Electrochemistry, spectroscopy and electrogenerated chemiluminescence of perylene, terrylene, and quaterrylene diimides in aprotic solution

Article abstract of DOI:10.1021/ja984188m

The electrochemistry, UV-vis spectrophotometry, photoluminescence, and electrogenerated chemiluminescence (ECL) of perylenedicarboxylic imide, perylenetetracarboxylic diimide (PDI), terrylenetetracarboxylic diimide (TDI), and quaterrylenecarboxylic diimide (QDI) were investigated. All compounds undergo two reversible one-electron reductions and one reversible one-electron oxidation reaction. The first reduction potential shifts to less negative values and the potential for oxidation to less positive values for the diimide series with increasing size of the aromatic core. These changes in potential correlate well with orbital energies from molecular orbital calculations. The difference in potential between the first and Second reduction waves decreased with increasing distance between the imide groups, so that TDI and QDI show only a single reduction wave, equivalent to a two- electron reduction. These reduction potentials provide estimates for the equilibrium constant for disproportionation of the radical anion. Very stable ECL spectra of PDI or TDI generated by sequential production of the radical cation and radical anion at an electrode were observed; these were identical to the photoluminescence spectra A consideration of the energetics of the electron transfer reaction and the singlet energy leads to the conclusion that emission occurs mainly via generation of triplets followed by triplet- triplet annihilation (the T-route).

Full text of DOI:10.1021/ja984188m

J. Am. Chem. Soc. 1999, 121, 3513-3520
3513
Electrochemistry, Spectroscopy and Electrogenerated
Chemiluminescence of Perylene, Terrylene, and Quaterrylene
Diimides in Aprotic Solution
Sang Kwon Lee,† Yanbing Zu,† Andreas Herrmann,‡ Yves Geerts,‡,§ Kla1us Mu1llen,‡ and
Allen J. Bard*,†
Contribution from the Department of Chemistry and Biochemistry, The UniVersity of Texas at Austin,
Austin, Texas 78712, and Max-Planck-Institut fu¨r Polymerforschung, Ackermannweg 10,
55128 Mainz, Germany
ReceiVed December 4, 1998
Abstract: The electrochemistry, UV-vis spectrophotometry, photoluminescence, and electrogenerated chemi-
luminescence (ECL) of perylenedicarboxylic imide, perylenetetracarboxylic diimide (PDI), terrylenetetracar-
boxylic diimide (TDI), and quaterrylenecarboxylic diimide (QDI) were investigated. All compounds undergo
two reversible one-electron reductions and one reversible one-electron oxidation reaction. The first reduction
potential shifts to less negative values and the potential for oxidation to less positive values for the diimide
series with increasing size of the aromatic core. These changes in potential correlate well with orbital energies
from molecular orbital calculations. The difference in potential between the first and second reduction waves
decreased with increasing distance between the imide groups, so that TDI and QDI show only a single reduction
wave, equivalent to a two-electron reduction. These reduction potentials provide estimates for the equilibrium
constant for disproportionation of the radical anion. Very stable ECL spectra of PDI or TDI generated by
sequential production of the radical cation and radical anion at an electrode were observed; these were identical
to the photoluminescence spectra. A consideration of the energetics of the electron transfer reaction and the
singlet energy leads to the conclusion that emission occurs mainly via generation of triplets followed by triplet-
triplet annihilation (the T-route).
Introduction
mophores for applications, such as reprographical processes,3
fluorescent solar collectors,4 photovoltaic devices,5 dye lasers,6
and molecular switches.7,8 Recently, several new terrylenetet-
racarboxylic diimides and quaterrylenetetracarboxylic diimides
were synthesized.2f,9 Owing to their extended π systems, these
molecules exhibit absorption and emission bands at much longer
wavelengths (λab ) 650-760 nm, λfl ) 665-780 nm) than the
corresponding perylene diimides (λab ) 526 nm, λfl ) 545 nm);
the fluorescence quantum yields of terrylenetetracarboxylic
diimides are near 1.0. These characteristics make them good
potential ECL labels for analytical purposes.
ECL is initiated by the generation of ions capable of
undergoing energetic electron-transfer reactions at electrodes.10
The general scheme involves the alternate electrochemical
production of radical cations and radical anions (eqs 1 and 2)
by stepping the electrode potential to appropriate values. The
We describe here the electrochemical reduction and oxidation
and radical ion annihilation electrogenerated chemiluminescence
(ECL) of perylene, terrylene, and quaterrylene diimides (Figure
1) in aprotic solution. We felt that these molecules deserved
study on the basis of the stability of their radical ions and the
good fluorescence efficiencies of compounds of this series,
which suggest good ECL properties. Moreover, as described
below, the electrochemistry allows examination of the energetics
during the stepwise formation of the radical anion and dianion
as a function of molecule size. Of particular interest is the long
wavelengths for emission from the diimides compared to the
hydrocarbon core, which is of importance in the design of ECL
labels that emit in the red and near-IR region.
Perylene dyes have been known since 1913 as highly
photostable pigments or vat dyes. Perylenetetracarboxylic di-
imides have been studied because of their brilliant color, strong
absorption and fluorescence, and good thermal, chemical, and
photochemical stability.1,2 They have been suggested as chro-
(3) Loufty, H. O.; Hor, A. M.; Kazmaler, P.; Tarn, M. J. Imaging Sci.
1989, 33, 151.
