Ordered and oriented supramolecular n/p-heterojunction surface architectures: Completion of the primary color collection

Article abstract of DOI:10.1021/ja9030648

In this study, we describe synthesis, characterization, and zipper assembly of yellow p-oligophenyl naphthalenediimide (POP-NDI) donor-acceptor hybrids. Moreover, we disclose, for the first time, results from the functional comparison of zipper and layer-by-layer (LBL) assembly as well as quartz crystal microbalance (QCM), atomic force microscopy (AFM), and molecular modeling data on zipper assembly. Compared to the previously reported blue and red NDIs, yellow NDIs are more π-acidic, easier to reduce, and harder to oxidize. The optoelectronic matching achieved in yellow POP-NDIs is reflected in quantitative and long-lived photoinduced charge separation, comparable to their red and much better than their blue counterparts. The direct comparison of zipper and LBL assemblies reveals that yellow zippers generate more photocurrent than blue zippers as well as LBL photosystems. Continuing linear growth found in QCM measurements demonstrates that photocurrent saturation at the critical assembly thickness occurs because more charges start to recombine before reaching the electrodes and not because of discontinued assembly. The found characteristics, such as significant critical thickness, strong photocurrents, large fill factors, and, according to AFM images, smooth surfaces, are important for optoelectronic performance and support the existence of highly ordered architectures.

Full text of DOI:10.1021/ja9030648

Published on Web 07/17/2009
Ordered and Oriented Supramolecular n/p-Heterojunction
Surface Architectures: Completion of the Primary Color
Collection
Ravuri S. K. Kishore,† Oksana Kel,‡ Natalie Banerji,‡ Daniel Emery,†
Guillaume Bollot,† Jiri Mareda,† Alberto Gomez-Casado,¶ Pascal Jonkheijm,¶
Jurriaan Huskens,¶ Plinio Maroni,§ Michal Borkovec,§ Eric Vauthey,*,‡
Naomi Sakai,*,† and Stefan Matile*,†
Departments of Organic, Inorganic and Analytical, and Physical Chemistry, UniVersity of
GeneVa, GeneVa, Switzerland, and Molecular Nanofabrication Group, UniVersity of Twente,
Enschede, The Netherlands
Received April 23, 2009; E-mail: eric.vauthey@unige.ch; naomi.sakai@unige.ch; stefan.matile@unige.ch
Abstract: In this study, we describe synthesis, characterization, and zipper assembly of yellow p-oligophenyl
naphthalenediimide (POP-NDI) donor-acceptor hybrids. Moreover, we disclose, for the first time, results
from the functional comparison of zipper and layer-by-layer (LBL) assembly as well as quartz crystal
microbalance (QCM), atomic force microscopy (AFM), and molecular modeling data on zipper assembly.
Compared to the previously reported blue and red NDIs, yellow NDIs are more π-acidic, easier to reduce,
and harder to oxidize. The optoelectronic matching achieved in yellow POP-NDIs is reflected in quantitative
and long-lived photoinduced charge separation, comparable to their red and much better than their blue
counterparts. The direct comparison of zipper and LBL assemblies reveals that yellow zippers generate
more photocurrent than blue zippers as well as LBL photosystems. Continuing linear growth found in QCM
measurements demonstrates that photocurrent saturation at the critical assembly thickness occurs because
more charges start to recombine before reaching the electrodes and not because of discontinued assembly.
The found characteristics, such as significant critical thickness, strong photocurrents, large fill factors, and,
according to AFM images, smooth surfaces, are important for optoelectronic performance and support the
existence of highly ordered architectures.
Introduction
efficient exciton dissociation without losses in charge mobil-
ity, bicontinuous electron-(e-) and hole (h+)-transporting
Directional energy, electron and hole transport along
sophisticated multicomponent donor-acceptor cascades, is
essential in biological photosynthesis.1 In bilayer organic
solar cells, electrons and holes generated by exciton dis-
sociation at the interface of n- and p-semiconductors travel
through gradient-free bulk layers.2a In bulk n/p-heterojunction
(BHJ) solar cells, the active interface is increased to maximize
the efficiency of exciton dissociation.2-5 Improved exciton
dissociation at increased interfaces usually comes at the cost
of decreased hole and electron mobility in less continuous
and organized n- and p-semiconducting materials. To achieve
channels that are aligned coaxially on the molecular level
are desired. This ideal architecture has been referred to as
supramolecular n/p-heterojunction (SHJ). Contrary to the
advanced situation with covalent donor-acceptor architec-
tures,6 synthetic “bottom-up” access to SHJs7-10 remains,
however, as challenging as the installation of redox
gradients.11-16 In the bioinspired ideal scenario, electrons
and holes generated by photoinduced charge separation (PCS)
would be rapidly carried away in opposite directions along
oriented multicolored antiparallel redox gradients (OMARGs)
in the adjacent n- and p-channels of SHJs. Ideal OMARG-
SHJ architectures can be expected to give not only maximal
short circuit photocurrents but also maximal fill factors5a as
well as open circuit voltages11,14a (see below).
† Department of Organic Chemistry, University of Geneva.
‡ Department of Physical Chemistry, University of Geneva.
§ Department of Inorganic and Analytical Chemistry, University of
Geneva.
¶ Molecular Nanofabrication Group, University of Twente.
(1) (a) Deisenhofer, J.; Michel, H. Science 1989, 245, 1463–1473. (b)
Nelson, N.; Ben-Shem, A. Nat. ReV. Mol. Cell Biol. 2004, 5, 971–
982. (c) Carmeli, I.; Frolov, L.; Carmeli, C.; Richter, S. J. Am. Chem.
Soc. 2007, 129, 12352–12353. (d) Barber, J. Chem. Soc. ReV. 2009,
38, 185–196.
(3) (a) Lee, J. K.; Ma, W. L.; Brabec, C. J.; Yuen, J.; Moon, J. S.; Kim,
J. Y.; Lee, K.; Bazan, G. C.; Heeger, A. J. J. Am. Chem. Soc. 2008,
130, 3619–3623. (b) Yang, X.; Loos, J.; Veenstra, S. C.; Verhees,
W. J. H.; Wienk, M. M.; Kroon, J. M.; Michels, M. A. J.; Janssen,
R. A. J. Nano Lett. 2005, 5, 579–583.
(2) (a) Tang, C. W. Appl. Phys. Lett. 1986, 48, 183–185. (b) Yu, G.; Gao,
J.; Hummelen, J. C.; Wudl, F.; Heeger, A. J. Science 1995, 270, 1789–
1791. (c) Thompson, B. C.; Fre´chet, J. M. J. Angew. Chem., Int. Ed.
2008, 47, 58–77. (d) Gnes, S.; Neugebauer, H.; Sariciftci, N. S. Chem.
ReV. 2007, 107, 1324–1338.
(4) Ma, C.; Fonrodona, M.; Schikora, M. C.; Wienk, M. M.; Janssen,
R. A. J.; Ba¨uerle, P. AdV. Funct. Mater. 2008, 18, 3323–3331.
(5) (a) Yang, F.; Shtein, M.; Forrest, S. Nat. Mater. 2005, 4, 37–41. (b)
O’Regan, B.; Gra¨tzel, M. Nature 1991, 353, 737–741.
9
11106 J. AM. CHEM. SOC. 2009, 131, 11106–11116
10.1021/ja9030648 CCC: $40.75  2009 American Chemical Society

Supramolecular n/p-Heterojunction Surface Architectures
A R T I C L E S
Figure 1. Frontier orbital energy levels of NDIs (general structures C2-G), POPs (general structure as in Figure 2), chlorophyll, and ubiquinone (solid lines,
HOMO; dashed lines, LUMO); dashed arrows, absorption of light (hν) with wavelength (nm) of maximal absorption (top) and emission (bottom); data from
refs 20-24, 33, and this report.
