C for 4.5 days gives the target molecule PI-PDI4, 12, in 88%
yield. Lower temperature solvents such as toluene, pyridine,
and DMF did not produce the target compound, most likely
because aggregation of the intermediate condensation product
with three PDI molecules attached inhibits subsequent conden-
sation of the fourth PDI to the core. Carrying out the condensa-
tion at higher temperatures reduces aggregation of the inter-
mediate allowing better access of 8 to the amine leading to a
higher yield of imide formation.
The synthesis of PDI5 begins with the reaction of dinitro-
dianhydride 13 with 1,4-diaminobenzene in refluxing pyridine/
imidazole for 6 h to give 14 in quantitative yield. Condensation
of the first pair of PDI chromophores is performed by reacting
14 with excess anhydride 8 in pyridine/imidazole for 24 h to
give 15 in 27% yield. Reduction of the nitro groups to the
amines using SnCl2 in THF under acidic conditions affords 16
in quantitative yield. Condensation of the remaining two PDI
chromophores to complete the synthesis of PDI5 is carried out
9
8288 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Supramolecular Light-Harvesting Arrays
A R T I C L E S
range below 0.05 Å-1 (below q2 ≈ 0.003 Å-2) the experimental
scattering data for PDI5 in toluene and THF (note: this is more
pronounced in THF) show concentration-dependent upward
deviations from Guinier behavior that are characteristic of
weakly associating molecular systems.68 This can be linked to
the propensity of PDI5 to form large aggregates in the solid
state as indicated by AFM measurements (see discussion below
and Figure 7). The Rg is not directly related to the molecular
structure but is the electron-density-contrast weighted sum of
the squares of the atomic distances from the center of mass and,
hence, depends on a combination of factors including electron
density contrast with the solvent and structure of the solvent-
excluded and solvent-associated layers.66,67 Thus, for PDI5 in
toluene Rg ) 16.0 ( 0.1 Å, and in THF Rg ) 12.7 ( 0.1 Å.
The variation in Rg values in two solvents may indicate
somewhat dissimilar structures and/or different solvent-solute
interactions in toluene vs THF. It should be noted that Rg values
for PDI5 indicate that it forms small aggregates in toluene and
THF solution.
Figure 3. Guinier fit (solid line) of SAXS data for PDI5 in toluene. Inset:
Guinier fit (solid line) for PDI5 in THF.
by reacting 16 with an excess of 8 in quinoline/zinc acetate at
200 °C for 24 h to give 17 in 29% yield.
The Guinier plot also yields the extrapolated intensity at zero
scattering angle, I(0), which can be used to estimate the
molecular weight of the aggregate by comparison to an I(0)
value of a reference compound of similar structure with known
molecular weight (provided both compounds demonstrate linear
Guinier plots).67 Our reference of choice was a π-stacked dimer
of a 3-fold symmetric tris(PDI) derivative, Figure S4A, which
demonstrates excellent SAXS characteristics (Figures S4B and
S4C), while its dimeric structure was independently established
by a variety of methods (see Supporting Information). The esti-
mated molecular weight of PDI5 is ∼7600 in toluene and ∼7800
in THF, indicating that PDI5 forms dimers in these solvents.
Characterization of Self-Assembled Structures. The mass
spectrum of each chromophoric array was obtained using
MALDI/TOF techniques. In addition to the parent ion (PI-PDI4
) 4154 and PDI5 ) 4328) significant peaks are seen at several
multiples of the molecular weight of each molecule. A distribu-
tion of aggregates with as many as 4 PI-PDI4 and 13 PDI5
molecules are readily seen in their mass spectra (Supporting
Information, Figures S1A and S1B, respectively). These results
are similar to that observed previously by us for ZnTPP-PDI4.20
Once again, these mass spectra give a strong indication of the
tendency of the PDI molecule to self-assemble into larger
structures. The 1H NMR spectra of PI-PDI4 and PDI5 in
deuterated chloroform, DMSO, toluene, and pyridine display
severely broadened lines even at concentrations down to 10-5M,
which indicates that each of these molecules forms aggregates
in these solvents. This is also the case for intermediates 15 and
16. The observed broadening is illustrated for PDI5 in chloro-
form in the Supporting Information, Figure S2.
We used gel permeation chromatography (GPC) to inde-
pendently determine the sizes of the PI-PDI4 and PDI5 ag-
gregates in solution at approximately 10-5 M, the concentration
at which the optical experiments reported here are performed.