(4) Seybold, G.; Wagenblast, G. Dyes Pigm. 1989, 11, 303.
(5) Schlettwein, D.; Wo¨hrle, D.; Karmann E.; Melville, U. Chem. Mater.
1994, 6, 3.
(6) Sadari, M.; Hadel, L.; Sauers, R. R.; Husain, S.; Krogh-Jespersen,
K.; Westbrook, J. D.; Bird, G. R. J. Phys. Chem. 1992, 96, 7988.
(7) O’Neil, M. P.; Niemczyk, M. P.; Svec, W. A.; Gosztola, D.; Gaines,
G. L.; Wasielewski, M. R. Science 1992, 257, 63.
(8) Tasch, S.; List, E. J. W.; Ekstrom, O.; Graupner, W.; Leising, G.;
Schlichting, P.; Rohr, U.; Geerts, Y.; Scherf, U.; Mu¨llen, K. Appl. Phys.
Lett. 1997, 71, 2883.
† The University of Texas at Austin.
‡ Max-Planck-Institut fu¨r Polymerforschung.
§ Present address: Universite´ Libre de Bruxelles, Macromolecular
Chemistry, CP 206/1, Boulevard du Triomphe, 1050 Bruxelles, Belgium.
(1) (a) Zolinger, H. Color Chemistry; VCH: New York, 1991. (b) Fabian,
J.; Zahradnik, R. Angew. Chem., Int. Ed. Engl. 1989, 28, 677.
(2) (a) Quante, H.; Geerts, Y.; Mu¨llen, K. Chem. Mater. 1997, 9, 495.
(b) Quante, H.; Mu¨llen, K. Angew. Chem., Int. Ed. Engl. 1995, 34, 1323.
(c) Feiler, L.; Langhals, H.; Polborn, K. Liebigs Ann. 1995, 7, 1229. (d)
Demmig, S.; Langhals, H. Chem. Ber. 1988, 121, 225. (e) Langhals, H.
Chem. Ber. 1985, 118, 4641. (f) Geerts, Y.; Quante, H.; Platz, H.; Mahrt,
R.; Hopmeier, M.; Bo¨hm, A.; Mu¨llen, K. J. Mater. Chem., in press.
(9) Holtrup, F. O.; Mu¨ller, G. R. J.; Quante, H.; Feyter, S. D.; Schryver,
F. C. D.; Mu¨llen, K. Chem. Eur. J. 1997, 3, 219.
(10) Faulkner, L. R.; Bard, A. J. In Electroanalytical Chemistry; Bard,
A. J., Ed.; Marcel Dekker: New York, 1977; Vol. 10, pp 1-95.
10.1021/ja984188m CCC: $18.00 © 1999 American Chemical Society
Published on Web 03/26/1999

3514 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
Lee et al.
report briefly describing the ECL of two different N-substituted
perylene diimide compounds.14
The species described here are perylenedicarboxylic imide
(PI), three perylenetetracarboxylic dimides (PDI-1, -2, and -3),
terrylenetetracarboxylic diimide (TDI), and quaterrylenecar-
boxylic diimide (QDI). The main focus of this study was to
correlate the structures of a homologous series of imides of
different sizes and orbital energies and their electrochemical
properties and ECL. We discuss the effect of the aromatic imide
structure and the N-substituent on redox potentials and the
disproportionation of the radical anions.
Experimental Section
Chemicals. Acetonitrile (MeCN, UV grade, Burdick and Jackson,
Muskegon, MI) and chloroform (anhydrous, 99+%, Aldrich) were used
as received after being transported unopened into an inert atmosphere
drybox (Vacuum Atmospheres Corp., Los Angeles, CA). Tetra-n-
butylammoniom hexafluorophosphate (TBAPF6, SACHEM, Inc., Aus-
tin, TX) was recrystallized from EtOH/H2O (4:1, v/v) three times and
dried at 100 °C in a vacuum oven and stored in the drybox before use.
All solutions used in UV-vis, fluorescence, electrochemistry, and ECL
measurements were prepared in the drybox under a helium atmosphere
and sealed in airtight cells.
Figure 1. Chemical structures used in this investigation: PI, (2,6-
diisopropylphenyl)-3,4-perylenedicarboxylic imide; PDI-1, N,N′-bis-
(2,6-dimethylphenyl)-3,4,9,10-perylenetetracarboxylic diimide; PDI-2,
N,N′-bis(2,6-diisopropylphenyl)-3,4,9,10-perylenetetracarboxylic diim-
ide; PDI-3, N,N′-bis(2,5-di-tert-butylphenyl)-3,4,9,10-perylenetetracar-
boxylic diimide; TDI, N-(2,6-diisopropylphenyl)-N′-octylterrylenetet-
racarboxylic diimide; QDI, N,N′-bis(2,6-diisopropylphenyl)-3,4,9,10-
quaterrylenetetracarboxylic diimide.
(2,6-Diisopropylphenyl)-3,4-perylenedicarboxylic imide (PI) (Ald-
rich), N,N′-bis(2,6-dimethylphenyl)-3,4,9,10-perylenetetracarboxylic di-
imide (PDI-1) (∼85%, Aldrich), and N,N′-bis(2,5-di-tert-butylphenyl)-
3,4,9,10-perylenetetracarboxylic diimide (PDI-3) (∼97%, Aldrich) were
used as received. N,N′-Bis(2,6-diisopropylphenyl)-3,4,9,10-perylene-
tetracarboxylic diimide (PDI-2), N-(2,6-diisopropylphenyl)-N′-octyl-
terrylenetetracarboxylic diimide (TDI), and N,N′-bis(2,6-diisopropy-
lphenyl)-3,4,9,10-quaterrylenecarboxylic diimide (QDI) were available
from previous studies.2f,8,9 Deionized water from a Millipore Milli-Q
system was used throughout.