Recently, we have introduced zipper assembly as a new
approach toward OMARG-SHJs and other applications in
molecular optoelectronics.8 Zipper architectures are composed
of e--transporting π-stacks aligned along h+-transporting strings
of rods. The π-acidic 1,4,5,8-naphthalenediimides (NDIs)17-26
have been selected as π-stacks in zipper architectures because
they can change spectral and redox properties without global
structural changes (Figure 1).9,20-23 Compared to the many other
possible multicolor collections (e.g., perylenediimides, Alexa
Fluors, fluoresceins, BODIPYs, cyanines, porphyrinoids, or
triangular dehydrobenzo[12]annulenes),27-30 the spectacular
combination of (1) decreasing HOMO/LUMO energies with
increasing HOMO/LUMO energy differences (Figure 1),24-26
(2) full coverage of the visible range, (3) global structural
preservation, (4) n-semiconductivity, (5) π-acidity, (6) planarity,
(7) compactness (“atom efficiency”), and (8) synthetic acces-
sibility offered by NDIs remains unique. Multicolor redox
(6) (a) Balzani, V.; Venturi, M.; Credi, A. Molecular DeVices and
Machines; Wiley-VCH: Weinheim, Germany, 2003. (b) Fukuzumi,
S. Bull. Chem. Soc. Jpn. 2006, 79, 177–195. (c) Wasielewski, M. R.
J. Org. Chem. 2006, 71, 5051–5066. (d) Hambourger, M.; Moore,
G. F.; Kramer, D. M.; Gust, D.; Moore, A. L.; Moore, T. A. Chem.
Soc. ReV. 2009, 38, 25–35. (e) Mart´ın, N.; Sa´nchez, L.; Herranz, M. A.;
Illescas, B.; Guldi, D. M. Acc. Chem. Res. 2007, 40, 1015–10124. (f)
Clifford, J. N.; Gu, T.; Nierengarten, J.-F.; Armaroli, N. Photochem.
Photobiol. Sci. 2006, 5, 1165–1172. (g) Imahori, H.; Mori, Y.; Matano,
Y. J. Photochem. Photobiol. C: Photochem. ReV. 2003, 4, 51–83.
(7) (a) 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.
(b) Yamamoto, Y.; Fukushima, T.; Suna, Y.; Ishii, N.; Saeki, A.; Seki,
S.; Tagawa, S.; Taniguchi, M.; Kawai, T.; Aida, T. Science 2006, 314,
1761–1764. (c) Wu¨rthner, F.; Chen, Z.; Hoeben, F. J. M.; Osswald,
P.; You, C.-C.; Jonkheijm, P.; Herrikhuyzen, J.; Schenning, A. P. H. J.;
van der Schoot, P. P. A. M.; Meijer, E. W.; Beckers, E. H. A.; Meskers,
S. C. J.; Janssen, R. A. J. J. Am. Chem. Soc. 2004, 126, 10611–10618.
(d) Yamamoto, Y.; Fukushima, T.; Saeki, A.; Seki, S.; Tagawa, S.;
Ishii, N.; Aida, T. J. Am. Chem. Soc. 2007, 129, 9276–9277.
(8) Sakai, N.; Sisson, A. L.; Bu¨rgi, T.; Matile, S. J. Am. Chem. Soc. 2007,
129, 15758–15759.
(17) (a) Bhosale, S. V.; Jani, C. H.; Langford, S. J. Chem. Soc. ReV. 2008,
37, 331–342. (b) Chaignon, F.; Falkenstrom, M.; Karlsson, S.; Blart,
E.; Odobel, F.; Hammarstro¨m, L. Chem. Commun. 2007, 42, 64–66.
(c) Gao, X.; Qiu, W.; Yang, X.; Liu, Y.; Wang, Y.; Zhang, H.; Qi,
T.; Liu, Y.; Lu, K.; Du, C.; Shuai, Z.; Yu, G.; Zhu, D. Org. Lett.
2007, 9, 3917–3920. (d) Lokey, R. S.; Iverson, B. L. Nature 1995,
375, 303–305. (e) Gabriel, G. J.; Iverson, B. L. J. Am. Chem. Soc.
2002, 124, 15174–15175. (f) Pantos, G. D.; Wietor, J. L.; Sanders,
J. K. Angew. Chem., Int. Ed. 2007, 46, 2238–2240. (g) Gabutti, S.;
Knutzen, M.; Neuburger, M.; Schull, G.; Berndt, R.; Mayor, M. Chem.
Commun. 2008, 2370–2372. (h) Iijima, T.; Vignon, S. A.; Tseng, H. R.;
Jarrosson, T.; Sanders, J. K. M.; Marchioni, F.; Venturi, M.; Apostoli,
E.; Balzani, V.; Stoddart, J. F. Chem.sEur. J. 2004, 10, 6375–6392.
(i) Mukhopadhyay, P.; Iwashita, Y.; Shirakawa, M.; Kawano, S.; Fujita,
N.; Shinkai, S. Angew. Chem., Int. Ed. 2006, 45, 1592–1595. (j)
Tanaka, H.; Litvinchuk, S.; Tran, D.-H.; Bollot, G.; Mareda, J.; Sakai,
N.; Matile, S. J. Am. Chem. Soc. 2006, 128, 16000–16001. (k)
Hagihara, S.; Tanaka, H.; Matile, S. J. Am. Chem. Soc. 2008, 130,
5656–5657. (l) Talukdar, P.; Bollot, G.; Mareda, J.; Sakai, N.; Matile,
S. J. Am. Chem. Soc. 2005, 127, 6528–6529. (m) Talukdar, P.; Bollot,
G.; Mareda, J.; Sakai, N.; Matile, S. Chem.sEur. J. 2005, 11, 6525–
6532.
(9) Sisson, A. L.; Sakai, N.; Banerji, N.; Fu¨rstenberg, A.; Vauthey, E.;
Matile, S. Angew. Chem., Int. Ed. 2008, 47, 3727–3729.
(10) (a) Kira, A.; Umeyama, T.; Matano, Y.; Yoshida, K.; Isoda, S.; Park,
J. K.; Kim, D.; Imahori, H. J. Am. Chem. Soc. 2009, 131, 3198–3200.
(b) Snaith, H. J.; Whiting, G. L.; Sun, B.; Greenham, N. C.; Huck,
W. T. S.; Friend, R. H. Nano Lett. 2005, 5, 1653–1657.
(11) Sista, S.; Yao, Y.; Yang, Y.; Tang, M. L.; Bao, Z. Appl. Phys. Lett.
2007, 91, 223508/1–223508/3.
(12) (a) Yamada, H.; Imahori, H.; Nishimura, Y.; Yamazaki, I.; Ahn, T. K.;
Kim, S. K.; Kim, D.; Fukuzumi, S. J. Am. Chem. Soc. 2003, 125,
9129–9139. (b) Takeda, T.; Lin, C.; Majima, T. Angew. Chem., Int.
Ed. 2007, 46, 6681–6683. (c) Kimura, S. Org. Biomol. Chem. 2008,
6, 1143–1148.
(13) (a) Mwaura, J. K.; Pinto, M. R.; Witker, D.; Ananthakrishnan, N.;
Schanze, K. S.; Reynolds, J. R. Langmuir 2005, 21, 10119–10126.
(b) Fushimi, T.; Oda, A.; Ohkita, H.; Ito, S. Langmuir 2005, 21, 1584–
1589. (c) Zhao, L.; Ma, T.; Bai, H.; Lu, G.; Li, C.; Shi, G. Langmuir
2008, 24, 4380–4387.
(18) (a) Gorteau, V.; Bollot, G.; Mareda, J.; Perez-Velasco, A.; Matile, S.
J. Am. Chem. Soc. 2006, 128, 14788–14789. (b) Gorteau, V.; Bollot,
G.; Mareda, J.; Matile, S. Org. Biomol. Chem. 2007, 5, 3000–3012.
(19) Mareda, J.; Matile, S. Chem.sEur. J. 2009, 15, 28–37.
(20) (a) Wu¨rthner, F.; Ahmed, S.; Thalacker, C.; Debaerdemaeker, T.