The GPC data, Supporting Information Figure S5, show that
the aggregates of both PI-PDI4 and PDI5 are dimers with a
polydispersity of 1.1, which agrees very well with the SAXS
data for PDI5. Dynamic light scattering experiments using a
632.8 nm laser source (Coulter N4 Plus) on solutions of PI-
PDI4 and PDI5 at concentrations as high as 10-3 M in toluene
show only very weak scattering signals, which indicates that,
even at this higher concentration, the aggregate sizes remain at
or below the ∼3 nm size measurement limit of the light
scattering instrument.
To estimate the PDI5 aggregate size we performed SAXS
measurements on toluene and THF solutions using a high-flux
synchrotron source (Advanced Photon Source (APS) at Argonne
National Laboratory; see Experimental Section). The scattering
intensity is a function of the scattering vector, q, which is related
to the scattering angle 2θ by the relation q ) (4π/λ)sinθ, where
λ is the X-ray wavelength, Figure S3. We performed SAXS
experiments in two different q regions: 0.02 Å-1 < q < 0.3
Å-1 and 0.1 Å-1 < q < 1 Å-1, using two different sample-to-
detector distances. In the low resolution scattering region below
q ) 0.1 Å-1, the calculated scattering follows the Guinier
The optical absorption spectra of PI-PDI4 and PDI5 in toluene,
2-methyltetrahydrofuran (MTHF), and chloroform shown in
Figures 4 and 5, respectively, provide evidence for the orienta-
tion of the transition moment axes of the PDI chromophores69
relative to one another within the aggregates. The lowest energy
absorption band of the PDI monomer is found at 550 nm,20
while, in the case of (PI-PDI4)2, there are two bands at 510 and
545 nm and, for (PDI5)2, these bands are at 515 and 550 nm.
Very similar spectra were observed for the fully covalent dimeric
PDI2 reference molecule, Figure 1,20 in which the transition
dipoles for the lowest energy transitions in each of the two PDI
chromophores, which lie along their N-N axes, are constrained
2
relationship,65-67 I(q) ) I(0) exp(-q2Rg /3), parametrized in
terms of the forward scattering amplitude, I(0), and the radius
of gyration, Rg. Guinier plots of the PDI5 data are presented in
Figure 3. The linearity of Guinier plot is a measure of the
monodispersity of the aggregate. The data for PDI5 show
essentially linear plots in both THF and toluene in the range
0.05 Å-1 < q < 0.09 Å-1 (0.003 Å-2 < q2 < 0.008 Å-2),
indicating that the PDI5 aggregates are monodisperse. In the q
(65) Guinier, A.; Fournet, G. Small Angle Scattering; Wiley: New York, 1955.
(66) Glatter, O. In Neutron, X-ray and Light Scattering; Lindner, P., Zemb, T.,
Eds.; Elsevier: Amsterdam, 1991; pp 33-82.
(67) Glatter, O.; Kratky, O. Small-Angle X-ray Scattering; Academic Press:
London, 1982.
(68) Ducruix, A.; Guilloteau, J. P.; Ries-Kautt, M.; Tardieu, A. J. Cryst. Growth
1986, 168, 28-39.
(69) Fritz, K. P.; Scholes, G. D. J. Phys. Chem. B 2003, 107, 10141-10147.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8289

A R T I C L E S
Ahrens et al.
The optical transition from the ground state of (PI-PDI4)2 and
(PDI5)2 to their lower exciton states, which is symmetry
forbidden in the molecular exciton model for H-aggregates, still
has significant oscillator strength as evidenced by the intensity
of the band at 550 nm. Extension of the simple molecular
exciton model to include vibronic coupling in the exciton states
of the dimer relieves the symmetry restrictions inherent to the
simple model.71,72 Thus, the 510 nm band is the transition from
the ground state to the V ) 0 vibronic level of the upper exciton
state, while the 550 nm band is the corresponding transition to
the V ) 1 vibronic level of the lower exciton state. The greater
oscillator strength of the 510 nm transition in PDI is consistent
with the geometry of face-to-face stacking. The presence of the
low energy transition is evident in the spectra of both (PI-PDI4)2
and (PDI5)2. This behavior is also observed in the solvent-
independent spectra of PDI2 in which the orientation of the two
PDI molecules relative to one another is constrained to a parallel,
cofacial geometry by the xanthene spacer.20 Nevertheless, it is
also possible that a small fraction of the residual 550 nm
absorption in the spectra of both (PI-PDI4)2 and (PDI5)2 is due
to residual monomeric PDI molecules in equilibrium with the
aggregates.
