Boundaries of anion/naphthalenediimide interactions: From anion-π interactions to anion-induced charge-transfer and electron-transfer phenomena

Article abstract of DOI:10.1021/ja303173n

The recent emergence of anion-π interactions has added a new dimension to supramolecular chemistry of anions. Yet, after a decade since its inception, actual mechanisms of anion-π interactions remain highly debated. To elicit a complete and accurate understanding of how different anions interact with π-electron-deficient 1,4,5,8-naphthalenediimides (NDIs) under different conditions, we have extensively studied these interactions using powerful experimental techniques. Herein, we demonstrate that, depending on the electron-donating abilities (Lewis basicity) of anions and electron-accepting abilities (π-acidity) of NDIs, modes of anion-NDI interactions vary from extremely weak non-chromogenic anion-π interactions to chromogenic anion-induced charge-transfer (CT) and electron-transfer (ET) phenomena. In aprotic solvents, electron-donating abilities of anions generally follow their Lewis basicity order, whereas π-acidity of NDIs can be fine-tuned by installing different electron-rich and electron-deficient substituents. While strongly Lewis basic anions (OH^- and F^-) undergo thermal ET with most NDIs, generating NDI^?- radical anions and NDI ^2- dianions in aprotic solvents, weaker Lewis bases (AcO^-, H₂PO₄^-, Cl^-, etc.) often require the photoexcitation of moderately π-acidic NDIs to generate the corresponding NDI^?- radical anions via photoinduced ET (PET). Poorly Lewis basic I^- does not participate in thermal ET or PET with most NDIs (except with strongly π-acidic core-substituted dicyano-NDI) but forms anion/NDI CT or anion-π complexes. We have looked for experimental evidence that could indicate alternative mechanisms, such as a Meisenheimer complex or CH anion hydrogen-bond formation, but none was found to support these possibilities.

Full text of DOI:10.1021/ja303173n

Article
pubs.acs.org/JACS
Boundaries of Anion/Naphthalenediimide Interactions: From
Anion−π Interactions to Anion-Induced Charge-Transfer and
Electron-Transfer Phenomena
Samit Guha, Flynt S. Goodson, Lucas J. Corson, and Sourav Saha*
Department of Chemistry and Biochemistry and Integrative NanoScience Institute, Florida State University, 95 Chieftan Way,
Tallahassee, Florida 32306, United States
S
* Supporting Information
ABSTRACT: The recent emergence of anion−π interactions has added a new dimension to
supramolecular chemistry of anions. Yet, after a decade since its inception, actual mechanisms of
anion−π interactions remain highly debated. To elicit a complete and accurate understanding of how
different anions interact with π-electron-deficient 1,4,5,8-naphthalenediimides (NDIs) under different
conditions, we have extensively studied these interactions using powerful experimental techniques.
Herein, we demonstrate that, depending on the electron-donating abilities (Lewis basicity) of anions
and electron-accepting abilities (π-acidity) of NDIs, modes of anion−NDI interactions vary from
extremely weak non-chromogenic anion−π interactions to chromogenic anion-induced charge-transfer
(CT) and electron-transfer (ET) phenomena. In aprotic solvents, electron-donating abilities of anions
generally follow their Lewis basicity order, whereas π-acidity of NDIs can be fine-tuned by installing
different electron-rich and electron-deficient substituents. While strongly Lewis basic anions (OH− and F−) undergo thermal ET
−
with most NDIs, generating NDI•− radical anions and NDI2− dianions in aprotic solvents, weaker Lewis bases (AcO−, H2PO4 ,
Cl−, etc.) often require the photoexcitation of moderately π-acidic NDIs to generate the corresponding NDI•− radical anions via
photoinduced ET (PET). Poorly Lewis basic I− does not participate in thermal ET or PET with most NDIs (except with strongly
π-acidic core-substituted dicyano-NDI) but forms anion/NDI CT or anion−π complexes. We have looked for experimental
evidence that could indicate alternative mechanisms, such as a Meisenheimer complex or CH···anion hydrogen-bond formation,
but none was found to support these possibilities.
naphthalenediimides (NDIs) under thermal and light-gated
conditions that generated NDI•− radical anions and NDI2−
dianions.
1. INTRODUCTION
Conceived in theory1 and confirmed experimentally,2 anion−π
interactions1−3 have emerged as a new paradigm of anion-
recognition chemistry4 that, for a long time, relied primarily on
hydrogen-bonding interactions5 and Lewis acid (transition
metal ions, boron, etc.) coordination6 of anions. Mirroring
cation−π interaction7 (e.g., Na+·C6H6 complex) between a
cation and an electron-rich π-system with a negative quadru-
pole moment, non-covalent interactions between anions and π-
electron-deficient systems (π-acids) with positive quadrupole
moments are broadly defined as anion−π interactions.1−3
Several anion/π-acceptor complexes have come to light since
the turn of the millennium, yet most anion−π (anion−
quadrupole) interactions are so weak that they do not usually
perturb electronic properties of the π-acceptors, nor do they
generate any optical response.2 Only a handful π-acceptors8
display optical changes upon interacting with certain anions,
and the resulting UV/vis absorption spectra show Mulliken
dependence,8,9 i.e., a linear relationship between the donor/
acceptor (D/A) electronic transition energies (νCT) and their
redox potentials. These anion/π-acceptor interactions8 have
been aptly attributed to charge-transfer (CT) interactions that
essentially resemble π-donor/acceptor CT interactions.10
Expanding the scope and boundary of anion−π interactions,
we were the first to demonstrate11 unique formal electron
transfer (ET) from Lewis basic anions to π-acidic 1,4,5,8-
In the realm of donor/acceptor chemistry, [D·A] CT states
and [D•+·A•−] ET states belong to the same energy
continuum.12 The manifestation of CT or ET depends on
the relative energies of HOMO and LUMO levels of electron
donors and acceptors, respectively, and their spectroscopic
signals are quite distinctive. For instance, if the HOMO of an
electron donor is located well above the LUMO of an electron
acceptor (Figure 1a), the thermal ET pathway should be turned
ON (ΔG°ET < 0), generating paramagnetic D•+ and A•− radical
ions.11b,12 In contrast, if the HOMO of an electron donor is
located below the LUMO of an electron acceptor, thermal ET
should be turned OFF (ΔG°ET > 0, forbidden). However, if the
donor HOMO is located above the photogenerated SOMO−1
of the electron acceptor, photoinduced electron-transfer (PET)
pathway could be turned ON (Figure 1b), generating the same
A•− radical anion. Regardless of the electron source (electro-
chemical vs chemical reduction, thermal ET vs PET), a given
A•− radical anion always displays the same characteristic signals
(UV/vis and EPR). The rate and extent of A•− formation
depend on the ET driving force (ΔG°ET and ΔG°PET). When
Received: April 2, 2012
Published: June 11, 2012
© 2012 American Chemical Society
13679
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
Figure 1. Energy diagrams show how the relative positions of the HOMO and LUMO levels of NDIs with respect to the HOMO of anions channel
anion/NDI interactions through anion-induced (a) thermal ET (ΔG°ET < 0), (b) PET (ΔG°ET ≥ 0, but ΔG°PET < 0), and (c) CT interactions. (a)
Thermal ET takes place when the HOMO of anion is located above the LUMO of NDI. (b) PET takes place when the HOMO of the anion lies
below the LUMO of NDI but still above its HOMO. (c) CT interaction takes place when the anion is weakly Lewis basic and cannot trigger formal
ET to NDIs. Lewis basicity trend: X− > Y− ≫ Z−.
