Core-modified naphthalenediimides generate persistent radical anion and cation: New panchromatic NIR probes

Article abstract of DOI:10.1021/ol302140x

The generation of the first persistent radical cation of naphthalenediimide with Cu²⁺ /Fe³⁺ under ambient conditions is reported. An alternate anionic trigger generates a persistent radical anion within the same motif. Steric protection and H-bonding enhances the half-life of radical cation by 290-fold. The radical anion and cation have orthogonal spin density, panchromatic and NIR optical bands, which can be applied as attractive multichannel probes.

Full text of DOI:10.1021/ol302140x

ORGANIC
LETTERS
2012
Vol. 14, No. 18
4822–4825
Core-Modified Naphthalenediimides
Generate Persistent Radical Anion and
Cation: New Panchromatic NIR Probes
M. R. Ajayakumar, Deepak Asthana, and Pritam Mukhopadhyay*
Supramolecular and Material Chemistry Lab, School of Physical Sciences,
Jawaharlal Nehru University, New Delhi 110 067, India
*m_pritam@mail.jnu.ac.in
Received August 1, 2012
ABSTRACT
The generation of the first persistent radical cation of naphthalenediimide with Cu2þ/Fe3þ under ambient conditions is reported. An alternate
anionic trigger generates a persistent radical anion within the same motif. Steric protection and H-bonding enhances the half-life of radical cation
by 290-fold. The radical anion and cation have orthogonal spin density, panchromatic and NIR optical bands, which can be applied as attractive
multichannel probes.
New π-extended organic radical and radical ions have
attracted huge interest for their spin-based properties.1 In
contrast, their attractive low-energy optical properties in
the visible (vis) and near-infrared (NIR) region due to
D0fDn transition have been mostly overlooked.2 Conse-
quently, their potential as new multichannel, wide-optical
probes remains untapped, primarily because of their tran-
sient nature and complex generation processes.
Most significantly, design and synthesis of an organic
π-motif that would generate a persistent radical anion as well
as a cation by alternate trigger of a chemical stimulus has
remained elusive. Such a molecule, by virtue of its distinct
electronic levels and spin/charge distribution, would have
the unique ability to harness and switch the orthogonal
optical, electronic, etc. properties. This would demand (a)
easily accessible HOMO and LUMO levels, (b) noninter-
feringfunctionalgroupsforseamlessgain/lossofelectrons,
and (c) precise structural and electronic factors to impart
stability to the radical anion and cation.
Herein, we demonstrate for the first time formation of a
persistent radical cation in a naphthalenediimide (NDI)
moiety and simultaneous generation of radical anion by
simple chemical methods, under ambient conditions. The
radical cation and anion reveal orthogonal spin distribu-
tion and nonoverlapping intense absorptions in the UV,
whole visible, and NIR region. Importantly, H-bonding
and steric protection at the NDI-core (c-NDI) can enhance
half-life of the radical cations by 290-fold. Furthermore,
the multichannel optical properties could be applied as
panchromatic, narrowband NIR probes for metal ions as
well as anions having diverse charge/size.
Extensive applications of the versatile NDI moiety
stem from its large π-surface and low-LUMO levels.3,4
Our group and others have recently shown that axially
(1) (a) Torrent, M. M.; Crivillers, N.; Rovira, C.; Veciana, J. Chem.
Rev. 2012, 112, 2506–2527. (b) Fukuzumi, S.; Ohkubo, K.; D’Souza, F.;
Sessler, J. L. Chem. Commun. 201210.1039/C2CC32848H. (c) Miller, J. S.
Chem. Soc. Rev. 2011, 40, 3266–3296. (d) Hicks, R. G. Org. Biomol.
Chem. 2007, 5, 1321–1338. (e) Nishinaga, T.; Komatsu, K. Org. Biomol.
Chem. 2005, 3, 561–569.