Figure 4. Normalized optical absorption spectra PI-PDI4.
For PDI5, we see a strong tendency to aggregate in toluene
and MTHF, and to a lesser degree in chloroform. Aggregation
is driven by strong π-π interactions between adjacent PDI
chromophores. For example, intermediate 15, which contains
three PDI chromophores in a linear array, also exhibits an intense
blue-shifted band at 510 nm (Supporting Information, Figure
S6). Given that the intramolecular chromophores are in an end-
to-end arrangement, the coplanar orientation of the PDI π
systems makes it possible for the PDI molecules to form parallel
stacked aggregates as indicated by the appearance of the 510
nm absorption feature. In the case of PI-PDI4, the same general
trends are observed; aggregate formation is more significant in
toluene and MTHF and less so in chloroform.
Two of the simplest possible stacking motifs for (PI-PDI4)2
and (PDI5)2 are illustrated in Figure 6A and B, respectively.
These structures were calculated using the MM+ force field.73
The PI-PDI4 dimer structure in Figure 6A is rather flat, while
two saddle-shaped PDI5 molecules nest to form the PDI5 dimer
shown in Figure 6B. The average distance between the stacked
PDI molecules in (PI-PDI4)2, as well as between the stacked
peripheral PDI molecules in (PDI5)2, is 3.5 Å. The distance
between the central PDI molecules in (PDI5)2 is 5.1 Å. Although
these structures are not unique due to rotational isomers about
the single bonds joining the peripheral PDI molecules to the
core, the optical spectra require that each PDI chromophore is
stacked on a partner PDI chromophore with their transition
moments parallel.
Figure 5. Normalized optical absorption spectra of PDI5.
to a parallel orientation. If the transition dipoles of two identical
chromophores are positioned in a parallel, stacked geometry,
the molecular exciton model predicts that coupling of the two
transition dipoles will cause the lowest energy electronic
transition of the dimer to split into two bands with the higher-
energy band having all of the oscillator strength.70 The presence
of the intense 510 and 515 nm bands in (PI-PDI4)2 and (PDI5)2
suggests that the PDI molecules in these assemblies adopt a
parallel, stacked geometry. However, the spectra do not preclude
the possibility that any two PDI molecules within the stacks
may deviate somewhat from a strictly cofacial orientation
relative to one another to minimize steric interactions. For
example, sideways slippage of parallel aromatic planes either
along the N-N axis of PDI (y-axis) or along the in-plane axis
perpendicular to it (x-axis) may occur. In addition, a small
degree of tipping of the aromatic planes may occur. Small
slippage of the planes along the x-axis will not alter the spectrum
greatly because the transition moments of the two PDI molecules
remain parallel to one another. Slippage of one PDI molecule
relative to another along the y-axis, as measured by the angle
between a vector normal to the aromatic planes (z-axis) and a
vector joining the centers of the two PDI molecules must be
considerably less than 54.7° (the magic angle) in order to
preserve the dominance of the blue-shifted absorbance.70 Finally,
significant tipping of the aromatic planes relative to one another
along the z-axis will result in a rapid decrease in exciton
coupling due to the 1/r3 dependence of this interaction.
The sizes and structures of aggregates in solution frequently
differ greatly from those deposited as films. We have used
scanning probe microscopy techniques74-76 to examine the
(71) Fulton, R. L.; Gouterman, M. J. Chem. Phys. 1964, 41, 2280-2286.
(72) Oddos-Marcel, L.; Madeore, F.; Bock, A.; Neher, D.; Ferencz, A.; Rengel,
H.; Wegner, G.; Kryschi, C.; Trommsdorff, H. P. J. Phys. Chem. 1996,
100, 11850-11856.
(73) MM+ calculations were performed using HyperChem(TM), Hypercube,
Inc., 1115 NW 1114th Street, Gainesville, FL 32601, USA.
(74) Jonkheijm, P.; Miura, A.; Zdanowska, M.; Hoeben, F.; Feyter, S.;
Schenning, A.; Schryver, F.; Meijer, E. Angew. Chem., Int. Ed. 2004, 43,
74-78.