both thermal and photoinduced ET are turned OFF because of
a significantly low-lying donor HOMO level, neutral D/A CT
complexes could be formed via orbital mixing (Figure 1c).11b
CT complexes show broad UV/vis absorption bands, the
locations of which (λCT) depend on the donor and acceptor
strengths (Mulliken dependence).8,10,12
F− and a similar spectrum of a strongly π-acidic core-
substituted dicyano-NDI (DCNDI) in the presence of less
basic I−. While the sharp UV/vis spectrum resulting from I−/
DCNDI interaction was loosely attributed to CT or ET from I−
(although these two processes are fundamentally different,12
vide supra), a [F−NDI]− Meisenheimer complex formation was
invoked8d to rationalize similar spectroscopic changes induced
by F−. Electrospray ionization mass spectrometry (ESIMS)
revealed a wide range of anion/π-acceptor [X−/A] adducts,8d,11
which could be interpreted as one of the following possibilities:
(1) [X−/A] complex formation via CT, ET, or anion−π
interaction, (2) [X•/A•−] radical pair formation after an anion-
induced ET, (3) CH···X− H-bonded complex formation, or (4)
covalent [X−A]− Meisenheimer complex formation. Although
covalent Meisenheimer complex formation between electron-
deficient dinitrotoluene and Lewis bases has been long believed
Given the similarities between π-donor/acceptor and anion/
π-acceptor CT interactions and well-known examples of
ascorbate-mediated reductions of the Lewis acidic transition
metal ions, e.g., Fe(III) to Fe(II) in cytochrome-b redox
protein13 and Cu(II) to Cu(I) in “click chemistry”,14 the
paucity of anion-induced formal ET to neutral π-acceptors
seems quite surprising. Herein, we demonstrate that an anion
can interact with a neutral π-acceptor through either non-
chromogenic anion−π interactions or chromogenic CT and ET
interactions, if a delicate interplay between (1) the anion’s
Lewis basicity, i.e., electron-donating ability, (2) the π-
acceptor’s electron-accepting ability, and (3) the effects of
solvents and counterions on the Lewis basicity of the anions
satisfies the requirement of each mechanism. In addition to
these anion/π-acceptor interactions, two other modes of
contacts, namely, (i) CH···anion H-bonding15 and (ii) covalent
Meisenheimer or σ-complex15a,16 formation, have also been
debated within the realm of anion/π-acceptor interactions. In
CH···anion H-bonding, the anion interacts with an acidic
proton along the σ-frame rather than interacting with the
positive quadrupole of the π-acceptor, a phenomenon that
usually causes a downfield chemical shift of the H-bonded
proton.17 Nucleophilic attack of an anion on a π-acceptor could
form a covalent σ-bond, and the resulting [X−A]− species is
called a Meisenheimer complex.18 Although Meisenheimer
complexes often display vibrant colors, unlike EPR-active
paramagnetic A•− radical anions, these covalent [X−A]−
intermediates are usually diamagnetic species.
to be the cause of ensuing colorimetric changes, Buhlmann et.
̈
al.18b recently demonstrated that, instead of forming
Meisenheimer complex, OH− and amines actually deprotonate
an acidic proton from the Me group of DNT, generating an
electron-delocalized, colorful anionic species. Furthermore, ab
initio calculations19 suggest that modes of anion/π-acceptor
interactions vary drastically in the presence of different solvent
molecules. For instance, a covalent F−−triazine Meisenheimer
complex or a CH···Cl− H-bonded triazine·Cl− complex
constitutes the lowest-energy structure in the gas phase (no
or little solvent), whereas anion−π interaction becomes the
most stable and preferable mode in the presence of solvent
molecules (MeCN or H2O).19 Thus, insufficient studies and
incomplete explanations caused further confusion in this highly
debated field of anion−π interactions.
In this article, we present a comprehensive analysis of
different modes of anion/π-acceptor interactions through
structure−property relationship studies using powerful exper-
imental techniques, such as UV/vis, NMR, and EPR spectros-
copies, electrochemistry, spectroelectrochemistry, isothermal
titration calorimetry, and X-ray photoelectron spectroscopy
(XPS). Density functional theory (DFT) and natural bond
orbital calculations have also been used to identify the electron-
deficient areas of NDIs and how they interact with different
anions. These studies unequivocally show that, depending on
the π-acidity of NDIs, the Lewis basicity of anions, solvents, and
experimental conditions, modes of anion/NDI interactions vary
from anion-induced thermal and light-gated ET phenomena to
While we demonstrated11 formal ET from Lewis basic anions
(e.g., F−, AcO−, H2PO4 , and Cl−) to tunable π-acidic NDI
−
receptors under thermal and photoinduced conditions, Dunbar
et al.8c demonstrated that π-acidic HAT(CN)6 receptor forms
CT complexes with halides, and Matile et al.8d reported
formation of various anion/NDI complexes. In the latter two
cases, π-acceptors displayed stronger affinity and selectivity
toward more Lewis basic anions (e.g., Cl− > Br− > I−). In
addition to anion/NDI CT complexes, Matile et al.8d also
observed a highly featured, new UV/vis spectrum of
unsubstituted NDIs in the presence of strongly Lewis basic
13680
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
Chart 1. Molecular Structures of NDI Derivatives
weak CT and anion−π interactions and rule out σ-complex
drawing substituents enhance the π-acidity of NDIs, as they
display less negative first reduction potentials (E1Red) and lower
LUMO levels, whereas electron-rich substituents diminish π-
acidity, render the corresponding NDIs more difficult to
reduce, and elevate their LUMO levels (Table 1). While the N-
formation and deprotonation of NDIs.