(3) (a) Holman, G. G.; Foote, M. Z.; Smith, A. R.; Johnson, K. A.;
Iverson, B. L. Nat. Chem. 2011, 3, 875–881. (b) Katz, H. E.; Lovinger,
A. J.; Johnson, J.; Kloc, C.; Seigrist, T.; Li, W.; Lin, Y.-Y.; Dodabalapur,
A. Nature 2000, 404, 478–481. (c) Mamada, M.; Bolιvar, C. P.; Anzenbacher,
P., Jr. Org. Lett. 2011, 13, 4882–4885.
(4) (a) Sakai, N.; Mareda, J.; Vauthey, E.; Matile, S. Chem. Commun.
ꢀ
€
(2) Rathore, R.; Abdelwahed, S. H.; Guzei, I. A. J. Am. Chem. Soc.
2004, 126, 13582–13583.
2010, 46, 4225–4237. (b) Wurthner, F.; Stolte, M. Chem. Commun. 2011,
47, 5109–5115.
r
10.1021/ol302140x
Published on Web 09/06/2012
2012 American Chemical Society

modified NDI (a-NDI) can accept electrons from specific
amines5a or anions5b,c to form persistent radical anions.
However, a huge increase in NDIs gamut of applications
can be anticipated if radical cation is generated within the
NDI moiety. Although a wide variety of donors have been
incorporated in c-NDI resulting in charge-separation and
fascinating optical properties,6 generation of persistent
radical cation in the NDI moiety has remained elusive
until now.
Toward this goal, in our design, we systematically
varied the donor (D) amino units by incorporating steric,
H-bonding and lipophilic groups (1aꢀe) (Scheme 1a).
Theoretical calculations predicted a significant increase
of the HOMO level to ꢀ5.5 eV (vs ꢀ7.00 eV in a-NDI) and
an increase in the LUMO level to ꢀ3.0 eV (vs ꢀ3.8 eV in a-
NDI).4 1aꢀe were synthesized from 2,3-dibromonaphtha-
lene-1,4,5,8-tetracarboxylic acid bisimide by reacting with
the respective amines. All of the molecules were character-
ized in detail (Supporting Information, SI).
A solvent system of MeCN/CHCl3 (8:2 v/v) was required
toattain uniform solubility of1aꢀe. The redox potential of
Cu2þ/Cuþ couple was found to be þ1.01 V vs SCE in this
solvent system. Thus, addition of Cu2þ to 1a, resulted in an
instantaneous color change from deep-blue to turquoise
blue. Generation of 1a•þ was further confirmed by ESR
spectroscopy with a strong signal at g = 2.0029 and peak-
to-peak width of 10G, while with excess of Cu2þ, char-
acteristic EPR signal of Cu2þ along with the signal of 1a•þ
was verified (Figures S2 and S3, SI).
Next, we explored the ability of 1a to form a radical
anion with reducing anions. Indeed, in the presence of
cyanide anion in THF, a sharp color change from deep-
blue to brown-red was observed within 10 min. The slow
color transition was attributed to the endergonic ET
(ΔGET > 0).7,81a•ꢀ was confirmed by ESR with a strong
resonance at g = 2.0033 and peak-to-peak width of 5G
(Figure S2, SI). Similar results were obtained for 1bꢀe.
Scheme 1. (a) Chemical Structures (1aꢀe) of c-NDI Molecules;
(b) Reaction Scheme toward Radical Anion and Radical Cation
Figure 1. (a) UVꢀvisꢀNIR absorption spectra of 1a with Cu-
(ClO4)2 (0ꢀ4 equiv) [1a; 2 ꢁ 10ꢀ4 M in MeCN/CHCl3 (8:2)] and
(b) 1a with TBACN (0ꢀ4 equiv), each spectrum recorded after
30 min of equilibration [1a; 2 ꢁ 10ꢀ4 M in THF].
We further confirmed the formation of 1a•þ and 1a•ꢀ by
UVꢀvisꢀNIR absorption spectroscopy (Figure 1a,b).