(70) Kasha, M.; Rawls, H. R.; El-Bayoumi, M. A. Pure Appl. Chem. 1965, 11,
371-392.
(75) Liu, D. F. S.; Grim, P.; Vosch, T.; Grebel-Koehler, D.; Wiesler, U.;
Berresheim, A.; Muellen, K. Schryver, F. Langmuir 2002, 18, 8223-8230.
9
8290 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Supramolecular Light-Harvesting Arrays
A R T I C L E S
Figure 6. MM+ calculated structures of (A) (PDI5)2 and (B) (PI-PDI4)2. Views are along the stacking direction (top) and an axis 90° to that direction (side).
Figure 7. AFM phase contrast images of PDI5. (A) PDI5 spin-cast from toluene onto methylated quartz; (B) PDI5 drop-cast from toluene onto HOPG.
nature of (PDI5)n in the solid state. To this end, thin solid films
of (PDI5)n were prepared on both methylsilylated quartz55 and
freshly cleaved highly ordered pyrolytic graphite (HOPG)
substrates and examined using atomic force microscopy (AFM).
Blank slides of both substrates were used to verify the
smoothness and morphology of each substrate. Spin-cast films
of PDI5 prepared using quartz slides produced columnar
structures with a broad distribution of sizes having an average
height of 138 ( 94 nm and an average width of 2.09 ( 0.89
µm, Figure 7A. No columnar structures were observed on
nonmethylsilylated quartz slides, where substantial dewetting
of the films on the hydrophilic quartz surfaces occurred. Drop-
cast films made on HOPG, Figure 7B, reveal that PDI5 is
substantially more ordered than observed for films on methyl-
silylated quartz. The film appears to consist of packed rod-
shaped nanostructures. When the rods are packed perpendicular
to the surface, they form a 19 ( 4 nm thick layer. The rods
that have grown in directions parallel to the surface are generally
much longer, exhibiting dimensions that average (11 ( 3) ×
(45 ( 10) × (131 ( 28) nm3. The growth of these structures in
the direction perpendicular to the surface is limited relative to
the lengths they can attain when they grow parallel to the
surface. As is usually the case, the solid-state morphology of
these films is affected by surface hydrophobicity, substrate
roughness, sample concentration, solvent choice, and method
of thin-film growth. Spin-cast films on methylsilylated quartz
exhibit ground-state absorption spectra that are similar to those
observed in solution (Supporting Information Figure S7). The
similarity between the spectra of the aggregates in solution and
on the methylsilylated quartz films suggests that the H-
aggregation motif observed in solution is also preserved to a
large degree in the films. The optical absorption spectra of the
HOPG based films could not be obtained due to the opaque
nature of the substrate.
Photophysics of Self-Assembled Structures. Steady-state
fluorescence measurements were performed on solutions of PI-
PDI4 and PDI5 in toluene, MTHF, and chloroform. Each sample
was excited at 500 nm, and rhodamine 640 served as the
standard for measuring emission quantum yields. This standard
was chosen for its well-characterized emission from 550 to 750
nm, which is nearly identical to the wavelength range of the
PDI emission in these molecules. As seen in Figure 8, the
emission of (PI-PDI4)2 in all three solvents is red-shifted
compared to the PDI monomer. In MTHF and toluene, this band
is broadened relative to the PDI monomer. A new red-shifted
emission band appears in both toluene and MTHF, while the
(76) Schenning, A.; Herrikhuyzen, J.; Jonkheijm, P.; Chen, Z.; Wu¨rthner, F.;
Meijer, E. J. Am. Chem. Soc. 2002, 124, 10252-10253.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8291

A R T I C L E S
Ahrens et al.
Table 2. Fluorescence Lifetime Data
530−640 nm
640−740 nm
τ1
τ2
τ1
τ2
compd
solvent
(ns)
A
1
(ns)
A
2
(ns)
A
1
(ns)
A
2
PI-PDI4 chloroform 4.54 1.00
4.53 1.00
PI-PDI4 MTHF
PI-PDI4 toluene
4.40 0.98 17.9 0.02 4.35 0.77 18.9 0.23
4.32 0.98 21.4 0.02 4.12 0.78 21.7 0.22
PDI5
PDI5
PDI5
PDI2
PDI2
PDI2
chloroform 5.58
6.15
MTHF
4.59 0.96 14.8 0.04 4.78 0.65 15.9 0.35
toluene
chloroform
MTHF
4.42 0.94 17.6 0.06 4.67 0.63 20.3 0.37
a
a
a
14.6 1.0
19.7 1.0
22.8 1.0
toluene
a All fluorescence decays of PDI2 are monoexponential.