2. RESULTS AND DISCUSSION
To understand how electronic and structural parameters of
anions and π-acceptors control the mechanisms of their
interactions, we chose π-electron-deficient, neutral NDI
compounds20 as a common platform (Chart 1). The π-acidity
of NDI receptors can be easily manipulated by installing
different electron-rich and -deficient substituents on two imide
N-centers11b as well as at the core naphthalene ring (cNDIs).21
We have previously demonstrated11 that N-substituents impart
modest impacts on NDI’s π-acidity, while others8d,21 have
demonstrated that electron-withdrawing core substituents
render the corresponding cNDIs strongly π-acidic.
2.1. Synthesis and Characterization. The N,N′-disub-
stituted NDI derivatives (e.g., N,N′-dipyridyl NDI (DPNDI),
NDIs 1−7) have been synthesized by standard bis-imidiza-
tion11 of commercially available naphthalene-1,4,5,8-tetracar-
boxydianhydride (NDA) with corresponding amines (Support-
ing Information, Scheme S1). To prepare dicyano-substituted
NDI (DCNDI), NDA was first brominated with dibromoiso-
cyanuric acid according to a literature procedure (Scheme
S2).21 The corresponding dibromo-NDI was then treated with
CuCN to afford DCNDI (Scheme S2).8d,21c All these
compounds have been purified by column chromatography
and characterized by NMR spectroscopies and high-resolution
ESIMS. 1H NMR signals of the central naphthalene protons of
symmetric N,N′-disubstituted NDI derivatives appear as a
singlet at δ ∼8.8 ppm, whereas the characteristic singlet peak of
strongly π-acidic DCNDI’s core protons appears farther
downfield at 9.07 ppm.
2.2. Electrochemistry. To quantify π-acidities of NDIs in
different solvents, cyclic voltammograms were recorded in
different polar (MeCN, DMF, and DMSO) and nonpolar (o-
dichlorobenzene (ODCB)) solvents. In both polar and
nonpolar solvents, the trend of NDI’s π-acidity remains the
same (Figure S1); however, polar solvents render the reduction
of NDIs slightly easier than nonpolar solvents, a phenomenon
that can be attributed to better stabilization of negatively
charged NDI•− and NDI2− species in polar media. HOMO and
LUMO energies of NDIs have been calculated from their redox
potentials (E1Redox) and electronic absorption spectra using the
following equations:21c
Table 1. NDIs’ Redox Potentials (vs Ag/AgCl in MeCN),
HOMO, and LUMO Energies
NDIs
E1Red/E2 (mV)
LUMO/HOMO (eV)
Red
DCNDI
NDI 1
NDI 2
NDI 3
DPNDI
NDI 4
NDI 5
NDI 6
NDI 7
+10/−500
−310/−790
−380/−830
−395/−885
−440/−890
−510/−900
−535/−965
−550/−975
−560/−980
−4.39/−7.49
−4.07/−7.17
−4.00/−7.10
−3.98/−7.08
−3.94/−7.04
−3.87/−6.97
−3.85/−6.95
−3.83/−6.93
−3.82/−6.92
substituents impart a modest impact on NDI’s π-acidity, which
is evident from more subtle changes in reduction potentials,
electron-withdrawing cyano- groups at the NDI core
dramatically enhance its π-acidity. The low-lying LUMO levels
of strongly π-acidic NDIs open up the possibility of ET from
several electron donors that possess higher HOMO levels,
rendering ET events less selective (Figure 1a). On the other
hand, moderately π-acidic NDIs possessing higher LUMO
levels are expected to undergo ET more selectively only from
the stronger Lewis basic anions (e.g., F−, OH−, CN−, etc.).
2.3. Spectroelectrochemistry. To characterize UV/vis
absorption properties of NDI•− and NDI2− species, we
conducted spectroelectrochemical analysis of NDIs using
−
nonbasic Bu4N+PF6 (TBAPF6) as a supporting electrolyte.
Constant potentials (Vap) commensurate with the first and
second reduction potentials were applied until NDI•− and
NDI2− spectra became saturated in stepwise fashion. Electro-
chemical oxidation of NDI•− and NDI2− regenerated neutral
NDIs as the original spectra reappeared, demonstrating
complete reversibility of redox processes.
Regardless of the substituents, all neutral NDIs (DPNDI,
DCNDI, and NDIs 1−7) are colorless compounds that absorb
in the UV region (340−400 nm), indicating that these
substituents do not significantly modify HOMO−LUMO
band gaps. However, NDI•− radical anion and NDI2− dianion
are intensely colorful species that display highly featured UV/
vis absorption spectra spanning across the visible−NIR region
(Figures 2a and S2).11,22 For instance, electrochemical
reduction of colorless DPNDI (λAbs = 343, 361, and 381 nm)
to an orange colored DPNDI•− at Vap = −450 mV (vs Ag/
AgCl) produces highly featured absorption spectrum that
displays prominent new peaks at 475 (λmax), 605, 711, and 791
nm and diminished intensities of the neutral DPNDI peaks in
EHOMO = [−4.8 eV − (E1Ox − 0.42)(V)] eV
(1)
ELUMO = [−4.8 eV − (E1 − 0.42)(V)] eV
(2)
(3)
Red
ΔEHOMO−LUMO = 1240/λonset
when the redox potentials are measured against a Ag/AgCl
reference electrode. CV data show that the electron-with-
13681
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
Figure 2. Spectroelectrochemical analysis of (a) DPNDI (in 0.1 M TBAPF6/DMF) and (b) DCNDI (in 0.1 M TBAPF6/MeCN). UV/vis spectra of
(a) neutral DPNDI (Vap = 0 mV, black trace), DPNDI•− (Vap = −450 mV, orange trace), and DPNDI2− (Vap = −900 mV, pink trace), and (b)
neutral DCNDI (Vap = +400 mV, black trace), DCNDI•− (Vap = −50 mV, orange trace), and DCNDI2− (Vap = −600 mV, pink trace).
the UV-region, establishing a clear isosbestic point at 394 nm
(Figure 2a). Further reduction of DPNDI•− to a pink colored
DPNDI2− dianion at Vap = −900 mV (vs Ag/AgCl) made
DPNDI•− signals disappear and a prominent new peak appear
at 542 nm. All NDI•− radical anions constitute a doublet state
(D0) and their longer wavelength absorption peaks correspond
to D0→D1 electronic transitions.22 Absorption features of all
NDI•− and NDI2− species obtained from different N,N′-
disubstituted NDIs are essentially the same (Figure S2), with
slight differences in peak positions (ca. 5 nm), indicating that
electrons are delocalized only within the NDI core and not
through the N-substituents.