Chem.sEur. J. 2002, 8, 4742–4750. (b) Thalacker, C.; Roger, C.;
Wu¨rthner, F. J. Org. Chem. 2006, 71, 8098–8105.
(14) (a) Abdelrazzaq, F. B.; Kwong, R. C.; Thompson, M. E. J. Am. Chem.
Soc. 2002, 124, 4796–4803. (b) Lahav, M.; Heleg-Shabtai, V.;
Wasserman, J.; Katz, E.; Willner, I.; Du¨rr, H.; Hu, Y.-Z.; Bossmann,
S. H. J. Am. Chem. Soc. 2000, 122, 11480–11487.
(15) (a) Guldi, D. M. J. Phys. Chem. B 2005, 109, 11432–11441. (b)
Martinson, A. B. F.; Massari, A. M.; Lee, S. J.; Gurney, R. W.; Splan,
K. E.; Hupp, J. T.; Nguyen, S. T. J. Electrochem. Soc. 2006, 153,
A527–A532. (c) Morisue, M.; Yamatsu, S.; Haruta, N.; Kobuke, Y.
Chem.sEur. J. 2005, 11, 5563–5574. (d) Huang, C.-H.; McClenaghan,
N. D.; Kuhn, A.; Bravic, G.; Bassani, D. M. Tetrahedron 2006, 62,
2050–2059.
(21) Ro¨ger, C.; Wu¨rthner, F. J. Org. Chem. 2007, 72, 8070–8075.
(22) Blaszczyk, A.; Fischer, M.; von Ha¨nisch, C.; Mayor, M. HelV. Chim.
Acta 2006, 89, 1986–2005.
(23) Bhosale, S.; Sisson, A. L.; Talukdar, P.; Fu¨rstenberg, A.; Banerji, N.;
Vauthey, E.; Bollot, G.; Mareda, J.; Ro¨ger, C.; Wu¨rthner, F.; Sakai,
N.; Matile, S. Science 2006, 313, 84–86.
(16) Decher, G. Science 1997, 277, 1232–1237.
9
J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009 11107

A R T I C L E S
Kishore et al.
Figure 2. Structure of yellow POP-Y dyads 1 and 2 and their zipper assembly with initiator 3 on gold substrates compared to the structure of the previously
reported blue POP-B dyads 4 and 5 and the red POP-RCl dyads 6 and 7 (all suprastructures are speculative representations that are consistent with experimental
data, 6 and 7 are mixtures of 2,6- and 3,7-regioisomers). (a) Inter- and intralayer interdigitation of POPs and NDIs, respectively, is thought to be driven by
the formation of NDI π-stacks and controlled by intrastack hydrogen-bonded chains and interstack ion pairing.
gradients obtained with quantum dots appear less suitable
because decreasing HOMO with increasing LUMO levels
produce low-energy electron traps.31
p-Oligophenyls (POPs) have been selected as p-semiconduct-
ing32 rigid-rod scaffolds in zipper architectures.9,20-23 Zipper
assembly of the new yellow POP-NDI hybrids 1 and 2 begins
9
11108 J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009

Supramolecular n/p-Heterojunction Surface Architectures
A R T I C L E S
with short p-quaterphenyls 3. These initiators 3 are equipped
with anionic NDI acceptors along the scaffold and disulfide tails
for covalent immobilization on gold (Figure 2). Incubation of
Au-3 monolayers with cationic p-octiphenyl propagators 1 is
expected to give interdigitating Au-3-1 bilayers, where the
lower half of the cationic NDIs of propagator 1 π-stacks with
the anionic NDI acceptors of initiator 3 and the upper half
remains free as a “sticky end” to zip up with the lower half of
the anionic propagator 2. The sticky ends in Au-3-1-2 would
then be available to zip up with the lower half of cationic
propagator 2, and so on. In the resulting POP-NDI zipper
architectures Au-3-(1-2-)n, the parallel alignment of e--
transporting NDI stacks along strings of h+-transporting POP
rods perpendicular to the electrode surface represents the ideal
architecture of an SHJ photosystem.
Results with blue zippers Au-3-(4-5-)n and red zippers
Au-3-(6-7-)n supported the existence of zipper architectures,
their functional significance, and their compatibility with
OMARG-SHJ architectures (Figure 2).8,9,34 However, the
HOMO of the blue NDI (B) is too high, and that of the red
NDI (RCl) is barely sufficient to accept electrons from POPs.
To improve optoelectronic matching in rod-stack SHJ archi-
tectures, either the HOMO of the rod had to be raised or the
HOMO of the stack had to be lowered. Yellow NDIs with their
low HOMO levels were ideal to explore the latter possibility.
The introduction of yellow zippers was of particular interest
to complete the primary color collection of POP-NDI zippers.
Their availability would already suffice to build mixed OMARG-
SHJ architectures that absorb most visible light and have an
oriented three-component “bluefredfyellow” redox gradient
in the e--transporting channel (Figure 1). In the following, we
describe the synthesis and complete characterization of yellow
POP-NDI zippers. Moreover, for the first time, we directly
Figure 3. Electrostatic potential surfaces and Qzz (MP2/6-311G**) of NDIs
8-10 (top, (30.0 kcal/mol; blue positive, red negative) and face-to-face
dimer of 9 (bottom, d ) 3.36 Å).
compare zipper and layer-by-layer (LBL) assembly and report
quartz crystal microbalance (QCM), atomic force microscopy
(AFM), and molecular modeling data on zipper assembly.
(24) Jones, B. A.; Facchetti, A.; Wasielewski, M. R.; Marks, T. J. J. Am.
Chem. Soc. 2007, 129, 15259–15278.
Results and Discussion
(25) (a) Miller, L. L.; Mann, K. R. Acc. Chem. Res. 1996, 29, 417–423.
(b) Katz, H. E.; Lovinger, A. J.; Johnson, J.; Kloc, C.; Siegrist, T.;
Li, W.; Lin, Y. Y.; Dodabalapur, A. Nature 2000, 404, 478–481. (c)
Yan, H.; Chen, Z.; Zheng, Y.; Newman, C.; Quinn, J. R.; Do¨tz, F.;
Kastler, M.; Facchetti, A. Nature 2009, 457, 679–686.
Molecular Modeling. Dichloro-NDI 8, dimethoxy-NDI 9, and
dimethylamino-NDI 10 were selected as tractable model systems
to compute the effect of increasingly electron-donating core
substituents on the nature of the naphthalene core (Figure 3).
Electrostatic potential surfaces were computed using the MP2/
6-311G** method.36 The obtained aromatic surfaces range from
weakly positive (light green) with potent diamino π-donors in
10 and more distinctly positive with the weaker dialkoxy donors
in 9 to deep blue with dichloro substituents in 8. As expected
from crystal structures,37 the optimized structures of NDIs were
totally planar, independent of the nature of the core substituent.
(26) Warman, J. M.; de Haas, M. P.; Dicker, G.; Grozema, F. C.; Piris, J.;
Debije, M. G. Chem. Mater. 2004, 16, 4600–4609.
(27) (a) Wu¨rthner, F. Chem. Commun. 2004, 1564–1579. (b) Shaller, A. D.;
Wang, W.; Gan, H.; Li, A. D. Q. Angew. Chem., Int. Ed. 2008, 47,
7705–7709. (c) Perez-Velasco, A.; Gorteau, V.; Matile, S. Angew.
Chem., Int. Ed. 2008, 47, 921–923. (d) Peneva, K.; Mihov, G.; Nolde,
F.; Rocha, S.; Hotta, J.; Braeckmans, K.; Hofkens, J.; Uji-i, H.;
Herrmann, A.; Mu¨llen, K. Angew. Chem., Int. Ed. 2008, 47, 3372–
3375.
(28) (a) Miller, R. A.; Presley, A. D.; Francis, M. B. J. Am. Chem. Soc.