Figure 8. Normalized fluorescence spectra of PI-PDI4 in toluene (s),
MTHF (- - -), and CHCl3 (-‚-‚-) compared to that of PDI in CHCl3 (‚‚‚‚).
PDI2 (see below). Thus, even though the PDI molecules are
associated in the ground state, and thus the state is not a true
excimer, the interaction between the PDI chromophores becomes
stronger in the excited state.78-80 The measured quantum yields
of emission from this band are significant, Table 1. For example,
the quantum yields of excimer-like emission for (PI-PDI4)2 and
(PDI5)2 in toluene are 0.15 and 0.19, respectively, and are similar
to that observed for PDI2.
Time-resolved fluorescence spectroscopy on PI-PDI4 and
PDI5 was carried out in toluene, MTHF, and chloroform, Table
2. The excitation intensity was kept low (1 nJ/pulse) to avoid
exciton annihilation processes (see below). Time constants down
to about 5 ps can be obtained by deconvolution of the IRF from
the data. The fluorescence lifetime data for both (PI-PDI4)2 and
(PDI5)2 reveal biexponential decays in both toluene and MTHF
and single exponential decays in chloroform. In the 530-640
nm wavelength range, the fluorescence lifetimes of (PI-PDI4)2
and (PDI5)2 in all solvents are dominated by a 4-5 ns
component, which is very similar to that observed for mono-
meric PDI in these solvents. However, in the 640-740 nm
wavelength range, a significant fraction of the total amplitude
corresponds to a much longer emission component, which varies
from 17 to 22 ns. The amplitude of this long component does
not dominate the decay because the radiative rate of the long
wavelength emitting species is smaller than that of the residual
PDI monomers. This results in contamination of the long wave-
length emission band by the tail of the PDI monomer emission.
It is interesting to note that our observed lifetime for the long
wavelength emission at 640-740 nm is very similar to the 17
ns lifetime observed for perylene excimers by Katoh et al..79
Moreover, our own results show that in PDI2 the excimer-like
emission dominates completely and exhibits monoexponential
fluorescence lifetimes that are long (15-23 ns), Table 2. Since
the absorption spectrum of PDI2 is only weakly dependent on
solvent, the small variation in PDI2 fluorescence lifetime and
fluorescence quantum yield is most likely due to small changes
in radiative rate and perhaps a heavy atom effect in CHCl3.
Femtosecond photoexcitation of (PI-PDI4)2 and (PDI5)2 in
toluene with either 400 or 550 nm laser pulses yields similar
transient absorption spectra. Typical spectra for (PDI5)2 are
shown in Figure 10, while that of (PI-PDI4)2 is shown in the
Supporting Information, Figure S8. The ground-state absorption
Figure 9. Normalized fluorescence spectra of PDI5 in toluene (s), MTHF
(- - -), and CHCl3 (-‚-‚-) compared to that of PDI in CHCl3 (‚‚‚‚).
Table 1. Fluorescence Quantum Yields
toluene
MTHF
chloroform
530−640
640−825
530−640
640−825
530−640
640−825
compound
nm
nm
nm
nm
nm
nm
PI-PDI4
PDI5
0.12
0.04
0.15
0.19
0.15
0.07
0.02
0.08
0.13
0.13
0.40
0.15
0.16
0.12
0.06
a
PDI2
PDIb
0.96
0.98
0.86
a PDI2 exhibits a single fluorescence emission band with λmax ) 700-
720 nm. b PDI exhibits a single fluorescence emission band with λmax
550 nm.
)
emission feature in chloroform is very similar to that of
monomeric PDI. This effect is even more pronounced for
(PDI5)2, Figure 9. The red-shifted emission of (PDI5)2 dominates
in both toluene and MTHF, and is even apparent to some degree
in chloroform. The changes in the emission spectra of PI-PDI4
and PDI5 as a function of solvent track the observed changes
in the ground state absorption spectra, where aggregate forma-
tion in chloroform is significantly reduced relative to that in
both toluene and MTHF. The fluorescence quantum yields given
in Table 1 also support this trend, where the highest fluorescence
quantum yields are observed in chloroform.