1)with Bu4N+ (TBA) and Et4N+ (TEA) salts of various
anions with decreasing Lewis basicity: OH− > F− > CN−
>
AcO− > H2PO4− > Cl− > Br− > I− ≫ PF6 .23 By adjusting the
Lewis basicity of anions and π-acidity of NDIs, anion-induced
ET to NDIs can be channeled through two distinct pathways
(Figure 1a).11b (1) When anion and NDI are strong electron
donors and acceptors, respectively, the HOMO of the anion is
located above the LUMO of NDI, a situation that turns ON a
thermal anion-to-NDI ET. (2) When the HOMO of a less
Lewis basic anion falls below the LUMO of an NDI but still
rests above its HOMO, the thermal ET is turned OFF, but light
can activate a PET pathway from the anion’s HOMO to the
photogenerated 1*NDI’s SOMO−1. The final electronic
configurations of NDI•− and NDI2− species generated by
direct electrochemical reduction or anion-induced ET are the
same. Therefore, irrespective of the electron source, the
reduced NDI species display identical spectroscopic signals.
When anions are poor electron donors but highly polarizable
(e.g., I− and Br−), anion-induced ET and PET to NDIs are
turned OFF, but CT and anion−π interactions could still take
place.
−
Using spectroelectrochemical analysis, for the first time, we
were able to identify the spectroscopic signatures of DCNDI•−
radical anion and DCNDI2− dianion upon stepwise electro-
chemical reduction of DCNDI (Figure 2b). Since electron-
withdrawing cyano substituents are in conjugation with the
central π-system, the DCNDI•− and DCNDI2− spectra are
significantly different from those displayed by reduced N,N′-
disubstituted NDIs. While neutral DCNDI absorbs light at 342,
358 (λmax), 379, and 394 nm, the electrochemically generated
DCNDI•− radical anion (Vap = −50 mV) displays significantly
diminished intensity of these signals and prominent new peaks
at 401, 454, 487 (λmax), 530, 605, 686, and 764 nm, establishing
a clear isosbestic point at 400 nm. The second reduction,
leading to DCNDI2− dianion formation (Vap = −600 mV),
diminishes the intensities of 487, 530, 686, and 764 nm peaks
of DCNDI•− radical anion and generates new peaks at 427, 454
(λmax), and 532 nm, establishing clear isosbestic points at 400,
465, 540, and 615 nm (Figure 2b). Electrochemical oxidation of
DCNDI2− dianion to neutral DCNDI (Vap = +100 mV) takes
place via an intermediate DCNDI•− formation, as DCNDI•−
signal emerges momentarily at the expense of DCNDI2−
spectrum before disappearing to regenerate the original
DCNDI spectrum. This new information is critical for an
accurate understanding of anion-induced spectroscopic changes
in NDIs and cNDIs, lack of which misled researchers8d to
prematurely attribute F−-induced spectroscopic changes of
NDIs to a covalent [F−NDI]− Meisenheimer complex
formation. In the following sections, we will show that
electrochemically generated NDI•− radical anion and NDI2−
dianion spectra, including the isosbestic points, are identical to
those generated by Lewis basic anions.
Thermal Anion-Induced Electron Transfer to N,N′-Dis-
ubstituted NDIs. UV/vis titration of a moderately π-acidic
DPNDI (E1
= −450 mV vs Ag/AgCl) in DMSO with
Red
strongly basic anions, e.g., OH− (pKa of H2O in DMSO = 32),
F− (pKa of HF in DMSO = 15), and CN− (pKa of HCN in
DMSO = 13) gradually bleached the original DPNDI
absorption peaks at 343, 361, and 381 nm and concurrently
produced prominent new peaks at 475 (λmax), 605, 711, and
791 nm, establishing a clear isosbestic point at 394 nm (Figures
3 and S3) as the solution turned orange. The entire spectrum,
including the isosbestic point (394 nm) of F−-generated orange
solutions of DPNDI, is essentially identical to that of the
electrochemically generated DPNDI•− radical anion produced
in the absence of these anions (Figures 2a and 3b, orange
traces), suggesting that DPNDI•− radical anion is formed via an
ET from strongly Lewis basic anions.11a Had the anion reacted
with DPNDI as a nucleophile, forming a covalent [anion−
DPNDI]− Meisenheimer complex en route DPNDI•− for-
mation, the isosbestic point (394 nm) during anion titrations
should have been different from that observed during
electrochemical generation of DPNDI•− radical anion, as
these two pathways would have been completely different.
Addition of higher equivalents of F− (also OH− and CN−,
Figure S3) turned the orange-colored DPNDI•− solution to
pink that displayed gradual disappearance of the characteristic
DPNDI•− signals with a concomitant emergence of a new peak
at 542 nm (Figure 3b, pink trace). UV/vis spectra of anion-
2.4. UV/Vis Studies of Anion/NDI Interactions. To
understand how the interplay between the π-acidity of NDIs
and Lewis basicity of anions regulates different modes of anion/
NDI interactions, we surveyed the interactions of a library of
NDI derivativesDPNDI, DCNDI, and NDIs 1−7 (Chart
13682
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
Figure 3. F−-induced (a) colorimetric and (b,c) UV/vis spectroscopic changes of DPNDI in DMSO from colorless (no F−) to orange (≤5 equiv of
F−) to pink (5−30 equiv of F−) and back to colorless upon NOBF4 treatment of colored solutions. (b) UV/vis titration of DPNDI with F−
(TBAF·3H2O). Black trace, neutral DPNDI (no F−); orange trace, DPNDI•− (5 equiv of F−); and pink trace, DPNDI2− (30 equiv of F−). (c)
Regeneration of neutral DPNDI spectrum upon oxidation of DPNDI•− with NOBF4.
generated pink solutions of DPNDI closely resemble an
electrochemically generated DPNDI2− dianion spectrum
(Figures 2a and 3b), suggesting the formation of DPNDI2−
in the presence of excess amounts of strongly Lewis basic
anions. ESIMS and ITC studies (vide infra) further
demonstrate the formation of a 1:1 F−·DPNDI complex (m/
z = 439.1). Although ESIMS data could be interpreted as a
[F−·DPNDI] precursor complex formation prior to the first ET
or a [F•·DPNDI•−] radical-pair complex formation as a result
of ET, UV/vis and EPR (vide infra) studies do not display any
electronic perturbation of DPNDI•− with F− or F•. Anion-
induced formation of DPNDI2− dianion was also observed
through ESIMS (m/z = 210.9). Due to electrostatic repulsions,
it is extremely unlikely that the reduction of DPNDI•− to
DPNDI2− with a second equivalent of strongly Lewis basic
anions would proceed through complexation of an anion with
DPNDI•−, leaving outer-sphere ET as the only possible
mechanism of the second reduction.