2007, 129, 3104–3109. (b) Koeppe, R.; Bossart, O.; Calzaferri, G.;
Sariciftci, N. S. Sol. Energy Mater. Sol. Cells 2007, 91, 986–995.
(29) Haugland, R. P. The Handbook. A Guide to Fluorescent Probes and
Labeling Techniques, 10th ed.; Invitrogen: Eugene, OR, 2005.
(30) (a) Rio, Y.; Rodriguez-Morgade, M. S.; Torres, T. Org. Biomol. Chem.
2008, 6, 1877–1894. (b) Tahara, K.; Fujita, T.; Sonoda, M.; Shiro,
M.; Tobe, Y. J. Am. Chem. Soc. 2008, 130, 14339–14345. (c) Lopalco,
M.; Koini, E. N.; Cho, J. K.; Bradley, M. Org. Biomol. Chem. 2009,
7, 856–859.
The global quadrupole moments Qzz of NDIs 8-10 were
computed using the MP2/6-311G** method. The Qzz ) +18.0
B found for dichloro NDI 8 was in the range of unsubstituted
NDIs.18b Diamino NDIs 10 remained with Qzz ) +2.3 B, weakly
π-acidic. The π-acidity of the intermediate dialkoxy NDI 9 was
with Qzz ) +8.0 B, close to that of hexafluorobenzene.18 The
magnitude of this quadrupole moment suggested that, with
interdigitating POP-Y dyads 1 and 2, the conductivity of the
resulting π-stacks could be very high.
(31) (a) Weiss, E. A.; Porter, V. J.; Chiechi, R. C.; Geyer, S. M.; Bell,
D. C.; Bawendi, M. G.; Whitesides, G. M. J. Am. Chem. Soc. 2008,
130, 83–92. (b) Kongkanand, A.; Tvrdy, K.; Takechi, K.; Kuno, M.;
Kamat, P. V. J. Am. Chem. Soc. 2008, 130, 4007–4015.
(32) Khanna, R. K.; Jiang, Y. M.; Creed, D. J. Am. Chem. Soc. 1991, 113,
5451–5453.
(33) Sakai, N.; Kishore, R. S. K.; Matile, S. Org. Biomol. Chem. 2008, 6,
3970–3976.
(36) (a) Moller, C.; Plesset, M. S. Phys. ReV. 1934, 46, 618–622. (b)
Headgordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett. 1988,
153, 503–506.
(34) Lista, M.; Sakai, N.; Matile, S. Supramol. Chem. 2009, 21, 238–244.
(35) Banerji, N.; Fu¨rstenberg, A.; Bhosale, S.; Sisson, A. L.; Sakai, N.;
Matile, S.; Vauthey, E. J. Phys. Chem. B 2008, 112, 8912–8922.
(37) Kishore, R. S. K.; Ravikumar, V.; Bernardinelli, G.; Sakai, N.; Matile,
S. J. Org. Chem. 2008, 73, 738–740.
9
J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009 11109

A R T I C L E S
Kishore et al.
Scheme 1a
a Reagents and conditions: (a) 1. dibromoisocyanuric acid; 2. EtI, EtOH, K2CO2, 35%;37 (b) NaOEt, 73%;37 (c) 1. KOH, i-PrOH; 2. 18, 19, AcOH,
24%;37 (d) PdCl2(PPh3)4, Bu3SnH, p-nitrophenol; (e) PdCl2(PPh3)4, PhSiH3 (crude); (f) HATU, di-t-Bu-pyridine, TEA, DMF (43% from 20); (g) HBr/AcOH,
TFA, thiophenol, pentamethylbenzene (quant); (h) 1. KOH, i-PrOH; 2. 18, 27, AcOH, 54%; (i) PdCl2(PPh3)4, PhSiH3 (crude); (j) HATU, di-t-Bu-pyridine,
TEA, DMF (29% from 28); (k) TFA, CH2Cl2 (quant); (l) several steps, as in ref 43; (m) TFA, CH2Cl2 (quant);43 (n) H2, Pd(OH)2, MeOH (quant): Proc )
n-propyloxycarbonyl.
The strength of the binding was evaluated with methods based
on Truhlar’s M06-L/6-311G** functional,38 which was reported
to give very good results for π,π interactions.39 The π,π-stacked
dimers of the yellow NDI 9 were therefore optimized with the
M06-L functional, and the BSSE-corrected interaction energies
were computed at the M06-2X/6-311G** level of theory. The
geometry-optimized suprastructure was characterized by a short
plane-to-plane distance d ) 3.36 Å and a twist of the two
aromatic planes by 44°. The strong face-to-face π,π interactions
in π-stacks implied by the short plane-to-plane distance were
confirmed by binding interaction energy of -27.2 kcal/mol. This
value is some 8 kcal/mol stronger than the value determined
for the face-to-face dimers of guanine-cytosine pairs (-19.0
kcal/mol)40 and clearly much stronger than that computed for
dimers of hexafluorobenzene (-1.7 kcal/mol)41 or benzene
(-2.8 kcal/mol).42
too reactive and destroyed other functional groups present in
the substrate. To address this problem, dianhydride 15 was
brominated with dibromoisocyanuric acid.37 Treatment of the
product mixture with iodoethane in EtOH afforded tetraester
16 in overall 35% yield. Reaction with ethanolate followed by
basic hydrolysis of 17 and reaction of the obtained tetraacid
with amines 18 and 19 gave 20.37
Chemoselective removal of the Alloc protecting group in 20
with Pd(0) and PhSiH3 was problematic because the produced
amine 21 cyclized spontaneously into amidine 22. This problem
was bypassed by using the crude amine 21 without workup.
The synthesis of POP 23 from biphenyl 24 and t-butyl
bromoacetate 25 has been extensively optimized over the past
decade.43 Reaction of POP 23 with Y 21 and deprotection of
the obtained POP-Y conjugate 26 gave cationic POP-Y dyad
1, the first target molecule. The anionic POP-Y dyad 2 was
synthesized by reacting the hydrolyzed 17 with the t-Bu-
protected glutamate derivative 27 and amine 18. The obtained
NDI 28 was selectively deprotected using Pd catalysis to give
amine 29, which in turn was reacted in situ with POP 23.
Deprotection of the obtained POP-Y conjugate 30 gave target
molecule 2. The initiator POP-N 3 was synthesized following
the previously established synthetic scheme (Figure 2).33 The
cationic Y monomer 31, needed for control experiments, was
readily prepared by hydrogenation/hydrogenolysis of 20
(Scheme 1).
Synthesis. The blue NDI 11 and red NDI 12 needed to
synthesize POP-B and the red POP-R dyads were obtained by
core substitution of 2,6-dichloro-NDIs 13 with isopropylamine
14 (Scheme 1).8,9,20,23 This approach did not lead to the yellow
POP-Y dyads 1 and 2 because the ethanolate nucleophile was
(38) (a) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157–167. (b)
Zhao, Y.; Truhlar, D. G. J. Chem. Phys. 2006, 125, 194101–194118.
(39) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 119, 525–525.
(40) Jurecka, P.; Sponer, J.; Cerny, J.; Hobza, P. Phys. Chem. Chem. Phys.
2006, 8, 1985–1993.
(41) Frontera, A.; Quinonero, D.; Costa, A.; Ballester, P.; Deya, P. M. New
J. Chem. 2007, 31, 556–560.
(43) (a) Sakai, N.; Majumdar, N.; Matile, S. J. Am. Chem. Soc. 1999, 121,
4294–4295. (b) Baumeister, B.; Sakai, N.; Matile, S. Org. Lett. 2001,
3, 4229–4232.
(42) Sinnokrot, M. O.; Sherrill, C. D. J. Phys. Chem. A 2006, 110, 10656–
10668.
9
11110 J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009

Supramolecular n/p-Heterojunction Surface Architectures
A R T I C L E S
be much more favorable than that in POP-R dyads with ∆EHOMO
) +0.16 eV or the disfunctional POP-B dyads with ∆EHOMO
) -0.40 eV (Figure 1). Favorable frontier orbital energy levels
together with the high electron mobility expected for more
electron-deficient NDI stacks implied that POP-Y architectures
should afford perfect SHJs.