Upon gradual addition of Cu2þ to 1a, new bands
evolved at 290, 464, 669, 768, and 874 nm with a con-
comitant decrease of the πꢀπ* 346 and 361 nm bands and
the internal charge-transfer (ICT) band at 613 nm. A 1:1
stoichiometric dependence of 1a/Cu2þ on the ET was
confirmed (Figure S4, SI). Compounds 1bꢀd show similar
spectra upon Cu2þ addition, while the lowest energy band
appears at 1025 nm in 1e. The D0fDn transitions qualita-
tively match with the calculated values (Table S2, SI).
On the other hand, with CNꢀ the πꢀπ* and ICT bands
diminish with formation of new bands at 494, 533, 678,
778, and 891 nm (Figure 1b). These signature absorption
bands fairly match with the theoretical results and with
that of the electrochemically generated c-NDI•ꢀ.9
First, theabilityofthec-NDIstodonateaswellasaccept
electrons was verified by cyclic voltammetry (CV) and
differential pulse voltammetry (DPV). Gratifyingly, 1a
show two well-separated reversible oxidation waves at
þ0.98 and þ1.34 V and one pseudoreversible reduction
wave at ꢀ0.89 V vs SCE. The experimentally determined
value of HOMO and LUMO for 1a was found to be ꢀ5.39
and ꢀ3.51 eV, respectively. Expectedly, 1bꢀe exhibit
similar redox properties, and the results are in line with
similar donors (Table S1 and Figure S1, SI).6c
The first oxidation potential E1ox of 1aꢀe being <1.0 V
(vs SCE), an exergonic thermal electron transfer (ET)
(ΔGET < 0) to Cu2þ to form c-NDI•þ can be predicted.7
(5) (a) Ajayakumar, M. R.; Mukhopadhyay, P. Chem. Commun.
2009, 25, 3702–3704. (b) Ajayakumar, M. R.; Yadav, S.; Ghosh, S.;
Mukhopadhyay, P. Org. Lett. 2010, 12, 2646–2649. (c) Guha, S.; Saha.,
S. J. Am. Chem. Soc. 2010, 132, 17674–17677.
Importantly, the clean spectral plots and clear isosbestic
points refer to uncluttered ET processes to form 1a•þ/1a•ꢀ.
Also, significantly, the 1a•þ/1a•ꢀ produce nonoverlapping
€
(6) (a) Bhosale, S.; Sisson, A. L.; Talukdar, P.; Furstenberg, A.;
€
€
Banerji, N.; Vauthey, E.; Bollot, G.; Mareda, J.; Roger, C.; Wurthner,
F.; Sakai, N.; Matile, S. Science 2006, 313, 84–86. (b) Naomi, S.; Lista,
M.; Kel, O.; Sakurai, S.-i; Emery, D.; Mareda, J.; Vauthey, E.; Matile, S.
J. Am. Chem. Soc. 2011, 133, 15224–15227. (c) Fin, A.; Petkova, I.;
Doval, D. A.; Sakai, N.; Vauthey, E.; Matile, S. Org. Biomol. Chem.
2011, 9, 8246–8252.
(8) (a) The endergonic reaction is explained by formation of stable
cyclic products from the cyanyl radical: Guo, T.; Illies, A.; Cammarata,
V.; Arndt, M.; Sonzogni, W. J. Electroanal. Chem. 2007, 610, 102–105.
(b) Additional stabilization is due to the well-known R-effect: Edwards,
J. O.; Pearson, R. G. J. Am. Chem. Soc. 1962, 84, 16–24.
(9) Chopin, S.; Chaignon, F.; Blart, E.; Odobel, F. J. Mater. Chem.
2007, 17, 4139–4146.