Akin to completely reversible electrochemical reductions of
DPNDI (DPNDI ⇆ DPNDI•− ⇆ DPNDI2−), the anion-
generated DPNDI•− and DPNDI2− species can be fully
oxidized back to neutral DPNDI using an oxidizing agent,
NOBF4, that regenerates the original DPNDI spectrum (Figure
3c). A neutral reducing agent, NH2NH2, also generates
DPNDI•− radical anion but cannot form DPNDI2− (Figure
S3), possibly because of its weaker reducing ability than
strongly Lewis basic anions. Thus, strong similarities between
chemically generated, electrochemically generated, and anion-
generated DPNDI•− and DPNDI2− spectra corroborate that all
these ET pathways are equivalent and that no covalent
Meisenheimer intermediate is involved. Weaker Lewis basic
and DPNDI2−, a phenomenon that can be attributed to high
solvation of small F− ion via H-bonding interactions
(−ΔHHydration of F− = 115 kcal/mol),24 which dramatically
diminishes its Lewis basicity and electron-donating ability.
Thermal Anion-Induced Electron Transfer to DCNDI.
Because of strong π-acidity of DCNDI, electron-rich neutral
solvents, such as DMSO, DMF, and DMAc, undergo thermal
ET, generating DCNDI•− spectra (Figure S5), rendering these
solvents unsuitable for the investigation of anion/DCNDI
interactions. To avoid solvent-mediated ET and PET, freshly
prepared MeCN solutions of DCNDI were titrated with anions
in the absence of light.
As in the case of spectroelectrochemical analysis, UV/vis
titrations of DCNDI with strongly Lewis basic F− and OH−
ions initially diminish the neutral DCNDI peaks at 342, 358
(λmax), 379, and 394 nm and generate prominent new peaks at
401, 454, 487 (λmax), 530, 605, 686, and 764 nm, establishing a
clear isosbestic point at 400 nm (Figure 4a,b). Addition of
more than 1 equiv of these anions diminishes the intensities of
DCNDI•− radical anion peaks at 487, 530, 686, and 764 nm
and generates new peaks at 427, 454 (λmax), and 532 nm,
establishing new isosbestic points at 400, 465, 540, and 615 nm.
The second step corresponds to DCNDI2− dianion formation.
While strongly Lewis basic anions (F− and OH−) generate up
to DCNDI2− dianion in two steps, weaker Lewis basic anions,
e.g., AcO−, Cl−, Br−, and I−, that do not reduce moderately π-
acidic DPNDI at all, can only produce DCNDI•− radical anion
but do not generate DCNDI2− dianion via thermal ET (Figures
4c and S5). NOBF4 oxidizes anion-generated DCNDI•− and
DCNDI2− species back to neutral DCNDI, as the original
spectrum reappears (Figure S5).
anions, e.g., AcO−, H2PO4 , Cl−, Br−, I−, and PF6 , do not
reduce DPNDI at all, as the spectra of DPNDI in the presence
of large excesses of these anions remain unchanged (Figure S4).
The fact that, in aprotic solvents strongly Lewis basic anions
(OH−, F−, and CN−) generate DPNDI•− and DPNDI2−, but
−
−
If F− and OH− ions indeed formed covalent Meisenheimer
complexes with NDIs, cleavage of these σ-bonds under
oxidative (NOBF4) conditions to regenerate neutral NDI
would be an unlikely event. The similarities between reversible
spectroscopic changes under electrochemical conditions and
with anions corroborate that anion-induced thermal ET events
are indeed responsible for DCNDI•− and DCNDI2− formation
and no Meisenheimer complex is involved.
less basic anions (AcO−, H2PO4 , ClO4 , Cl−, Br−, I−, and
PF6 ) do not, clearly shows that the former anions have much
−
−
−
better electron-donating ability than the latter ones. In protic
solvents (H2O and MeOH), F− no longer generates DPNDI•−
13683
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
Figure 4. UV/vis spectroscopic changes of DCNDI in MeCN (black trace) upon two-step reduction to DCNDI•− (orange trace) and DCNDI2−
(pink trace) with (a) OH− and (b) F−, and (c) one-step reduction to DCNDI•− with 100 equiv of I−.
Figure 5. Anion-generated UV/vis spectra of NDI•− radical anions of NDIs 1 (violet), 2 (indigo), 3 (cyan), 4 (green), 5 (gray), 6 (red), and 7
(orange) (each 30 μM in ODCB) in the presence of (a) F− (10 equiv), (b) AcO− (50 equiv), (c) H2PO4− (50 equiv), and (d) Cl− (100 equiv). The
spectra show that while (a) F− and (b) AcO− generate NDI•− radical anions of 1−7, (c) H2PO4− does so only for NDIs 1 and 2, and (d) Cl− does so
only for NDI 1. The extent of NDI•− formation diminishes with decreasing π-acidity of NDIs and Lewis basicity of anions.
Regulating Anion-Induced Thermal ET. To examine
whether anion-induced thermal ET to NDIs can be turned
ON/OFF by adjusting the π-acidity (LUMO level) of NDIs
with respect to Lewis basicity (HOMO level) of anions, we
surveyed the interactions between NDIs 1−7 with gradually
isosbestic points are essentially identical to those of electro-
chemically generated NDI•− radical anions in the absence of
any electron-donating anions, confirming that anion-to-NDI
ET events are indeed responsible for these spectroscopic
changes and ruling out a Meisenheimer complex formation. In
summary, the extent of anion-induced NDI•− formation is
greater when the energy gap (ΔG°ET) between the anion’s
HOMO and the NDI’s LUMO is larger. ΔG°ET decreases as
anions and NDIs become weaker electron donors and
acceptors, respectively, diminishing the NDI•− radical anion
formation via thermal ET.