Apparatus and Procedures. A charge-coupled device (CCD)
camera (Photometrics CH260) cooled to below -135 °C and interfaced
to a personal computer was used to obtain ECL spectra. The camera
was focused on the output of a grating spectrometer (concave grating,
1 mm entrance slit, Holographics, Inc.). The CCD camera and general
configuration of the spectra acquisition have been described previ-
ously.15
Cyclic voltammetry and bulk electrolysis for spectroelectrochemistry
experiments were carried out with either the Model 660 electrochemical
workstation (CH Instruments) or a PAR Model 173/175. The working
electrode consisted of an inlaid platinum disk (1.5 or 2.1 mm diameter)
that was polished on a felt pad with 0.05 µm alumina (Buehler, Ltd.)
and sonicated in absolute EtOH for 1 min before each experiment. A
platinum wire was used as a counter electrode. A silver wire served as
quasi-reference electrode, and its potential was calibrated vs aqueous
SCE (for comparison with earlier studies) by addition of ferrocene as
an internal standard (taking E(Fc/Fc+) ) 0.424 V vs SCE in the mixed
solvent of CHCl3/MeCN).
subsequent annihilation reaction between these two ions gener-
ates excited states that lead to emission (eqs 3 and 4).
R - e f R•+
R + e f R•-
•+ + R•- f 1R* + R
1R* f R + hν
(1)
(2)
(3)
(4)
R
In the case of an energy-deficient system, i.e., a system where
the enthalpy of the annihilation reaction is smaller than the
energy needed to produce the excited singlet, the luminescent
singlet state can be produced via the triplet-triplet annihilation
reactions (the T-route).10
R
•+ + R•- f 3R* + R
3R* + 3R* f 1R* + R
(5)
(6)
Fluorescence spectra were obtained with a SLM Aminco SPF-500
spectrofluorometer, and UV-vis spectra were recorded on a Milton
Roy Spectronic 3000 array spectrophotometer.
Calculations. Digital simulation for electrochemical reactions of TDI
and QDI utilized DigiSim (BAS, West Lafayette, IN) run on a Gateway
2000 model P5-90 personal computer. The Hyperchem 5.0 software
package was used for molecular mechanics and semiempirical molecular
orbital calculations. The structures of the compound molecules were
optimized using the MM+ force field. Nonbonded electrostatic
interactions were calculated using bond dipole interactions. PM3 was
chosen as the semiempirical method to calculate the molecular orbitals
and electron density at each atom since it has been reported to perform
well on peri-fused polycyclic hydrocarbons16 and is probably more
Many polyaromatic hydrocarbons, the first compounds ex-
amined for ECL applications, produce emission by this reaction
scheme.10-13 In particular, the ECL of perylene12 and, more
recently, dibenzotetraphenylperiflanthene containing a perylene
spacer13 has been described. There has been one previous
(11) (a) Knight, A. W.; Greenway, G. M. Analyst 1994, 119, 879. (b)
Debad, J. D.; Lee, S. K.; Qiao, X.; Pascal, R. A., Jr.; Bard, A. J. Acta
Chem. Scand. 1998, 52, 45. (c) Maloy, J. T.; Bard, A. J. J. Am. Chem. Soc.
1971, 93, 5968. (d) Bezman, R.; Faulkner, L. R. J. Am. Chem. Soc. 1972,
94, 6324.
(12) Werner, T. C.; Chang, J.; Hercules, D. M. J. Am. Chem. Soc. 1970,
92, 5560.
(13) (a) Debad, J. D.; Morris, J. C.; Lynch, V.; Magnus, P.; Bard, A. J.
J. Am. Chem. Soc. 1996, 118, 2374. (b) Debad, J. D.; Morris, J. C.; Magnus,
P.; Bard, A. J. J. Org. Chem. 1997, 62, 530.
(14) Salbeck, J.; Kunkely, H.; Langhals, H.; Saalfrank, R. W.; Daub, J.
Chimia 1989, 43, 6.
(15) McCord, P.; Bard, A. J. J. Electroanal. Chem. 1991, 318, 91.
(16) Plummer, B. F.; Steffen, L. K.; Herndon, W. H. Struct. Chem. 1993,
4, 279.

Electrochemical BehaVior of Diimide Compounds
J. Am. Chem. Soc., Vol. 121, No. 14, 1999 3515
Table 1. Cyclic Voltammetric Results for PI, PDI (-1, -2, -3),
TDI, and QDI Compoundsa
E1o
E1r
E2r
∆E°or ∆E°r
(V)
compd (V vs SCE) (V vs SCE) (V vs SCE) (V)
Ke
PI
1.37b
(1.31c)
1.62b
-1.10b
(-1.05c)
-0.60b
-1.66b
-
2.36 ∼0.56 3.1 × 109
PDI-1
PDI-2
PDI-3
TDI
-0.81b
(-0.75c)
-0.80b
(-0.75c)
-0.79b
(-0.73c)
-
2.11
2.18
2.17
1.75
1.40
0.21 3.6 × 103
0.22 5.3 × 103
0.23 7.9 × 103
0.07 15
(1.56c)
1.71b
(-0.55c)
-0.58b
(1.65c)
1.76b
(-0.53c)
-0.55b
(1.67c)
1.21b
(-0.50c)
-0.73b
(1.12c)
0.88b
(-0.63d)
-0.62b
(-0.70d)
-
QDI
0.04 5
(0.82c)
(-0.58d)
(-0.62d)
a Scan rate: 0.2 V/s in CHCl3:MeCN (see caption, Figure 2 for
ratios):0.1 M TBAPF6. b Peak potential. c Half-wave potential. d Half-
wave potential obtained by digital simulation. e Equilibrium constant
for comproportion reaction between R and R2-
.