Steady-State Spectroscopy. The absorption spectra of Y 31
and POP-Y 1 in MeOH is dominated by a band around 470 nm
due to the NDI S0-S1 transition (Figure 5). The energy of this
charge transfer type transition depends strongly on the nature
of the core substituents (Figures 1 and 5). In POP-Y 1, this
band (S0 f NDI-LES (locally excited singlet state)) is substan-
tially broader and slightly red-shifted (∼180 cm-1) compared
to that of Y 31. A similar effect, which has been ascribed to an
excitonic interaction between the chromophoric units, has
already been seen with other NDIs on p-oligophenyl (POP) and
oligophenylethynyl (OPE) scaffolds.35,44
The second absorption band centered around 350 nm, with a
vibrational progression and observed with all chromophores, is
due to a π-π* transition involving the NDI center. The
vibrational progression is not so visible with POP-Y 1 as this
band overlaps with another one peaking at 320 nm and
originating from the POP scaffold.
The fluorescence spectra of Y 31 and POP-Y 1 upon local
NDI excitation are very similar, with a maximum at 489 nm.
However, the fluorescence quantum yield, Φf, of POP-Y 1 is
substantially smaller than that of Y 31 (i.e., 0.008 vs 0.078).
Interestingly, the fluorescence quantum yield of Y 31 is close
to that of the corresponding RCl monomer (Φf ) 0.08) but
substantially smaller than that of the B monomer (Φf ) 0.32).
This reduced yield can be ascribed to the occurrence of
intermolecular, hydrogen-bond-assisted nonradiative deactiva-
tion.45 This process is most probably inhibited in the blue NDI
because of the formation of stronger intramolecular hydrogen
bonds.
Time-Resolved Fluorescence. The fluorescence decay of Y
31 monomer was recorded using the time-correlated single-
photon counting technique (TCSPC).46 Upon excitation at 395
nm with a ∼60 ps laser pulse, biexponential decay with 2.1
and 7.4 ns lifetimes was obtained (Table 2). Such decay points
to the presence of two emitting populations or to a distribution
of populations with distinct excited-state lifetimes. This could
be due to H bonds between the alkoxy oxygens and the
solvent competing with H bonding to the carbonyl oxygens.
The early fluorescence dynamics of Y 31 at 490 nm recorded
by fluorescence up-conversion (FU)47 exhibits a small
additional 28 ps decay component that can be reasonably
ascribed to both vibrational and solvent relaxation of the
excited state (Figure 6).
Figure 4. Cyclic voltammograms of (b) Y 20 compared to (a) unsubstituted
NDIs (N),33 (c) R 12,33 (d) B 11,33 (e) POP 3233 and Fc as internal standard.
Cyclic Voltammetry. Redox potentials were determined by
cyclic voltammetry (CV) in dichloromethane with the ferrocene/
ferricenium couple (Fc/Fc+) as external or internal standard.
The first NDI/NDI- and the second NDI-/NDI2- reduction of
Y 20 were both reversible and well detectable (Figure 4b). The
redox potential E0red1 ) -1.09 V for the first NDI/NDI- reduction
was approximated as midpoint between the separate cathodic
and anodic peaks. To calculate the energy level of the LUMO
of the Y 20 against vacuum, the onset of the first reduction
onset
potential E
) -0.98 V was taken (Table 1, entry 3).
red1
Subtraction from the -4.8 eV for the Fc/Fc+ standard gave
ELUMO ) -3.82 eV (Table 1, entry 3, and Figures 4b and 1).
This value was in agreement with reported values for similar
compounds. The proposed alternative use of -5.1 eV for the
Fc/Fc+ standard would reduce the value for yellow NDIs
correspondingly to ELUMO ) -4.12 eV.4
The redox potential at E0ox1 ) +1.48 V for NDI/NDI+
oxidation was approximated from the oxidation wave; a
maximum for the reduction wave as not detectable. The onset
onset
of the first oxidation potential E
) +1.36 V corresponded
ox1
to an EHOMO ) -6.16 eV (Table 1, entry 3). The resulting
electrochemical HOMO/LUMO gap ECgV ) +2.34 eV compared
well with the optical HOMO/LUMO gap Eogpt ) +2.49 eV
onset
obtained from the onset of the lowest energy absorption at λ
max1
) 497 nm. Small differences between electrochemical and
optical HOMO/LUMO gap are expected not only because of
the intrinsic experimental uncertainties but also because charged
molecules are compared neutral excited ones.4
The fluorescence decay of POP-Y 1 is much faster than that
of Y 31, in agreement with the smaller fluorescence quantum
yield (Figure 6). Global analysis of the TCSPC and FU profiles
requires not less than the sum of five exponential functions with
From the detectable oxidation potential of POP 32, the redox
potential for POP/POP+ oxidation was approximated at E0red1
)
)
onset
+1.10 V. The onset of the oxidation potential was at E
ox1
+0.95 V gave EHOMO ) -5.75 eV for POPs (Table 1, entry 5,
Figure 4e).
(44) Banerji, N.; Duvanel, G.; Perez-Velasco, A.; Maity, S.; Sakai, N.;
Matile, S.; Vauthey, E. J. Phys. Chem. A, in press.
(45) (a) Flom, S. R.; Barbara, P. F. J. Phys. Chem. 1985, 89, 4489–4494.
(b) Fu¨rstenberg, A.; Vauthey, E. Photochem. Photobiol. Sci. 2005, 4,
260–267. (c) Sherin, P. S.; Grilj, J.; Tsentalovitch, Y. P.; Vauthey, E.
J. Phys. Chem. B 2009, 113, 4953–4962.
Compared to R 12 and B 11, reduction of Y 20 is easier,
whereas oxidation is much harder (Figures 4b-d and 1). LUMO
and HOMO of Y 20 remain clearly above those of unsubstituted
NDIs N (Figures 4a,b and 1). Y can thus accept electrons from
B and R and donate electrons to N (Figure 1).
(46) Fu¨rstenberg, A.; Julliard, M. D.; Deligeorgiev, T. G.; Gadjev, N. I.;
Vassilev, A. A.; Vauthey, E. J. Am. Chem. Soc. 2006, 128, 7661–
7669.
The HOMO of Y was ∆EHOMO ) +0.41 eV below that of
POP. This value suggested that PCS in POP-Y dyads should
(47) Morandeira, A.; Engeli, L.; Vauthey, E. J. Phys. Chem. A 2002, 106,
4833–4837.
9
J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009 11111

A R T I C L E S
Kishore et al.
Table 1. Frontier Orbital Energy Levels of Y from Cyclic Voltammetry (CV) and Absorption Spectroscopy in Comparison to Other NDIs (R,
N, B) and POPsa
cpdb
E
(V)c
E
(V)d
λ
(nm)e
onset
max1
HOMO (eV)f
LUMO (eV)g
ECgV (eV)h
Eogpt (eV)i
ref
onset
ox1
onset
red1
entry
1
2
3
4
5
B
R
Y
N
0.55
1.11
1.36
nd
–1.24
–1.02
–0.98
–0.81
nd
665
–5.35
–5.91
–6.16
–7.07j
–5.75
–3.56
–3.78
–3.82
–4.01
–2.35k
1.79
2.13
2.34
nd
1.86
2.18
2.49
3.06
3.40
33
33
570
497
405
365
33
33
POP
0.95
nd
a For original data and conditions, see Figures 4 and 5. b Measured for monomeric yellow NDI 20 and the corresponding NDI analogues, and POP
c
32; for generic and specific structures, see Figure 1, Scheme 1, and references. Onset value of the first oxidation potential in volts against Fc/Fc+.
d Onset value of the first reduction potential in volts against Fc/Fc+. e Onset wavelength of the longest wavelength absorption. f Energy of the highest
g Energy of the lowest unoccupied molecular orbital against vacuum,
onset
occupied molecular orbital against vacuum, from -4.8 eV for Fc/Fc+ minus E
.
ox1
i
onset
from -4.8 eV for Fc/Fc+ minus E
.
h Electrochemical HOMO/LUMO gap ECg V (eV) ) LUMO - HOMO. Optical HOMO/LUMO gap Egopt (eV) )
red1
j
onset
1240/λ
(nm). From LUMO - Eogpt k From HOMO + Eogpt
. .
max1
the B and RCl monomers reveal that both anion and cation have
similar absorption spectra.