−
decreasing π-acidity and F−, AcO−, H2PO4 , and Cl− ions with
gradually decreasing Lewis basicity.11b UV/vis titrations of
NDIs 1−7 in ODCB with strongly Lewis basic F− generate
prominent NDI•− spectra (Figure 5a), however, as the π-acidity
of NDIs gradually diminishes, the extent of NDI•− formation
decreases. Similarly, as the Lewis basicity of anions decreases
8c,11b,23
(F− > AcO− > H2PO4 > Cl−),
their electron-donating
Photoinduced Electron Transfer from Anions to N,N′-
Disubstituted NDIs. Next, we turned our attention to explore
whether a PET from the HOMO of an anion to the SOMO−1
−
abilities and HOMO levels diminish,8c first weakening and
eventually turning OFF thermal ET to NDIs (Figure 1a). As a
result, less basic AcO− generates weaker NDI•− signals from
NDIs 1−7 (Figure 5b) than those produced by F−, while
H2PO4− generates very weak NDI•− signals only with NDI 1−4
(Figure 5c), and Cl− triggers extremely weak thermal ET only
to NDI 1 in ODCB (Figure 5d). Regardless of the N-
substituents, anion-generated NDI•− absorption spectra and
1
of the *NDI excited state (Figure 1b) could generate NDI•−
radical anion when thermal anion-to-NDI ET is turned OFF.11b
Solvents have an important effect on anion’s Lewis basicity.
Protic solvents (H2O, MeOH) and aprotic solvents containing
acidic protons (MeCN) can better solvate anions through H-
bond formation, which diminishes anions’ Lewis basicity and
13684
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
Figure 6. (a−c) UV/vis spectra of NDI 1 in MeCN in the absence of any anion (black traces) and in the presence of large excess (≥50 equiv) of (a)
−
AcO−, (b) H2PO4 , and (c) Cl−, showing negligible 1•− formation in the dark (red traces). Irradiation of these solution mixtures (a−c) with a W-
1
lamp significantly enhances the 1•− signals (blue traces) upon anion-mediated photoreduction of *NDI 1. (d) At high concentrations (0.5 mM),
NDI 1 displays CT absorption band with excess I− (100 equiv) but does not form any 1•−
.
electron-donating ability.8c,11b Therefore, while AcO−, H2PO4 ,
and Cl− anions could trigger thermal ET to several NDIs in
ODCB (Figure 5), many of these thermal ET pathways are
turned OFF in MeCN (Figure 6), which opened up the
possibility of PET. To test this hypothesis, we irradiated the
NDI solutions in the presence of anions that showed negligible
or no NDI•− formation in the dark (no thermal ET) with a W-
lamp and periodically recorded UV/vis spectra.
acidic N,N′-disubstituted NDIs via ET or PET. However, large
excess of I− (ca. 100 equiv) produces a characteristic CT band
(λCT = 505 nm) with a concentrated solution of NDI 1 (the
most π-acidic core-unsubstituted NDI in the series apart from
DCNDI) in both ODCB and MeCN, but does not generate
any 1•− radical anion (Figure 6d). These observations further
corroborate that, in aprotic solvents, I− is indeed a weaker
electron donor than strongly Lewis basic OH− and F− that
form NDI•− and NDI2−. Thus, by adjusting the NDI’s π-acidity
and anion’s Lewis basicity, the entire spectrum of ET, PET, and
CT phenomena can be accessed (Figure 1).
−
Irradiation of a MeCN solution of NDI 1 in the presence of
AcO−, H2PO4 , and Cl−, which originally showed extremely
−
weak or no thermal ET interaction, immediately turns ON the
anion-to-1*NDI PET, significantly enhancing the intensity of
NDI 1•− absorption spectra that reached saturation within 10
min of irradiation (Figure 6a−c).11b Similarly, NDIs 2−4
2.5. NMR Titrations of NDIs with Anions. NMR
experiments were conducted to determine the effects of
anion/NDI interactions on their structural and electronic
undergo PET from AcO− and H2PO4 in MeCN (Figure S6).
properties. H NMR experiments can distinguish between the
−
1
Anion-induced PET to NDI is also observed in ODCB when a
thermal ET is negligible or turned OFF. For instance, in
ODCB, NDIs 6 and 7 undergo photoreduction with AcO− and
formation of either paramagnetic NDI•− radical anions (NDI
signals should disappear), covalent Meisenheimer complex
(signals of originally symmetric NDIs should split because of
the ensuing loss of symmetry), or CH···anion H-bonds (NDI
signals should shift downfield). 19F NMR experiments were
introduced to gain insights into the nature of F−/NDI
interactions. If F− reduces NDI to NDI•− and NDI2−, it should
be oxidized into a paramagnetic F• and the F− signal should
disappear. However, if a C−F σ-bond is formed in a
diamagnetic [F−NDI]− Meisenheimer complex, it should
display new F-signals. In the case of CH···F− H-bonds, F−
signal should shift downfield. We looked for all this evidence to
determine which mechanism actually takes place and to
eliminate the others.
NDIs 1−5 with H2PO4 , generating characteristic NDI•−
−
radical anion signals (Figure S7). When a facile thermal
anion-to-NDI ET process generates strong NDI•− signals by
directly populating the NDI’s LUMO in the ground state,
irradiation of these samples does not enhance the NDI•− signal
intensity, indicating that anion-to-1*NDI PET is turned OFF.
On the other hand, when the thermal ET interactions are
extremely weak or absent, π→π* transition in NDIs opens the
door for an energetically favored PET (ΔG°PET < 0) from the
anion’s HOMO to the 1*NDI’s SOMO−1 level (Figure 1a). In
control experiments, irradiation of NDIs in the absence of
anions did not produce any NDI•− radical anion (Figure S8),
ruling out the possibilities of intramolecular or solvent-
mediated PET.
1H NMR Titrations of N,N′-Disubstituted NDIs with Anions.