The large Stokes shift and the width of the new broad
emission feature that appears in Figures 8 and 9 relative to that
of monomeric PDI suggest that this transition is due to formation
of an excimer-like state. A strongly red-shifted emission has
been observed previously in covalent dimers of PDI77 including
(78) Michl, J.; Bonacic-Koutecky, V. Electronic Aspects of Organic Photo-
chemistry; John Wiley and Sons: New York, 1990; pp 274-385.
(79) Katoh, R.; Sinha, S.; Murata, S.; Tachiya, M. J. Photochem Photobiol., A
2001, 145, 23-34.
(80) Stevens, B. In AdVances in Photochemistry; Pitts, J. N., Hammond, G. S.,
Noyes, W. A., Eds.; Wiley-Interscience: 1971; Vol. 8, pp 194-198.
(77) Langhals, H.; Ismael, R. Eur. J. Org. Chem. 1998, 1915-1917.
9
8292 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Supramolecular Light-Harvesting Arrays
A R T I C L E S
Figure 11. Transient absorption kinetics for (PDI5)2 in toluene monitored
at 520 nm following excitation with 400 nm, 80 fs laser pulses having 0.25
(s), 0.5 (- - -), 2.0 (‚‚‚‚‚), and 2.5 (-‚-‚-) µJ per pulse focused to a 200 µm
diameter spot size.
Figure 10. Transient absorption spectra of (PDI5)2 in toluene following
excitation at 400 nm. Inset: kinetics at 715 nm.
spectrum bleaches immediately upon excitation and is ac-
companied by a broad excited-state absorption that extends from
580 nm to past 750 nm. The broad, ill-defined nature of the
580-750 nm absorption is similar to that observed for PDI
within PDI220 and is indicative of a strong interaction between
the excited PDI molecule and its neighboring PDI. The inset
shows the transient absorption kinetics of this band at 715 nm.
The transient absorption kinetics for the recovery of the ground-
state bleach at 510 nm (not shown) match those for the decay
of the excited state at 715 nm. The appearance kinetics for these
features is IRF limited, and the decay kinetics display two
components for (PI-PDI4)2, τ ) 19 ps and >3 ns, as well as for
(PDI5)2, τ ) 17 ps and >3 ns. The transient spectra and kinetics
are independent of whether 400 or 550 nm excitation is used,
so that the fast component is not due to relaxation of an upper
excited state, such as the upper exciton state.
In both (PI-PDI4)2 and (PDI5)2, a significant fraction of the
excited-state population decays to the ground state with τ ) 19
and 17 ps, respectively. The spectra of the ground-state bleaches
do not evolve in time, indicating that the ground-state product
of the photoinduced event is the same as the initial state, i.e.,
the PDI chromophores remain aggregated throughout the
excited-state decay process. It is well-known that ultrafast laser
excitation of dye aggregates including photosynthetic antenna
proteins having large absorption cross sections often leads to
multiple excitations of the chromophore array resulting in
exciton annihilation processes.81-85 To probe whether the rapid
initial transient absorption changes observed for (PI-PDI4)2 and
(PDI5)2 are due to this process, transient absorption kinetics as
a function of the laser pump pulse energy were measured with
400 nm excitation, which are illustrated for (PDI5)2 in Figure
11. The kinetics at 520 nm, normalized at the maximum ∆A,
clearly show that the fraction of molecules that undergo fast
deactivation increases as the pulse energy increases, which is
typical of singlet-singlet exciton annihilation.
the array.86 Dependent on the size of the chromophore array
and the degree of electronic interaction between the chromo-
phores, migration of the singlet excitons within the array leads
to singlet-singlet exciton annihilation as an important decay
pathway for the excitons. When two excitons collide, they can
form a variety of products,85,87,88 the most typical being an upper
excited singlet state and a ground singlet state.81,89 The upper
excited singlet state quickly relaxes back to the lowest excited
singlet state, so the net effect is the loss of one exciton. These
dynamics exhibit classic “predator-prey” type behavior; i.e., the
more excitons that are available, the more annihilations occur.89
However, this quickly depletes the number of excitons available,
so fewer annihilations can occur. Increased pump power creates
more excitons and, thus, more annihilation events, leading to
an increase in the ∆A for this process.81,84 It is important to
note that the conditions for annihilation, multiple excitons in
an array, generally only occur when very high intensity laser
pulses are employed. This behavior is usually not seen in
conventional steady-state measurements at low photon fluxes,
where it would manifest itself as a decrease in fluorescence
quantum yield.