However, energetic considerations point to a CSS with the
hole on the POP scaffold. Indeed, according to the Weller
equation,49 and using the redox potentials of Y monomer and
POP together with a value of 2.6 eV for the energy of the LES,
the driving force of the photoinduced CS between two yellow
NDIs amounts to -0.03 eV only, whereas that for the CS
between the POP scaffold and a yellow NDI is -0.41 eV. Thus,
even if CS to between two NDIs is operative, that with the POP
can be expected to be much faster. Furthermore, even if the
NDI/NDI CSS is populated, hole transfer from the NDI•+ to
the POP should be expected.
Figure 5. Absorption and emission spectra of Y 31 (yellow and green) in
MeOH compared to the corresponding RCl (red),35 B (blue and violet),35
and air mass (AM) 1.5 solar spectral irradiance.
The presence of slow decay components observed by time-
resolved fluorescence as well as a contribution of the LES
population to the late TA spectra of POP-Y indicates that the
time scales of CS and CR are partially overlapping, and
therefore, the CR dynamics cannot be directly extracted
from the decay of the NDI anion TA band. This situation is
similar to that encountered with red POP-RCl, and therefore,
the same procedure, described in detail in ref 35, has been used
to extract the intrinsic CR dynamics of POP-Y. Figure 8 shows
the TA profile of POP-Y 1 at 500 nm after removal of the
contribution arising from the LES population, the latter being
determined from the fluorescence time profile.35 This trace thus
reflects the temporal dependence of the CSS population. The
continuous line is the best fit of eq 1 in ref 35, which is a
convolution of the intrinsic CR dynamics with the buildup rate
of the CSS population. The CSS profile could be well-
reproduced assuming a biexponential CR with 200 ps and >5
ns time constants and with 0.73 and 0.27 relative amplitudes.
The CR dynamics of POP-Y 1 is quite similar, although
slightly slower, to that found with red POP-RCl and much slower
than that of blue POP-B (Figure 8). This difference can be
unambiguously attributed to the nature of the CSS. For energetic
reasons, only the NDI/NDI CSS can be populated in the POP-
B, whereas both NDI/NDI and POP/NDI CSS are accessible
with POP-RCl and POP-Y. The longer-lived POP/NDI CSS can
be explained by a superior spatial confinement of the charges
as already discussed.35 On the other hand, the similar CR
dynamics of POP-Y and POP-RCl are in agreement with their
almost identical driving force.
time constants ranging from about 1 ps to 5.5 ns (Table 2).
The fluorescence dynamics of POP-RCl and POP-B are also
multiexponential, but the associated time constants exhibit a
narrower dispersion.35 However, the relative amplitudes of the
smaller decay rate constants measured with POP-Y are very
small, indicative of very efficient nonradiative deactivation
pathways of the LES.
Transient Absorption Spectroscopy. Information on the origin
of the ultrafast nonradiative deactivation of POP-Y 1 was
obtained from the transient absorption (TA) spectra recorded
upon 400 nm excitation (Figure 7A).48 The TA spectra consist
of a positive band located between 470 and 570 nm. The shape
of this band exhibits substantial changes during the first 10 ps
after excitation. The intensity on its long wavelength side
decreases, whereas that on the high wavelength side increases.
From 10 ps onward, the band shape stays essentially unchanged
but decreases on the 100 ps time scale to a value that remains
almost constant within the time window of the experiment.
These initial spectral changes can be ascribed to an intramo-
lecular charge separation (CS) in the POP-Y dyad. Indeed, the
red side of the TA band recorded at early time can be ascribed
to the absorption of the LES, whereas its blue side and the band
measured after 10 ps can be assigned to a charge-separated state
(CSS). This interpretation is comforted by the S1 state absorption
spectrum measured with monomer Y 31 in MeOH and by the
spectrum of Y radical anion, recorded with Y 31 in the presence
of the electron donor, N,N-dimethylaniline (DMA, Figure 7B).
Whereas in the CSS the electron is on a NDI unit, the location
of the hole, the POP scaffold, or another NDI chromophore
cannot be deduced from the TA spectra. As discussed in ref
35, POP•+ is expected to absorb weakly around 450 nm.
Moreover, spectroelectrochemical measurements performed with
Zipper and LBL Assembly. Freshly cleaned gold electrodes
were dipped into 1 mM solutions of POP-N 3 in aqueous TFE
in the presence of 500 mM NaCl and 0.5 mM phosphate buffer
of pH 7.0, at ambient temperature for 1 week.8,9,33,34 Initiation
kinetics were followed by photocurrent generation (Figures 9
(48) Duvanel, G.; Banerji, N.; Vauthey, E. J. Phys. Chem. A 2007, 111,
5361.
(49) Rehm, D.; Weller, A. Isr. J. Chem. 1970, 8, 259–271.
9
11112 J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009

Supramolecular n/p-Heterojunction Surface Architectures
A R T I C L E S
Table 2. Fluorescence Decay and Quantum Yield of Y and POP-Ya
c
entry
cpd
τf1
Αf1
τf2
Αf2
τf3
Αf3
τf4
Αf4
τf5
Αf5
<τf >
Φf
1
2
31
1
28 ps
0.54 ps
0.12b
0.27
2.1 ns
4.1 ps
0.74
0.39
7.4 ns
37 ps
0.14
0.21
2.59 ns
319 ps
0.078
0.008
2.2 ns
0.12
5.5 ns
0.01
a Time constants τ with relative amplitudes A were obtained from the global analysis of the fluorescence time profiles of Y 31 and POP-Y 1. b This
component is not due to the decay of the excited-state population (see text). c Amplitude-averaged lifetime.
aqueous solutions of the cationic POP-Y propagator 1, then into
aqueous solutions of the anionic POP-Y propagator 2, and so
on (Figure 2).
Comparisonwithconventionallayer-by-layer(LBL)assembly13,16
was essential to verify existence and significance of zipper
assembly (Figure 10). For LBL assembly of POP-Y architec-
tures, gold electrodes are covered with lipoic acid 35 instead
of POP-N initiator 3. The obtained Au-35 electrodes have an
anionic surface to bind the cationic POP-Y 1, which, according
to extensive studies on LBL assembly,13,16 should produce an
cationic surface to bind the anionic POP-Y dyad 2. The obtained
Au-35-1-2 should then bind cationic 1, Au-35-1-2-1
Figure 6. Intensity-normalized time profiles of the fluorescence intensity
anionic 2, and so on. Beyond the first layer with and without
of Y 31 (green, X) and POP-Y 1 (green, b) compared to POP-RCl 6 (red,
unsubstituted NDI acceptors, the expected differences between
zippers Au-3-(1-2-)n and LBL assemblies Au-35-(1-2-)n
O)9,35 and POP-B 4 (blue, 0)23,35 and best multiexponential fits.
are subtle and of supramolecular nature only. Driven by ion
pairing and conceivably containing π-stacks with interdigitating
NDIs, POP interdigitiation is not essential for LBL assembly.
The main difference with zipper assembly is the presence of
intra- and interlayer recognition motifs introduced by initiator
3 and propagated by mismatched rigid-rod scaffolds to produce
POP-NDI architectures of higher order.