CV of PI revealed one reversible 1e reduction peak at -1.10
V vs SCE (V ) 0.2 V/s) with a second irreversible reduction
peak at -1.66 V. The electrochemistry of PI compared to its
parent compound, perylene, reflects the strong electron-accepting
nature of the carboximide substituent in PI. Thus, the first 1e
reduction potential of PI is less negative than that of perylene
(-1.68 V vs SCE).18
The measured redox potentials for the series of compounds
with different aromatic imide structures and N-substitutuents
are listed in Table 1. The first reduction potential shifts positively
for the imide compounds in the following order: PI < PDI,
TDI, and QDI. While there is little effect of the nature of the
N-substituent on the reduction potential, a large effect is
observed with changes in the aromatic imide structure. The
reduction potential of the conjugated hydrocarbon structure
reflects the electron affinity of the system. Comparing PDI to
PI, the potential of first reduction peak shifts from -1.10 to
-0.60 V, showing that the addition of another imide group to
the perylene makes the molecule much easier to reduce. The
potential difference between the first and second 1e reduction
waves is greatly reduced for PDI, indicating that the addition
of the second electron is also much easier with comparison to
PI.
Figure 2. Cyclic voltammograms of (a) 0.8 mM PI in CHCl3/MeCN
(3:2, v/v), (b) 0.3 mM PDI-1 in CH3CN/MeCN (2:1, v/v), (c) 0.8 mM
PDI-2 in CHCl3/MeCN (3:2, v/v), (d) 0.21 mM TDI in CHCl3/MeCN
(4:1, v/v), and (e) 0.1mM QDI in CHCl3/MeCN (4:1, v/v). Scan rate:
(a, b) ) 0.2 V/s; (c-e) ) 0.5 V/s (electrolyte: 0.1 M TBAPF6).
suitable for compounds containing nitrogen and oxygen.17 Normal
precision was used.
With the addition of fused rings and an increase in the
distance between the two imide groups, the spacing between
the two successive 1e reduction waves decreases until a single
2e wave (for TDI and QDI) results, although the reduction
potential of the first wave is not significantly affected. For TDI
and QDI, the current of the reduction peak is about twice that
of the oxidation peak, suggesting that the potential difference
between the first and second 1e reduction waves is greatly
reduced and the two peaks completely overlap. This result
suggests that in TDI and QDI the electronic charge is largely
localized in the imide groups, and there is relatively little
interaction between the groups. We discuss this further after
the molecular orbital calculations are described. The proposed
reduction mechanism for the diimide systems is shown in
Scheme 1, where the 1e transfer processes lead to the formation
of a radical anion and then the dianion. It is reasonable to assume
that the major portion of the electron density resides on the
carbonyl oxygen due to the electron-withdrawing nature of the
oxygen atom as indicated in possible resonance forms in Scheme
1. For PDI, the addition of the second electron is more difficult
than the first because of Coulombic repulsion between the pair
Results and Discussions
Electrochemistry. Cyclic voltammetry (CV) was used to
obtain information about the energetics for formation of the
species R•-, R2-, and R•+ and their stability. The chemical
structures of the compounds of interest are given in Figure 1.
The CV for these in CHCl3/MeCN solutions is shown in Figure
2. The perylene compounds (PDI), at scan rates (V) above 0.2
V/s, show chemically reversible waves for the stepwise reduction
[two one-electron (1e) transfers] and a 1e oxidation. In the cases
of TDI and QDI, however, the two-electron (2e) reduction
occurs in a single wave, and the oxidation in a 1e wave (about
one-half the height of the reduction wave, Figure 2d,e). At scan
rates below 0.2 V/s, for PDI-1 and TDI, the oxidation waves
become less reversible, suggesting that the radical cations
undergo slow decomposition, although in a previous study of
PDI-3 in MeCN, good reversibility was seen at 0.05 V/s.14
Nevertheless, the radical cation was sufficiently stable for ECL
generation, as discussed below. The radical cation of QDI was
more stable, and the reversibility of the QDI oxidation wave
varied little for scan rates as low as 0.05 V/s.
(17) HyperChem Release 5.0 for Windows Reference Manual, Hyper-
cube, Inc, Publication HC50-00-02-00, October 1996.
(18) Parker, V. D. J. Am. Chem. Soc. 1976, 98, 98.

3516 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
Lee et al.