(7) ΔGET = Eox ꢀ Ered ꢀ e2/dε, in polar solvent, the Coulombic term
is neglected. ΔGET (1a•þ) = 0.98 V ꢀ 1.01 V = (ꢀ)0.03 V; Eox (CNꢀ) =
0.87 V in THF.
Org. Lett., Vol. 14, No. 18, 2012
4823

intense bands in the UVꢀvis-NIR region. Importantly, the
addition spectra of 1a•þ and 1a•ꢀ cover substantial part of
the solar spectra (Figure 2a). This would be highly attrac-
tive as new panchromatic,10a NIR dyes and for optimum
solar energy utilization in solar cells.10b
Furthermore, 1a•þ could regenerate 1a upon removal of
the MeCN solvent under Ar, which in turn produced the
1a•ꢀ upon addition of CNꢀ in THF (Figure 2b). The
radical anion could also be switched to the radical cation
upon addition of excess Cu2þ, making it important for
applications in molecular switches, logic gates, etc.
Figure 3. (a) Atom numbering scheme. Spin density of (b) 1b•þ
and (c) 1b•ꢀ. Resonance structures of (d) c-NDI•þ and (e)
c-NDI•ꢀ
.
Figure 2. UVꢀvisꢀNIR absorption spectra of (a) 1a•þ, 1a•ꢀ and
their addition spectra; (b) switching of 1a•þ to 1a upon removal
of MeCN/CHCl3 under a stream of Ar and formation of 1a•ꢀ
with TBACN in THF.
Table 1. Spin Densitya Distributions Obtained by DFT Calcu-
lations at the B3LYP/6-31þ G (d,p) Level
spin density
spin density
atom
1b•þ
1b•ꢀ
atom
1b•þ
1b•ꢀ
To confirm the existence of differential spin density and
charge distribution in the radical anion and cation, DFT
calculations at the B3LYP/6-31þG(d,p) level of theory11
were performed with the smallest entities of 1b•þ/1b•ꢀ. The
spin density surface contour of 1b•þ clearly shows that the
spin is predominantly delocalized along the A-B molecular
axis of 1b•þ (Figure 3b). As expected, the two donor N
atoms N21 and N22 possess the maximum spin density
followed by the core-carbons C1, C5, C3, C7, C9 and C10
(Table 1). On expected lines, spin contours on the acceptor
>CdO groups and the imide Ns, N13 and N18 are of
negligible intensity.
In 1b•ꢀ, the spin density is orthogonally distributed to
that of 1b•þ and is delocalized along the A0-B0 axis
(Figure 3b). Core carbons C1, C4, C5, C8, and >CdO
groups have major spin density.
These results are in good agreement with the resonance
structures(Figure3d,e) and supporttheorthogonal optical
properties of c-NDI•þ/c-NDI•ꢀ.The delocalized spin dis-
tribution should add to the stability of radical ions.
Next, the effect of amino substituents on the stability of
radical ions was evaluated by time-dependent UVꢀvis-
NIR analysis of the optical bands (Figure 4a). Theoretical
studies along with IR, and NMR spectroscopy confirmed
C1,C5
þ0.119
þ0.009
þ0.064
ꢀ0.010
þ0.045
þ0.010
þ0.048
þ0.021
þ0.186
þ0.059
ꢀ0.022
þ0.066
C12,C17
N13,N18
C14,C19
O15,O20
N21,N22
þ0.001
ꢀ0.002
ꢀ0.010
þ0.022
þ0.254
þ0.045
ꢀ0.025
þ0.043
þ0.069
þ0.028
C2,C6
C3,C7
C4,C8
C9,C10
O11,O16
a Spin densities are calculated by natural population analysis. Note
about the þve/ꢀve signs: spin delocalization or the direct term is always
positive or null, whereas the spin polarization or the indirect term is
characterized by the alternation of positive or negative spin density.
intramolecular NꢀH---O H-bonding between the NꢀH
and CdO groups in 1aꢀd; e.g., the stretching frequency for
the CdO group is decreased by 22 and 5 cmꢀ1 in 1a vs 1e
(SI, FigureS5). Compound 1a•þ with bulky cyclohexyl and
H-bonding donor groups was found to be the most stable
and decayed slowly over a period of ten days under
ambient condition. The half-life of 1a•þ was found to be
40 h. In sharp contrast, 1b•þꢀd•þ with considerably less
bulky alkyl chains decay within 20ꢀ55 min. Compounds
1b, 1c, and 1d were found to have half-lives of 0.57, 0.39,
and 0.25 h, respectively. Compound 1e•þ devoid of any
H-bonding donor groups was found to be the least stable
with a half-life of only 0.14h. Therefore, steric protection1d
of the reactive radical cation site along with H-bonding
enhances the stability of 1a•þ by 290-times in comparison
to 1e•þ (Figure 4b).