Singlet-singlet annihilation is a second-order kinetic process,
so that the simplest description of excited singlet state decay in
the presence of annihilation is given by eq 1:87,90
d∆A
dt
1
2
-
) γ1∆A + γ2(∆A)2
(1)
where γ1 is the overall rate constant for unimolecular decay of
the excited singlet state and γ2 is the time independent rate
constant for singlet-singlet annihilation defined as the annihila-
tion rate per pair of excitons in a domain and has the dimensions
of s-1 84,91,92
The annihilation rate constant γ2 depends critically
.
on the number of electronically connected chromophores in the
aggregate (the domain size), as well as the strength of the
Laser excitation of large arrays of chromophores frequently
results in the formation of several excited chromophores within
(86) Den Hollander, W. T. F.; Bakker, J. G. C.; van Grondelle, R. Biochim.
Biophys. Acta 1988, 725, 492-507.
(87) Swenberg, C. E.; Geacintov, N. E. Org. Mol. Photophys. 1973, 18, 489-
565.
(81) Bittner, T.; Irrgang, K.-D.; Renger, G.; Wasielewski, M. R. J. Phys. Chem.
1994, 98, 11821-11826.
(88) Khairutdinov, R. F.; Serpone, N. J. Phys. Chem. B 1997, 101, 2602-2610.
(89) Pope, M.; Swenberg, C. E. Electronic Processes in Organic Crystals and
Polymers, 2nd ed.; Oxford University Press: New York, 1999; Vol. 56,
pp 157-161.
(82) Bittner, T.; Viogt, J.; Irrgang, K.-D.; Renger, G. Photochem. Photobiol.
1993, 57, 158-162.
(83) van Grondelle, R. Biochim. Biophys. Acta 1985, 811, 147-195.
(84) Gillbro, T.; Sandstrom, A.; Spanfort, M.; Sundstrom, V.; van Grondelle,
R. Biochim. Biophys. Acta 1988, 934, 369-374.
(85) Geacintov, N. E.; Breton, J. In Biological EVents Probed by Ultrafast Laser
Spectroscopy; Alfano, R. R., Ed.; Academic Press: New York, 1982; pp
158-191.
(90) Paillotin, G.; Swenberg, C. E.; Breton, J.; Geacintov, N. E. Biophys. J.
1979, 25, 513-534.
(91) Paillotin, G.; Swenberg, C. E.; Breton, J.; Geacintov, N. E. Biophys. J.
1979, 25, 513-534.
(92) Barzda, V.; Gulbinas, V.; Kananavicius, R.; Cervinskas, V.; van Amerongen,
H.; van Grondelle, R.; Valkunas, L. Biophys. J. 2001, 80, 2409-2421.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8293

A R T I C L E S
Ahrens et al.
Conclusions
coupling. For example, studies on the LH2 antenna protein from
the purple bacterium Rhodospirillum rubrum found an annihila-
tion rate constant, τa, of 1 × 1012 s-1 for a domain of 30
chromophores.93 More recent studies on this system used a
Two different multi-chromophoric arrays PI-PDI4 and PDI5
based on the PDI chromophore were synthesized and character-
ized. These arrays self-assemble into partially ordered supra-
molecular dimers (PI-PDI4)2 and (PDI5)2 in solution as indicated
by both SAXS and GPC data, where the PDI molecules are
approximately parallel and cofacial to one another. Photoexci-
tation of these supramolecular dimers gives direct evidence of
strong π-π interactions between the excited PDI chromophore
and other PDI molecules nearby based on the observed
formation of an excimer-like state in <130 fs with a lifetime
of about 20 ns. Multiple photoexcitation of these supramolecular
dimers leads to fast singlet-singlet annihilation of the excimer-
like states, which indicates that the electronic interaction among
the chromophores is strong. While singlet-singlet annihilation
is well-known in J-aggregates,88,97,98 relatively few studies have
demonstrated singlet-singlet annihilation in H-aggregates due
to the forbidden nature of the direct transition to ground state.88,99
The population of the nearby excimer-like state in (PI-PDI4)2
and (PDI5)2, followed by its allowed radiative transition to the
ground state, makes the annihilation events observable. Atomic
force microscopy reveals that solutions of (PDI5)2 assemble into
large rod-shaped nanostructures with an average length 130 nm
on flat substrates. The supramolecular PDI antenna structures
presented here are promising as efficient energy transport
materials, a quality that is highly desirable in both solar cell
and light-emitting diode applications.