Photocurrent Generation. Photocurrent generation by POP-Y
zipper architectures was evaluated with the gold electrode as
electron acceptor, a Pt electrode as cathode, and triethanolamine
(TEOA) as electron donor (Figure 9).8,9,33,34 The use of
dithiothreitol as more powerful (DTT, Figure 9h) or ascorbate
as alternative electron donors gave similar results. The slightly
stronger I- and hydroquinone gave clearly weaker and very
strong but transient photocurrents, respectively. The use of Co3+
(Figure 9i) or the slightly stronger methylviologen acceptor
together with TEOA gave less satisfactory results.
Photocurrent generation and termination by POP-Y zippers
Figure 7. (a) Transient absorption spectra recorded at different time delays
Au-3-(1-2-)n was instantaneous, and the obtained currents
after excitation of POP-Y 1 in MeOH and (b) TA spectra recorded after
were as stable as with POP-B and POP-RCl zippers.8,33 The
after excitation of monomer Y 31 with (blue) and without (red) 3 M DMA
photocurrents generated by zipper Au-3-(1-2-)n increased
overall almost linearly with increasing number of layers (Figure
10a, b).
in MeOH.
In early stages, cationic POP-Y 1 produced more significant
photocurrent increases, whereas anionic POP-Y 2 could even
cause a small decrease. AFM images suggest that these
differences could occur by substantial overzipping with cationic
POP-Y 1 despite charge repulsion, which then are subsequently
displaced by the stronger binding anionic POP-Y 2 (see below).
Alternative explanations for these differences include facilitated
and hindered electron transfer from TEOA to cationic and
anionic zipper surfaces, respectively.
LBL assembly Au-35-(1-2-)n generated clearly less
photocurrent (Figure 10a, X). Both zipper and LBL assembly
depended strongly on experimental conditions and, importantly,
varied for different POP-NDI architectures. Overall, decreasing
differences between zipper and LBL assembly with decreasing
ionic strength and solvent polarity indicated that contributions
Figure 8. Intensity-normalized time profiles of the CSS population (see
text) in the transient absorption of POP-Y 1 (green) with best fit to eq 1 in
ref 35 compared to POP-B 4 (blue) and POP-RCl 6 (red).35
and 10) and cyclic voltammetry and continued until saturation
values were reached. The absence of redox waves for K3Fe(CN)6
demonstrated that the final Au-3 electrodes do not contain large
uncovered areas. The obtained Au-3 was dipped first into
(50) Kalayanasundaram, K.; Kiwi, J.; Gra¨tzel, M. HelV. Chim. Acta 1978,
61, 2720–2730.
9
J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009 11113

A R T I C L E S
Kishore et al.
Figure 9. POP-Y photosystems for photocurrent generation. After absorp-
tion of light (hν) by yellow NDIs Y, the design envisions (a) PCS by
POPfY electron transfer (ET), followed by (b) YfY, (c) POPfPOP, and
(d) YfN ET along formal OMRG-SHJs, (e) NfAu, (f) TEOAf POP,
and (g) Ptf(decomposed)50 TEOA ET. Excitation of POP could be followed
by energy (or electron) transfer,44 that of N by energy or hole transfer to
Y. (h) Use of DTT instead of TEOA gave similar photocurrent. (i) Use of
Co3+ together with TEOA gave less photocurrent.
Figure 11. Current-layer profiles of POP-B architectures obtained by
zipper assembly Au-3-(4-5-)n (b) and LBL assembly Au-35-(4-5-)n
(X); lines are added to guide the eye.
Au-3-(4-5-)n generated increasing current with increasing
thickness until a maximum was reached at around 10 layers
(Figure 11). Decreasing photocurrents with zipper assembly
beyond the critical 10 layers yielded a bell-shaped JL profile
for Au-3-(4-5-)n zippers.
The maximal photocurrent Jmax ) 0.8 µA cm-2 of blue
Au-3-(4-5-)n zippers was clearly inferior compared to the
Jmax ) 20 µA cm-2 found for the yellow Au-3-(1-2-)n.
Although the significance of this comparison should not be
overestimated, this clear difference was consistent with the
formation of SHJs with the yellow but not with the blue zippers.
Comparable without reservation was the critical thickness of
POP zippers. With yellow zippers Au-3-(1-2-)n, photocur-
rents stopped increasing after around 20 layers of deposition.
The critical thickness of blue Au-3-(4-5-)n zippers was
already reached at about 10 layers. The critical thickness of both
architectures was clearly better than in related systems.10a These
findings suggested that zipper assembly provides access to
higher-order architectures with significant properties. The clearly
better critical thickness of yellow compared to blue zippers
indicated that the yellow NDIs produce more ordered zipper
architectures. This interpretation was in agreement with the
stronger π-stacking expected from more π-acidic yellow NDIs.
Photocurrent saturation occurs either because the zipper stops
growing or because electrons and holes start to recombine in
thicker architectures before reaching the electrodes. QCM results
suggest that the latter applies for the yellow zippers
Au-3-(1-2-)n (see below).
Current-Voltage Curves. Current-voltage (JV) curves were
measured to determine short circuit current density JSC, open
circuit potential VOC, and the fill factor FF (Figure 12). Fill
factors FF relate to the power generated with light, benefit from
highly organized surface architectures,5a and are of critical
importance for device efficiency. Minimal fill factors of linear
JV curves are FF ) 0.25, fill factors of optimized organic BHJ
solar cell with thick polymer films prepared by top-down
methods such as spin coating have a FF ) 0.61.2b
The JV curve of zipper Au-3-(1-2-)9-1 revealed FF )
0.53 (Figure 12a). This value was higher than the FF ) 0.48
obtained by LBL assembly Au-35-(1-2-)9-1 (Figure 12b),
which confirmed access to highly ordered architectures.5a FF
) 0.53 found for POP-Y zippers compared also very well with
Figure 10. Current-layer profiles of zipper assembly Au-3-(1-2-)n
(b) and assembly LBL Au-35-(1-2-)n (X) with incubation for (a) 48 h
and (b) 24 h per layer.
from hydrophobic interactions are less specific. Moreover,
charge repulsion decreases with increasing ionic strength. The
difference between zipper and LBL assembly was detectable
only after incubation for 48 h (Figure 10a). The photocurrents
generated by zippers Au-3-(1-2-)n and LBL assembly
Au-35-(1-2-)n after 24 h of incubation were roughly the
same (Figure 10b). This finding confirmed that the assembly
of zipper architectures is slower than that of LBL architectures.
Together with the previous capping experiments,8 the differences
found for zipper and LBL assembly confirmed that (a) zipper
assembly exists, (b) differs significantly from LBL assembly,
and (c) generates more photocurrent than the latter (Figure 10).
The presence of unsubstituted NDI acceptors in POP-N
initiators 3 is essentially ignored in the discussion of POP-Y
photosystems, although it introduces a formal redox gradient
in the e--transporting channel of formal POP-Y/N SHJs. This
neglect is appropriate because little difference was found
between the corresponding POP-RCl/N and POP-RCl zippers.33
To secure comparative values and validate the approach, the
previously reported blue POP-B zippers8 were exposed to direct
comparison with LBL assembly, as well. Under identically
optimized conditions, LBL assembly Au-35-(4-5-)n failed
to generate significant photocurrent, whereas zipper assembly
9
11114 J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009

Supramolecular n/p-Heterojunction Surface Architectures
A R T I C L E S
Figure 12. Current-voltage profile of (a) zipper Au-3-(1-2-)9-1 and
(b) LBL Au-35-(1-2-)9-1 with indication of short circuit current density
(JSC), open circuit voltage (VOC), maximum power rectangle, maximum
power point, and fill factor FF ) maximum power/(VOC × JSC) ) (Vm
×
Jm)/(VOC × JSC).
Figure 14. Tapping mode AFM overview (a) and magnified (b,c) amplitude
(a,b) and height images (c) of multilayer zippers Au-3-(1-2-)3-1.
Figure 13. Action spectrum of POP-Y zippers (b, green; Au-3-(1-2-)8)
compared to the corresponding POP-RCl (O, red) and POP-B zippers (X,
blue).9 IPCE values are normalized at 350 nm.