(10) (a) Asthana, D.; Ajayakumar, M. R.; Pant, R. P.; Mukhopadhyay,
P. Chem. Commun. 2012, 48, 6475–6477. (b) Kolemen, S.; Bozdemir,
O. A.; Cakmak, Y.; Barin, G.; Ela, S. E.; Marszalek, M.; Yum, J.-H.;
€
Zakeeruddin, S. M.; Nazeeruddin, M. K.; Gratzel, M.; Akkaya, E. U.
Chem. Sci. 2011, 2, 949–954.
(11) (a) For details of DFT calculations, see the Supporting Informa-
The next line of theoretical studies confirmed that
intramolecular H-bonding is greatly promoted in c-NDI•þ
with H-bond donors; e.g., the NꢀH, NꢀH---O, N---O,
tion. (b) For Mulliken atomic charge distribution of 1a, 1b•þ, and 1b•ꢀ
see the Supporting Information (Table S4).
,
4824
Org. Lett., Vol. 14, No. 18, 2012

Figure 4. (a) Time-dependent UVꢀvisꢀNIR abs spectra of
1a•þꢀe•þ and (b) decay profile of 1a•þꢀe•þ; [1aꢀe; 2 ꢁ 10ꢀ4
M in MeCN/CHCl3 (8:2), 1 equiv of Cu(ClO4)].
and CdO bond distances in 1b•þ were 1.027, 1.790, 2.635,
and 1.236 A, respectively, while for 1b, the distances were
1.018, 1.845, 2.659, and 1.241 A, respectively. In the
absence of H-bonding in 1e•þ, the N---O distance is
increased to 2.870 A and the CdO bond is shortened to
1.224 A (Table S3, SI). In the subseries 1bꢀd, a direct
dependence of lipophilicity on the stability of 1b•þꢀd•þ
was observed. In going from 1b to d, lipophilicity calcu-
lated from AlogP12 value increases from 3.61 to 11.79.
Hence, 1d•þ would be more reactive compared to 1b•þ and
1c•þ driven by aggregation in this polar solvent, which
explains the overall stability 1a•þ > 1b•þ > 1c•þ > 1d•þ
> 1e•þ.
Figure 5. (a) UVꢀvisꢀNIR abs spectra and (b) selectivity profile
of 1a with different metal ions; (c) UVꢀvisꢀNIR abs spectra
and (d) selectivity profile of 1a with different anions. [1a; 2 ꢁ
10ꢀ4 M in MeCN/CHCl3 (8:2). Metal ions: 2 ꢁ 10ꢀ4 M in
MeCN/CHCl3 (8:2). Anions (TBA salts): 1 ꢁ 10ꢀ3 M in THF].
redox reaction. This might limit the use of the redox probe
in specific cases, e.g. in case of complete signal overlap, the
identity of an oxidizing/reducing analyte has to be verified
with a nonredox reference sensor.
Stability of the c-NDI•ꢀ was also dependent on intra-
molecular H-bonding (Table S3, SI); e.g., the N---O dis-
tance of 1b•ꢀ was 2.652 A, significantly shortened in
comparison to 2.790 A in 1e•ꢀ. The effect of the N
substituent and its lipophilicity on the distant imide radical
anion was significantly less pronounced. Compound 1a•ꢀ
had a half-life of 18 h, which is 14 times greater than that of
1e•ꢀ (Figure S6, SI).