domain size of 12 and found a rate of 2 × 1012 s-1 94
. Workers
studying the Fenna-Matthews-Olsen antenna protein from the
green sulfur bacterium Chlorobium tepidum found a rate of 1.4
× 1011 s-1 for a domain of 21 units.95 Bittner et al.81 have shown
that τa for the trimeric LHCII light-harvesting protein from
photosystem II of green plants, which contains 21 chlorophyll
a molecules, is 25 ps, which was corroborated by the more
recent study of Bardza et al..92 For both (PI-PDI4)2 and (PDI5)2
in which the exciton can migrate between a small number of
sites, the observed time constant for the annihilation process,
τa, can be related to the site-to-site hopping time constant, τhop
,
using eq 2:96
τa ) 2γ2-1 ) (π-1N ln N + 0.20N - 0.12)τhop
(2)
where N is the number of sites. Equation 2 assumes that every
collision of two excitons results in rapid annihilation, so that τa
is limited by the exciton hopping rate τhop, not by the trapping
rate. In addition, eq 2 assumes the exciton hops between nearest
neighbor sites on a square lattice with equal probability. Since
the excimer-like state in both (PDI4-PI)2 and (PDI5)2 forms
within the 130 fs instrument response function of the experiment,
we will assume that the migrating exciton is initially delocalized
between two adjacent π-stacked PDI molecules within each
dimer. In that case, the exciton hops between four sites within
(PDI4-PI)2 and five sites within (PDI5)2. Substituting N ) 4
and 5 and the 19 and 17 ps annihilation time constants for
Acknowledgment. This paper is dedicated to Professor
Frederick D. Lewis on the occasion of his 60th birthday. This
work was supported by the National Science Foundation (CHE-
0102351) and the Office of Naval Research (N00014-02-1-
0381). Work at the Argonne National Laboratory was supported
by the Division of Chemical Sciences, Office of Basic Energy
Sciences, U.S. Department of Energy under Contract W-31-
109-Eng-38. The authors wish to thank Emily Weiss for many
helpful discussions, Prof. P. Messersmith for the use of his DLS
apparatus, and Ms. Jennifer Bash for carrying out the GPC
studies.
(PDI4-PI)2 and (PDI5)2, respectively, into eq 2 yields τhop
)
7.8 ps and τhop ) 4.9 ps for (PDI4-PI)2 and (PDI5)2, respec-
tively. These hopping times are very similar to those estimated
for the trimeric LHC-II photosynthetic light harvesting antenna
complex containing 21 chromophores, where typically τhop
=
4-5 ps (for a recent summary, see ref 92). It is most likely that
the value of τhop differs for exciton hops between sites that are
covalently bonded versus those that are not. Moreover, different
covalent connections between the PDI molecules will also result
in different values of τhop. In the work presented here our
experimental data cannot distinguish between these routes. In
future studies we will explore specific structures designed to
differentiate between these pathways.
Supporting Information Available: Details regarding the
synthesis and characterization of the molecules used in this
study. This material is available free of charge via the Internet
at http://pubs.acs.org.
JA039820C
(93) Deinum, G.; Aartsma, T. J.; van Grondelle, R.; Amesz, J. Biochim. Biophys.
Acta 1989, 976, 63-69.
(96) Montroll, E. W. J. Math. Phys. 1969, 10, 753-765.
(97) Ozcelik, S.; Akins, D. L. J. Phys. Chem. B 1997, 101, 3021-3021.
(98) Holzer, W.; Penzkofer, A.; Stockman, R.; Meysel, H.; Liebegott, H.;
Horhold, H. H. Synth. Met. 2002, 125, 343-357.
(94) Valkunas, L.; Trinkunas, G.; Liuolia, V.; van Grondelle, R. Biophys. J.
1995, 69, 1117-1129.
(95) Gulbinas, V.; Valkunas, L.; Kuciauskas, D.; Katilius, E.; Liuolia, V.; Zhou,
W.; Blankenship, R. E. J. Phys. Chem. 1996, 100, 17950-17956.
(99) Ern, J.; Bock, A.; Oddos-Marcel, L.; Rengel, H.; Wegner, G.; Trommsdorff,
H. P.; Kryschi, C. J. Phys. Chem. A 1999, 103, 2446-2450.
9
8294 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004