Scheme 1. Proposed Two-Step Reduction Mechanism for
the Diimide Molecules
Scheme 2. Proposed Oxidation Reaction for Generating the
Diimide Radical Cation
Figure 3. Scan rate variations of cyclic voltammograms by the
experimental (s) and digital simulations (0) of 0.21 mM TDI in CHCl3/
CH3CN (4:1, v/v), (electrolyte: 0.1 M TBAPF6): (a) ) 0.2, (b) ) 2.0,
and (c) ) 20 V/s at a 1.4 mm diameter Pt electrode. The experimental
data were corrected for Ohmic drop. Simulation parameters: reduction
reaction E1r ) -0.63 V vs SCE, E2r ) -0.70 V vs SCE, R ) 0.5, ks
) 1.0 cm/s; oxidation reaction E1o ) + 1.21 V vs SCE, R ) 0.5, ks )
0.0035-0.05 cm/s, kf ) 0.1 s-1
.
out of the plane of the hydrocarbon, and substituents on the
N-phenyl thus cause only small effects. Structures and energies
based on molecular orbital calculations are discussed in a later
section. Thus, the redox behavior of the diimide molecules is
not affected significantly by N-substituents, indicating little
contribution of these to the HOMO and LUMO. With extension
of the aromatic system, the oxidation of the diimide compound
becomes easier. Second irreversible oxidation waves for the PDI,
TDI, and QDI are also observed at more positive potentials,
corresponding to the oxidation reaction of the radical cation to
a reactive dication.
Digital Simulation for Electrochemical Reactions of TDI
and QDI. As shown in Figure 2, the height of the reduction
peak for TDI and QDI is almost double that of the oxidation
peak. We assumed above that two 1e transfer processes occur
within a very narrow potential range and that the reduction peaks
are superimposed. Digital simulation of the CVs, using the
proposed reduction and oxidation mechanisms in Schemes 1
and 2, was carried out and compared with the experimental
results. This allowed an estimate of the potentials of the two
superimposed reduction peaks. In performing the digital simula-
tions, it is necessary to assume rate constants for the hetero-
geneous electron-transfer reactions (k°) and compare the curves
to experimental data corrected for uncompensated Ohmic drop.
of electrons interacting through the π-orbitals of the intervening
perylene spacer. The second reduction potential should be
affected by the resonance energy and the ability to delocalize
the first electron density. The extended aromatic fused rings
will increase the resonance energy and delocalize the negative
charge more effectively. Therefore, the second 1e reduction
potentials become much closer to the first reduction potentials
in the case of TDI and QDI, and the Coulombic repulsion in
the dianions is rendered weaker. The reduction of the diimide
systems gives rise to an increased quinoidal character due to
charge separation by the ring structure, which minimizes the
electron-electron repulsion.19 The dianion form in Scheme 1
is represented by reduced cis-carbonyls. It can also be repre-
sented by reduced trans-carbonyls, but this requires the loss of
aromaticity in the pery-, terry-, or quaterry-lene spacers.
The proposed 1e oxidation reaction for generation of the
diimide radical cation is shown in Scheme 2. The odd electron
is delocalized over the system. A comparison of PDI-1, -2, and
-3 shows that the N-substituents have little influence on the
oxidation reaction. In all cases, the N-phenyl rings are twisted
(19) Viehbeck, A.; Goldberg, M. J.; Kovac, C. A. J. Electrochem. Soc.
1990, 137, 1460.

Electrochemical BehaVior of Diimide Compounds
J. Am. Chem. Soc., Vol. 121, No. 14, 1999 3517
Figure 5. Normalized absorption (s) and fluorescence spectra (‚‚‚)
of (a) 1.0 × 10-5 M PDI-1, (b) 1.0 × 10-5 M TDI, and (c) 1.0 × 10-5
M QDI in CH3Cl/CH3CN (3:2, v/v).
For QDI, the two 1e reduction potentials are closer than for
TDI. To avoid any uncertainty caused by other parts of the CV,
only the reduction wave was simulated. As mentioned above,
the reduction of QDI can be treated as Nernstian only for scan
rates below 5 V/s; simulations for these were carried out (Figure
4). The potentials of the two 1e reduction waves were given as
-0.58 and -0.62 V vs SCE, respectively.
The equilibrium constant for the comproportionation reac-
tion20
R + R2- ) 2R-
for formation of the radical anion (K) can be calculated by
(7)
Figure 4. Scan rate variations of cyclic voltammograms by the
experimental (s) and digital simulations (0) of 0.15 mM QDI in CHCl3/
CH3CN (4:1, v/v), (electrolyte: 0.1 M TBAPF6): (a) ) 0.05, (b) )
0.2, and (c) ) 1 V/s at a 3.4 mm diameter Pt electrode. The
experimental data were corrected for Ohmic drop. Simulation param-
eters: reduction reaction E1r ) -0.58 V vs SCE, E2r ) -0.62 V vs
SCE, R ) 0.5, ks ) 0.5 cm/s.
log K ) (E°2r - E°1r)/0.059 ) ∆E°r/0.059
(8)
The K values are listed in Table 1. The difference between the
first and second reduction potentials (∆E°r) provides a measure
of the thermodynamic stability of the radical anion. For PI and
PDI, the values of K are quite large, suggesting good stability
of the radical anions. The much smaller K values for TDI and
QDI imply facile reduction, low concentrations of radical ions,
and low ECL intensities.
However, one can only say that the k° values are large, i.e., g1
cm/s, since the waves are near those expected for Nernstian
response and the measured k° values are very sensitive to
uncompensated resistance.
UV-Vis Spectroelectrochemistry and Photoluminescence.