FF ) 0.42 for POP-B zippers (4 layers)9 and FF ) 0.49-0.53
for POP-RCl zippers (3-5 layers).33 Comparison of open circuit
photocurrents was more appropriate with the relative values
accessible in action spectra (see below). Absolute currents and
efficiencies are best appreciated considering an up to 280-fold
quenching via energy transfer to gold surfaces.12a
Action Spectra. The action spectrum of zipper Au-3-(1-
2-)8 revealed a strong peak around 480 nm (Figure 13, b).
Coinciding with the absorption maximum of the yellow NDI,
this peak demonstrated that the NDI component in POP-Y zipper
architectures generates photocurrent. The ratio IPCE480/IPCE350
) 0.85 indicated that the maximal ICPE (incident photon-to-
current efficiency) accessible around 480 nm was almost as good
as that at high energy. For POP-RCl zippers, the corresponding
IPCE550/IPCE350 ) 0.51 was clearly lower (Figure 13, O). In
POP-B zippers, the blue NDI was with IPCE620/IPCE350 ) 0.02
essentially inactive (Figure 13, X).9 The comparably high IPCE
of the yellow NDI was in excellent agreement with favorable
electron transfer from POP donors to NDI acceptors (Figure 1)
in response to light to result in long-lived PCS (Figure 8) and
thus the formation of functional SHJs.
Figure 15. Possible mechanisms of mixed zipper architectures integrating
overzipped domains.
proposed to be the key to significant function in molecular
optoelectronics, including high photovoltaic efficiency.2
Atomic force microscopy (AFM) images further revealed
occasional clumps. They already occurred with monolayers
similar to Au-3 and are of unknown origin, particularly
considering that the layers were grown on common, piranha-
cleaned gold surfaces without special purity precautions.
Presumably more important, AFM images also revealed sharp
steps of 1.5 nm height between lower and higher domains. With
Au-3, such steps may occur at the edges of “open”, high and
soft domains where initiators stand vertically on the surface (i.e.,
Au-3⊥-) and “closed”, low and hard domains where initiators
lay flat on the surface (i.e., Au-3), Figure 15). Addition of
cationic propagators 1 to mixed Au-3⊥-/Au-3) monolayers
could then cause the laying initiators to stand up to give a
homogeneous, soft and open bilayer Au-3-1- (Figure 15).
With Au-3-1 and higher Au-3-(1-2-)n-1 zippers, the
occasionally occurring 1-2 nm higher domains could then
Morphology. The POP-Y zipper architectures had exception-
ally smooth surfaces (Figure 14). The surface of multilayer
zipper Au-3-(1-2-)3-1 was at least as smooth as uncovered
gold surfaces and those of monolayers similar to Au-3.
Constant or decreasing surface roughness with increasing zipper
growth was contrary to increasing surface roughness observed
with conventional LBL assembly.13 The final surfaces of
advanced zippers were clearly smoother than those of LBL and
top-down spin-coated BHJ photosystems.2,3,13 For example,
root-mean-square (rms) roughness as large as 2.6, 3.2, and 4.7
nm has been reported for the LBL films with 1, 5, and 11
bilayers, respectively.13c Low surface roughness has been
9
J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009 11115

A R T I C L E S
Kishore et al.
organization of POP-Y zipper architectures is extraordinary and
matters for function.
Conclusions
The characteristics of the here introduced yellow POP-Y
photosystems are outstanding. Compared to the blue and red
architectures as well as related LBL systems, yellow zippers
excel with optoelectronic matching, long-lived photoinduced
charge separation, strong π-acidity, strong photocurrents, as well
as significant critical thickness and smooth surfaces.
The consistency of results with the complete primary color
series of POP-Y, POP-RCl, and POP-B hybrids is remarkable.
The decreasing HOMO and LUMO energies from blue over
red to yellow NDIs coincide with increasingly facile electron
transfer from POP donors to excited NDI acceptors. This
increasingly favorable rod-stack PCS correlates perfectly with
increasing photocurrent generation by blue < red < yellow NDI
chromophores in the action spectra of their POP-NDI architec-
tures. Increasing π-acidity with blue < red < yellow NDIs is
expected to strengthen π-stacking in zipper assemblies, increase
charge mobility, and organize zipper architectures. These trends
conceivably account for excellent critical thickness, fill factors,
and surface smoothness of POP-Y architectures.
These results demonstrate that the here reported insights from
direct comparison with LBL assembly, QCM, AFM, and
molecular modeling are essential to understand zipper assembly.
They identify the new POP-Y zipper architectures as ideal for
the creation of optoelectronic devices including OMARG-SHJs.
The availability of the primary color collection of POP-NDI
zippers is particularly important because it suffices, in principle,
to already build mixed OMARG-SHJ architectures that absorb
most of the visible light and have an oriented “bluefredfyellow”
redox gradient in the e--transporting channel (Figure 1). With
access to multicolor redox gradients in the e--transporting
channel secured, we currently try to better integrate red and
blue NDIs as well as to install redox gradients in the h+-
transporting channel. Linear conjugated oligomers with high
HOMO levels and topologically matching repeat distances such
as oligophenylethynyls (OPEs) or oligothiophenes are currently
envisioned to solve both problems. In parallel, we try to develop
access to surfaces other than gold to avoid quenching from
plasmon surface resonance and thus increase zipper efficiency.
Figure 16. QCM frequencies (with linear curve fit) as a function of the
number of theoretical layers of zipper assembly Au-3-(1-2-)n (b, 9)
and LBL assembly Au-35-(1-2-)n (X).
originate from additional cationic propagators 1 that zip up
weakly despite charge repulsion. Addition of anionic propagators
2 would then displace these weakly bound overzipped cationic
propagators 1 to afford Au-3-1-2 and higher Au-3-(1-
2-)n zippers with some overzipped Au-3-1-2-2 to account
for the 1-2 nm higher planes. Stronger photocurrents obtained
after addition of cationic propagators 1 could then suggest that
overzipping is more likely with 1 than with anionic propagators
2. This interpretation appears meaningful considering acidity
and basicity of the involved anions and cations.
Quartz Crystal Microbalance. The mass deposited by zipper
assembly of POP-Y architectures was determined electrome-
chanically with a quartz crystal microbalance (QCM) (Figure
16). The mass deposited by zipper assembly clearly exceeded
that deposited by LBL assembly. This result suggested that the
differences found for photocurrent generation by zipper and LBL
assembly originate to a good part from the different amounts
deposited per layer (Figure 10a vs 16). Nevertheless, QCM data
demonstrated continued deposition in the LBL assembly, in
contrast to the saturation behavior found in the photocurrent
generation after about 4 layers. Better charge mobility with
higher long-range order in zipper assembly is likely to contribute
to the found strong relative photocurrent, as well.
In contrast to photocurrent generation, zipper assembly did
not show saturation in the QCM signal (Figure 10a vs 16). This
lack of saturation indicated continuous growth of the surface
assembly. Photocurrent saturation in Au-3-(1-2-)n-1 zippers
thus occurred because charges start to recombine before reaching
the electrodes and not because POP-Y zippers stop growing.
This finding revealed that the critical thickness of
Au-3-(1-2-)n-1 is reached around ∼20 layers (i.e.,
Au-3-(1-2-)9-1, ∼30 nm). Comparable systems in the
literature have inferior critical thickness.10a,13c The important
critical thickness found thus confirmed that the long-range
Acknowledgment. We thank D.-H. Tran and L. Maffiolo for
contributions to synthesis, D. Jeannerat, A. Pinto, and S. Grass for
NMR measurements, P. Perrottet, N. Oudry, and G. Hopfgartner
for MS, the Swiss National Supercomputing Center (CSCS) in
Manno for CPU time, and the University of Geneva, the Swiss
NSF, and the Nanotechnology network Nanoned for financial
support.
Supporting Information Available: Experimental details. This
material is available free of charge via the Internet at http://
pubs.acs.org.
JA9030648
9
11116 J. AM. CHEM. SOC. VOL. 131, NO. 31, 2009