The 1H NMR spectrum of DPNDI (DMSO-d6) shows a singlet
at δ = 8.75 ppm corresponding to four identical NDI core
protons (Ha) and two doublets at 7.58 and 8.81 ppm
corresponding to Hb and Hc of two pyridine rings, respectively
(Figure 7).11a During the titration with F−, all DPNDI signals
Anion/NDI Charge-Transfer Interactions. While very weak
Lewis basic I− can reduce extremely strong π-acidic DCNDI to
DCNDI•−, it does not produce any NDI•− from weakly π-
13685
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
Figure 7. 1H NMR titrations of DPNDI (DMSO-d6, 298 K) with (a) F− shows DPNDI•− formation via thermal ET, as the Ha signal disappears and
(b) excess I− shows no DPNDI•− formation. 1H NMR spectra of NDI 1 (CD3CN, 298 K) with (c) 1 equiv of F− (trace ii) shows 1•− formation via
thermal ET, as indicated by the disappearance of the HA signal, and (d) excess Cl− shows no 1•− formation through thermal ET in the dark (trace ii),
but 1•− formation via PET (trace iii) upon irradiation with a W-lamp. In both cases, NOBF4 oxidation of 1•− regenerated neutral NDI 1 [trace (iii)
in (c) and trace (iv) in (d)].
became broad but none shifted at all, ruling out the possibility
of a CH···F− H-bond formation. At 1 equiv of F−, Ha signal
disappeared completely (Figure 7a), indicating the formation of
a paramagnetic DPNDI•− radical anion. EPR spectrum of this
species confirmed the formation of DPNDI•− radical anion
(vide infra). The fact that the Ha signal does not split during the
titration as a result of F−/DPNDI interactions rules out the
possibility of a covalent C−F bond formation. Had a
diamagnetic, covalent [F−DPNDI]− Meisenheimer complex
formed, the symmetry of DPNDI would have been destroyed,
which should have caused splitting of the Ha signal. Oxidation
of DPNDI•− with NOBF4 completely regenerated the original
DPNDI spectrum, as the Ha signal returned to full glory
NDI’s HA signal does not split or shift from its original position
before disappearing rules out a nonsymmetric covalent anion−
NDI intermediate formation or a CH···anion interaction
involving HA protons. Anion-generated NDI•− radical anions
are stable in dark, inert conditions, and they can be oxidized
back to neutral NDI with NOBF4, which brings the
characteristic NDI signals back to full glory (Figures 7).
Oxidative regeneration of NDIs confirms NDI•− formation via
ET in the first place and rules out a covalent Meisenheimer
complex formation.
1H NMR Titrations of DCNDI with Anions. To verify
whether both F− and I− ions trigger ET to strongly π-acidic
DCNDI generating DCNDI•− in the same way or their
interactions occur via different mechanisms, as predicted
−
(Figure 7a). Titrations of DPNDI with AcO−, H2PO4 , Cl−,
earlier,8d we conducted H NMR titrations in CD3CN. If I−
1
Br−, and I− ions did not cause any NMR spectroscopic change,
even at high concentrations of anions (Figures 7b), confirming
that these anions do not participate in ET with DPNDI.
Anion-induced thermal ET and PET to NDIs can also be
generated a paramagnetic DCNDI•− radical anion via an ET,
the NMR signals of DCNDI should disappear. On the other
hand, if F− formed a diamagnetic [F−DCNDI]− Meisenheimer
complex via nucleophilic attack, the NMR signal of DCNDI
core protons should split, since the symmetry of DCNDI would
be lost as a result of a covalent C−F bond formation.
11b
1
distinguished by H NMR experiments. 1H NMR titrations
of NDI 1 in CD3CN demonstrate that while F−, AcO−, and
−
H2PO4 generate paramagnetic 1•− through thermal ET,
leading to the disappearance of its core HA signal even in the
dark (Figure 7c and S9), weaker Lewis basic Cl− cannot do so
in the absence of light (Figure 7d). Irradiation of NDI 1 in the
presence of Cl− quickly generates paramagnetic 1•− as its HA
signal disappears. Poor Lewis bases Br− and I− do not produce
any NDI•− radical anion via thermal ET or PET (Figure S9).
The fact that, in the presence of electron-donating anions,
In CD3CN, the aromatic region of symmetrical DCNDI
displays a singlet at 9.07 ppm that corresponds to two core
NDI protons (Ha), as well as a triplet centered on 7.38 ppm
(Hc) and a doublet (Hb) at 7.30 ppm that correspond to the
2,6-dimethylphenyl substituents on imide N-centers. The Ha
signal of DCNDI became broad and ultimately disappeared in
the presence of 1 equiv of F− or I− anion without displaying any
13686
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
1
Figure 8. H NMR spectra of DCNDI (CD3CN, 298 K) before and after addition of (a) 1 equiv of F− and (b) 1 equiv of I−. Both generated
paramagnetic DCNDI•− radical anion as the Ha signal disappeared. NOBF4 oxidized DCNDI•− back to neutral DCNDI, regenerating the Ha signal.
splitting or shifting (Figure 8). The Hb and Hc signals also
became broader but did not shift, split, or disappear. The anion-
induced disappearance of Ha signal clearly indicates a
paramagnetic DCNDI•− radical anion formation with both F−
and I−. The facts that both anions led to identical spectroscopic
changes and the core DCNDI signal did not split with F− ruled
out a Meisenheimer complex formation. In both cases, all NMR
signals returned to full glory after oxidation with NOBF4
(Figure 8), demonstrating the reversibility of the entire process,
which is consistent with an ET event. In the absence of core
cyano substituents, less π-acidic NDIs do not generate NDI•−
with weakly basic I− (Figure S10).
support this scenario. The HF2− signal (−142.5 ppm) remained
unaffected, and its intensity did not increase upon the addition
of more and more DPNDI, ruling out a proton abstraction
from the DPNDI, a process that should have generated more
HF2 during the titration. Highly reactive F• can rapidly react
−
with silica, solvent, or counterions, making its capture into a
well-defined product extremely difficult. Nevertheless, 19F
NMR spectroscopy shows F• formation at the expense of F−,
and oxidative regeneration of neutral NDIs with NOBF4
suggests that F• or F− does not covalently bind with NDIs at
any stage.
Fate of Anions after ET. After ET, anion-generated X•
radicals act as sacrificial agents that could react with solvents,
counterions, or, in case of F•, glassware (silica), forming strong
Si−F bonds. The sacrificial nature of X• radicals prevents
NDI•−-to-X• back-ET, allowing NDI•− radical anions to persist
in solutions for an extended time. Although radicals generated
from inorganic anions (halides, OH−) could not be trapped so
far, the F• formation at the expense of F− has been observed
through 19F NMR spectroscopy. Since NDIs can be reduced to
NDI•− and NDI2− by certain anions and NOBF4 can oxidize
them back to neutral NDIs, we examined whether NOBF4 can
directly oxidize OH−, F−, Cl−, Br−, and I− anions (Figure S11).