The absorption and emission spectra of PDI-1, TDI, and QDI
are shown in Figure 5 (see also ref 2f). PDI-1 exhibits maxima
at 521 and 536 nm in the absorption and fluorescence spectra,
respectively. PDI-2 and 3 closely resemble PDI-1 in both fine
structure of the absorption and fluorescence, and location of
the maxima, indicating little sensitivity to the N-substituents in
the HOMO and LUMO of the diimide compound. TDI shows
maxima in the absorption and fluorescence red-shifted from PDI
at wavelengths of 650 and 673 nm, respectively, while those
of QDI reach the near-IR region with wavelengths of 762 and
783 nm, respectively. The photoluminescence is discussed in
detail elsewhere.2f The lowest ground-state absorbance in the
UV-vis region and the S1-S0 fluorescence emission of the
diimide species shifts to longer wavelengths with increasing
π-conjugation. As additional aromatic rings are added to the
diimide system, additional bonding and antibonding levels are
introduced with a resulting reduction of the energy of the lowest
π f π* transition. This trend is also reflected by the decrease
of the potential differences of the first oxidation wave and the
first reduction wave (∆E°or) as the π-conjugated system is
The scan rate dependence of the reduction and oxidation
current peaks of TDI showed a linear relationship between ipc
and ipa and V1/2 at scan rates up to 20 V/s, indicating diffusion
control. For QDI, however, at scan rates above 5 V/s, the data
for the reduction deviated from the line, indicating, perhaps,
quasi-reversibility (results for TDI and QDI are given in the
Supporting Information). Experimental and simulated voltam-
mograms at different scan rates are shown in Figures 3 and 4.
For TDI, the theoretical and experimental voltammograms match
well for the reduction wave over the scan rate range from 0.2
to 20 V/s. For the oxidation, however, there are small deviations,
especially at lower scan rates. To obtain a satisfactory fit to the
observed cyclic voltammogram, a decomposition reaction of the
TDI radical cation was taken into consideration. The rate
constant for this reaction was assumed to be 0.1 s-1. At high
scan rates, such as V ) 20 V/s (Figure 3c), the simulation was
not affected by this reaction. Some deviation in this wave may
also arise from failure to consider the effect of the second
oxidation wave in the simulations. The simulation allows one
to estimate the first and second reduction potentials, E°1r and
E°2r, as -0.63 and -0.70 V vs SCE, respectively.
(20) Michaelis, L. Chem. ReV. 1935, 16, 243.

3518 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
Lee et al.
Table 2. Energies Available in the Annihilation Reactions and
Needed To Generate Excited Singlet Forms
-∆G°ann (eV)a
-∆H°ann (eV)
Es (eV)b
PDI-1
TDI
QDI
2.11
1.75
1.40
2.01
1.65
1.30
2.31
1.84
1.58
a Obtained on the basis of the difference between half-wave potentials
of first oxidation and first reduction peaks in CV. b Calculated from
the highest energy emission peak in fluorescence spectrum.
of the lower resolution of the spectrometer used to measure the
ECL.
The relative ECL efficiency (φECL) was measured using Ru-
2+
2+
(bpy)3 as the reference standard (φECL of Ru(bpy)3 taken
as 0.05 in MeCN at 25 °C for the annihilation reaction).22 The
values are around 0.04 and 0.03 for PDI-2 in CHCl3/MeCN
(3:2, v/v) and TDI in CHCl3/MeCN (4:1, v/v), respectively. The
ECL of QDI was much weaker than PDI and TDI, and the
spectrum could not be obtained using the CCD camera. We
could only detect the undispersed ECL signal with a photo-
multiplier.
Figure 6. Normalized ECL spectra of (a) 8.0 × 10-4 M PDI-2 in
CHCl3/CH3CN (v/v, 3:2) and (b) 2.0 × 10-4 M TDI in CHCl3/CH3CN
(4:1, v/v). The potential was stepped (a) from -0.65 to +1.80 V and
(b) from -0.80 to +1.40 V. The exposure time was 15 min
(electrolyte: 0.1 M TBAPF6).
For both ECL and fluorescence spectra, the singlet is the
emitting state. The energy available in the radical cation/radical
anion annihilation reaction is given (in eV) by10
extended (Table 1), since this potential difference is also a
measure of the HOMO-LUMO energy gap.
The UV-vis spectra of PDI-1 and TDI during reduction
under potentiostatic control are shown in Figures S2 and S3 in
the Supporting Information. The spectra of PDI-1 during
electrolysis at -0.60 V show the growth of new absorbance
peaks at 678, 696, 705, 761, and 791 nm, which are attributed
to the formation of the radical anion. The radical anion forms
are stable and have higher absorbances in the visible and IR
regions than the corresponding neutral forms. Further reduction
of PDI-1 at -0.80 V gives rise to the growth of the dianion
peaks at 530, 564, and 596 nm. These results are consistent
with the UV-vis data for reduced PDI-3 previously re-
ported.14,19,21 PDI-1 can be reversibly cycled among the neutral
(yellow), radical anion (green), and dianion (violet) forms. For
TDI, an absorbance appears at around 900 nm when the
reduction is carried out at -0.65 V. No new peaks were
observed in the UV-vis region when QDI was reduced at -0.6
V.
-∆H°ann ) -∆G°ann - T∆S° ≈ (E°1,0 - E°1,r) - 0.1 eV
(9)
∆G°ann can be obtained from the difference between the standard
potentials of the first oxidation and first reduction waves in the
cyclic voltammogram, yielding the values for the compounds
studied here and listed in Table 2. The energy for the first excited
singlet generation, Es , was obtained from the emission spectrum.