In conclusion, the NDI moiety widely perceived as a
strong π-acceptor is shown for the first time to form a
persistent radical cation. With an alternate trigger the NDI
moiety alsoforms a persistent radical anion. Incorporation
of steric protection and H-bonding systematically en-
hances the half-life of the radical cation by 290-fold. The
orthogonal spin distribution and excellent optical property
of the radical ions coupled with the recent advances on c-
NDI modification,14 should cultivate enormous interests
pervading orthogonal ET reactions, switchable organic
materials and so on. We are currently working on the
supramolecular aspects of these ET reactions.
Finally, we explored the panchromatic, NIR optical
property of c-NDI•þ/c-NDI•ꢀ as multi-ion sensor probes
(Figure 5a-d). Compound 1a was found to be highly
selective and sensitive toward Cu2þ/Fe3þ, while Naþ,
Kþ, Mg2þ, Ca2þ, Mn2þ, Co2þ, Ni2þ, Agþ, and Pb2þ
remained silent. Cu2þ was 1.5 times more selective than
Acknowledgment. We thank DBT, CSIR, DST-
PURSE, and DST FIST-II for financial support, AIRF,
JNU, for NMR and IR data, Mr. J. Singh, CIF, JNU, for
MALDI-TOF MS, and Dr. R. P. Pant, NPL, New Delhi,
for EPR. We acknowledge Dr. A. Dey, IACS, Kolkata, for
his input on the theoretical calculations.
Fe3þ 13a,b
In addition, complete fluorescence turn-off of 1a
.
led to detection of Cu2þ at63ppb (FigureS7, SI). Similarly,
1a was found to be highly selective for CNꢀ.13cꢀe In
particular, 1a•þ with an intense NIR band at 874 nm, ε of
30300 Mꢀ1 cmꢀ1 and half-width (Δυ1/2) of only 652 cmꢀ1
h
(Figure S8, SI) makes it a promising probe in the NIR
region.
Supporting Information Available. Experimental pro-
cedures, synthesis, and various spectroscopic and theore-
tical data. This material is available free of charge via the
Internet at http://pubs.acs.org.
It is to be noted that signal transduction in the NDI
moiety is to a large extent dependent on the ΔGET of the
(12) AlogP was calculated with the software AlogPS 2.1, available
from VCCLAB, http://www.vcclab.org.
(13) Cu(II) NIR probes: (a) Sanna, E.; Martı
ꢀ
´
nez, L.; Rotger, C.;
(14) (a) Alvey, P. M.; Iverson, B. L. Org. Lett. 2012, 14, 2706–2709.
(b) Polander, L. E.; Romanov, A. S.; Barlow, S.; Hwang, D. K.;
Kippelen, B.; Timofeeva, T. V.; Marder, S. R. Org. Lett. 2012, 14,
918–921. (c) Bhosale, S. V.; Bhosale, S. V.; Bhargava, S. K. Org. Biomol.
Chem. 2012, 10, 6455–6468.
~
Blasco, S.; Gonzalez, J.; Garcı
´
a-Espana, E.; Costa, A. Org. Lett. 2010,
12, 3840–3843. (b) Sreenath, K.; Thomas, T. G.; Gopidas, K. R. Org.
Lett. 2011, 13, 1134–1137. CNꢀ probes: (c) Saha, S.; Ghosh, A.; Mahato,
P.; Mishra, S.; Mishra, S. K.; Suresh, E.; Das, S.; Das, A. Org. Lett. 2010,
12, 3406–3409. (d) Guliyev, R.; Ozturk, S.; Sahin, E.; Akkaya, E. U. Org.
Lett. 2012, 14, 1528–1531. (e) Liu, Y.; Lv, X.; Zhao, Y.; Liu, J.; Sun,
Y-Q; Wang, P.; Guo, W. J. Mater. Chem. 2012, 22, 1747–1750.
The authors declare no competing financial interest.
Org. Lett., Vol. 14, No. 18, 2012
4825