For ECL, via the S-route, the following equation must be
satisfied:
-∆Hann g Es
(10)
Thus, in an energy-sufficient system, the energy provided by
the annihilation reaction is large enough to produce a singlet
directly. Table 2 clearly shows that, for the diimide compounds
studied in this paper, the energy supplied by the ion-annihilation
reaction is smaller than that required for direct production of
the singlet state. In these energy-deficient systems, the ECL
cannot undergo the S-route as expressed by eqs 1-4. However,
the singlet can be produced via the T-route (eqs 5 and 6), where
the annihilation reaction produces a triplet, and the luminescent
singlet state is formed via triplet-triplet annihilation.10 The
proposed T-route involves the inefficient triple-triple annihila-
tion process, which results in the relatively low ECL efficiency.
There are several reasons for the much lower ECL efficiency
of QDI besides the inefficient T-route. First, as shown in Table
1, the value of K for QDI is very small, suggesting much poorer
stability of the QDI radical anion. When the Pt electrode is
pulsed to generate radical cations and anions stepwisely, the
disproportionation of QDI•- will greatly reduce the ECL
efficiency. Second, the photoluminesence efficiency of QDI is
low, ∼0.05,2f and its emission band lies in the near-IR region,
where excited states are easily quenched.
Electrogenerated Chemiluminescence. The PDI compounds
are good candidates for ECL because of their stable radical ions
and high fluorescence efficiencies. When a Pt working electrode
immersed in a solution of CHCl3/CH3CN (3:2, v/v) containing
PDI-2 (∼0.8 mM) was pulsed between the compound’s first
oxidation and reduction waves, ECL emission was produced
by the annihilation reaction. The ECL spectrum (Figure 6a) was
in the same region as the photoluminescence (PL) spectrum with
a maximum at 555 nm and a shoulder around 600 nm. The
ECL spectrum of 0.2 mM TDI in CHCl3/CH3CN (4:1, v/v)
(Figure 6b) shows an emission maximum at 679 nm. The ECL
emissions from both PDI and TDI have slightly different shapes
and maxima compared to their fluorescence spectra. The PL
and ECL spectra should be identical, since the same singlet state
is responsible for the emission in both cases. The slight red-
shift and change in relative peak heights of the ECL spectra
can be explained by self-absorption at the shorter wavelengths
(an inner-filter effect) in the ECL solution, which was more
concentrated than the solutions used for fluorescence measure-
ments. The ECL spectra in Figure 6 also do not exhibit the fine
structure shown in the fluorescence spectra (Figure 5b) because
Molecular Orbital Calculations. To correlate the observed
electrochemical behavior of this series of compounds with their
molecular structure and to obtain a better understanding of the
effect of core size on the spacing of the two reduction waves,
molecular orbital calculations were undertaken to examine
(21) Ford, W. E.; Hiratsuka, H.; Kamat, P. V. J. Phys. Chem. 1989, 93,
6692.
(22) Wallace, W. L.; Bard, A. J. J. Phys. Chem. 1979, 83, 1350.

Electrochemical BehaVior of Diimide Compounds
J. Am. Chem. Soc., Vol. 121, No. 14, 1999 3519
Table 3. Selected Structural Data for the Various N-Substituents in
the PDI Species
N-substituents with the core in the PDI series is consistent with
the rather small differences in electrochemical potentials for
oxidation and reduction when this substituent is changed.
bond angle of
atoms (deg)
torsion angle of
atoms (deg)
The calculated (PM3) energies of the HOMOs and LUMOs
of the parent compounds of this series are shown in Figure 7.
The oxidation of these compounds should correlate with the
energies of the HOMOs (Figure 8a). The addition of a second
imide group to PI to form PDI results in a decrease in the
HOMO energy, consistent with the more difficult oxidation of
PDI. Within the diimides, extension of the aromatic fused rings
results in an increase in the HOMO energy, again consistent
with the observed less positive potentials of electrochemical
oxidation of TDI and QDI compared to PDI.
substituent
1, 4, 5 1, 4, 6 1, 2, 4, 5 1, 3, 4, 6
R1 ) R4 ) methyl, isopropyl, ∼118.8 ∼118.8 ∼90
∼90
tert-butyl; R2 ) R3 ) H
R1 ) R3 ) tert-butyl;
R2 ) R4 ) H
∼123.9 ∼114.5 ∼95.3
∼84.7
The reduction of the compounds should correlate with the
location of the LUMO. The remarkably lower energies of the
LUMOs of the diimide compounds compared to PI are also
consistent with the observed potentials for reduction (Figure
8b). The differences between the LUMO and HOMO of a given
molecule are also in good agreement with the trends seen in
∆E°or and the location of the spectral adsorption and emission
bands (Figure 8c). Of particular interest is the merging of the
first and second reduction waves with the increase in the size
of the aromatic core. This can be understood from the electron
distribution in the parent, radical anion, and dianion and the
relative splitting of the first and second LUMOs. The charge
distribution at each atom was calculated for each diimide and
its reduced forms. The electronic charge is largely located at
the carbonyl oxygen atoms (Supporting Information Table S1),
with additional charge distributed over the connecting aromatic
fused ring system. The potentials of the reduction steps are
affected by the locations of the LUMOs. The reduction to the
radical anion occurs by addition of an electron to the lowest
LUMO. It is known from studies of many aromatic species that
Figure 7. Molecular orbitals calculated with PM3.
orbital energies and charge distributions. First, the structure of
each molecule was geometrically optimized using the MM+
force field to find the structure of lowest energy. In all cases,
the diimide core was found to be planar, consistent with the
structure that would yield the largest resonance stabilization.
The N-substituents, i.e., the phenyl rings or alkyl groups are
located in a plane that is essentially perpendicular to the diimide
core. Several bond angles and torsion angles illustrating this
are listed in Table 3. With PDI-3, the N-phenyl ring is twisted
a bit because of the steric hindrance caused by the unsym-
metrical substituents. The general lack of interaction of the
Figure 8. Correlation of experimental data with molecular orbital calculation results: (a) oxidation potential versus HOMO energy, (b) reduction
potential versus LUMO energy, (c) potential difference between the first oxidation and first reduction waves versus energy difference between
HOMO and LUMO, and (d) potential difference between the first and second reduction waves versus energy difference between the first and
second LUMO.

3520 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
Lee et al.
addition of a second electron to the same orbital, corrected for
better solvation of the dianion, results in a potential difference
of about 0.5-0.6 V.23 This results from the considerable
repulsion energy for two electrons in the same orbital (pairing
energy). Addition of a second electron into the next LUMO,
equivalent to the addition of an electron into the second imide
group, can involve a lower energy. Although this orbital is not
degenerate with the lowest LUMO, when one takes into account
the larger stabilization of the dianion by solvation and ion
pairing, one can rationalize the very close spacings of the two
reduction waves in TDI and QDI. The trend in ∆E°r values for
the series follows the observed decrease between the two
LUMOs (Figure 8d).
The question of the extent of interaction of two identical
electroactive groups on the same molecule and its effect on the
voltammetric response has been the subject of a number of
studies.24,25 Strong interaction results in a large splitting between
the reduction waves on the electroactive moieties, while no
interaction produces a single wave for the simultaneous reduc-
tion of both, with the appearance of a single 1e wave
representing the statistical spacing of the E°’s of (2RT/nF) ln 2
(36 mV at 25 °C). When the electroactive groups are coupled
with a saturated chain, even as short as -C2H4-, the groups
show little or no interaction. Polyene chains show interactions
over longer distances. Studies of groups such as ferrocene,
viologen, or Ru(bpy)32+ linked by polyenes of different length
have shown that the interaction decreases with the number of
interposed double bonds, n. Typically, there is little interaction
when n ) 5 or 6.26 If the chain is a polymethine-type, where
solitonic behavior is possible, interactions over longer distances
(e.g., n g 9) can occur.27 A chain that includes aromatic moieties
may show smaller interactions over similar distances, e.g.,
because of charge localization on these.28 The work described
here suggests similar decreased interactions across rigid aromatic
units.
Conclusions
The electrochemical behavior of several diimide compounds
(PDI, TDI, and QDI) was examined and compared to that of
the parent imide, PI. While the N-substituents have little
influence on the electrochemical behavior, a large effect is noted
with extension of the aromatic core between the imide groups.
As the core size increases, oxidation becomes easier and the
potentials for the two reduction steps become closer, so that
for TDI and QDI only a single overlapped 2e wave is seen.
Thus the tendency for the radical anion to disproportionate
follows the series QDI > TDI . PDI. These results can be
rationalized by molecular orbital calculations.
The adsorption and emission spectra exhibit a significant red
shift from PDI to the near-IR with QDI that reflects orbital
energy changes. ECL emission is observed at a Pt electrode for
these systems, when the potential is pulsed between the first
oxidation and first reduction waves. The ECL efficiency is
smaller for TDI and QDI, probably because of the extensive
disproportionation of the radical anions and lower fluorescence
quantum efficiencies.
Acknowledgment. Funding by the Robert A. Welch Foun-
dation, Texas Advanced Research Program, and IGEN Inc. is
gratefully acknowledged.
Supporting Information Available: Scan rate dependence
for cyclic voltammetry of TDI and QDI, UV-vis spectra for
reduction of PDI-1 and TDI, and calculated fractional electron
densities (PDF). This material is available free of charge via
the Internet at http://pubs.acs.org.
(23) Peover, M. E. In Electroanalytical Chemistry; Bard, A. J., Ed.;
Marcel Dekker: New York, 1967; Vol. 2, Chapter 1.
(24) Ammar, F.; Save´ant, J. M. J. Electroanal. Chem. 1973, 47, 115.
(25) Flanagan, J. B.; Margel, S.; Bard, A. J.; Anson, F. C. J. Am. Chem.
Soc. 1978, 100, 4248 and references therein.
(26) (a) Ribou, A.-C.; Launay, J.-P.; Sachtlebeu, M. L.; Li, H.; Spangler,
C. W. Inorg. Chem. 1996, 35, 3735. (b) Benniston, A. C.; Goulk, V.;
Harriman, A.; Lehn, J.-M.; Marczinke, B. J. Phys. Chem. 1994, 98, 7798
and references therein.
JA984188M
(27) (a) Tolbert, L. M.; Zhao, X.; Ding, Y.; Bottomley, L. A. J. Am.
Chem. Soc. 1995, 117, 12891. (b) Tolbert, L. M. Acc. Chem. Res. 1992,
25, 561.
(28) See, for example: Tiemann, B. A.; Cheng, L.-T.; Marder, S. R. J.
Chem. Soc., Chem. Commun. 1993, 9, 735.