NOBF4 treatment of these anions (as TBA+ salts) emanated
gases with characteristic colors (NO2, brown; F2, yellow; and
Cl2, greenish) and, in the case of Br− and I−, samples turned red
and brown, respectively, (Br2 and I2).26 Anions pretreated with
NOBF4 are already oxidized and can no longer reduce NDIs.
2.6. EPR Spectroscopy. EPR spectroscopy unequivocally
confirms a paramagnetic NDI•− radical anion formation via ET
from strongly Lewis basic anions. While neutral DPNDI is
diamagnetic and does not display any EPR signal, a 1:1 mixture
of DPNDI and F− in DMF displays characteristic EPR signals
of delocalized DPNDI•− radical anion (g = 2.0030)11,22b with
hyperfine splitting patterns that match perfectly with the
simulated EPR spectra of DPNDI•− radical anion (Figure 10,
orange trace). Excellent agreement between F−-generated and
simulated DPNDI•− EPR spectra indicates that F− or F• does
not perturb the electronic and magnetic properties of
DPNDI•− radical anion. If a covalent [F−DPNDI]−
Meisenheimer complex or a CH···F− H-bonded complex
were formed, it would have been a diamagnetic species and,
therefore, should not have displayed any EPR signal.
19F NMR Studies. 19F NMR spectrum of TBAF·3H2O in
DMSO-d6 shows (Figure 9) a singlet peak at −102 ppm
corresponding to the F− ion and a doublet at −142.5 ppm that
11a,25
−
corresponds to HF2 .
Titration of TBAF with DPNDI
initially caused broadening and gradual upfield shift of the F−
signal (Figure 9), which disappeared completely in the presence
of 1 equiv of DPNDI. While the upfield shift of F− signal could
be attributed to shielding by DPNDI π-system, its disappear-
ance at 1:1 TBAF/DPNDI clearly indicates a F• formation as a
result of F−-induced ET to DPNDI that generates DPNDI•−
simultaneously. Although a C−F bond formation was
considered as a possible alternative, no new F-signal
corresponding to a new covalent C−F bond was found to
Figure 9. 19F NMR titration (DMSO-d6, 298 K) of TBAF·3H2O with
DPNDI, showing the disappearance of the F− signal with 1 equiv of
DPNDI, indicating a F• formation.
13687
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
greater F− affinities than weaker π-acidic NDIs 4−7 that could
potentially form CH···anion H-bonds involving 2,6-protons on
their N-substituents. These results suggest that the affinity of
anion/NDI interactions primarily depends on electron donor
and acceptor strengths of anions and NDIs, respectively.
2.8. ESIMS Analysis. ESIMS provides physical evidence of
anion/NDI binding, although it cannot differentiate between a
noncovalent contact (CT, ET, or anion−π) and a covalent
attachment, nor can it confirm whether a [X−·NDI] precursor
complex formation precedes a CT or ET event or a [X•·NDI•−]
radical pair is formed as a result of the ET. ESIMS reveals
(Figure S13) m/z signals of 1:1 complexes 1·F− (m/z 1030.9),
1·AcO− (1070.7), 1·H2PO4 (1108.7), 2·F− (510.4), 2·AcO−
−
(549.1), 2·H2PO4− (587.0), 5·F− (437.1), and DPNDI·F− (m/z
439.1), all of which led to the corresponding NDI•− radical
anion formation via ET. ESIMS also shows several 1:1
complexes of DPNDI (420.1) with weaker Lewis basic anions:
Figure 10. EPR spectrum (DMF, 298 K) of 1:1 DPNDI/F− solution
(orange trace) displays hyperfine splitting identical to a simulated
DPNDI•− spectrum (blue trace), confirming a paramagnetic DPNDI•−
formation (g = 2.0030) from diamagnetic DPNDI (black trace).
Microwave frequency, 9.3902 GHz; power, 1 mW; and modulation
amplitude, 1 G.
Cl− (455.1), Br− (501.4), I− (547.4), NO2 (466.1), NO3
−
−
(482.4), AcO− (479.1), PhCO2 (541.5), H2PO4 (517.4),
−
−
HSO4 (517.4), ClO4 (519.4), and TfO− (570.8), none of
which triggered any CT or ET interactions, indicating that weak
anion−π (anion−quadrupole) interactions are sufficient for the
existence of these weak complexes. However, because of the
lack of any measurable changes (vide supra), these weak
interactions could not be quantified in solutions using UV/vis,
NMR, and ITC experiments. ESIMS also shows DPNDI2− (m/
z 210.4) formation in the presence of excess F−, but due to
electrostatic repulsion, it no longer binds any F− ion.
−
−
2.7. ITC Studies. ITC experiments were conducted to
determine the stoichiometry, binding constants (Ka), and
thermodynamic parameters (ΔH, TΔS, and ΔG) of anion/NDI
interactions (Figure S12). ITC studies show exothermic 1:1
interactions between F− and NDIs 1−5 with gradually
decreasing Ka values (Table 2) as π-acidity of NDIs diminishes.
Similarly, the anion affinity of NDI 1 diminishes with
2.9. XPS Analysis. XPS was introduced to further probe the
nature (noncovalent vs covalent) of F−/NDI interactions by
measuring the binding energy of F1s electrons in a 1:1 F−/
DPNDI mixture and comparing it with F-containing ionic and
covalent compounds. F1s electron binding energy is smaller for
an ionic F− salt than that of covalently bound fluorine. For
example, F1s binding energies in ionic CsF and NaF salts27 and
a fluorinated compound NDI 2 (containing covalent C−F
bonds) are 682.2, 683.5, and 687.1 eV, respectively. F1s binding
energy measured for a 1:1 F−/DPNDI mixture is 682.9 eV
(Figure 11), which is in excellent agreement with that of ionic
F− compounds, indicating that no covalent C−F bond has been
formed.
2.10. Computational and Structural Analysis. Electro-
static potential (ESP) maps of NDIs show that (Figure S14)
the imide rings are the most electron-deficient areas. Consistent
with the electron density map, B3LYP/6-31+G** energy
minimization of anion·NDI complexes suggests that anions
should preferentially interact with the electron-deficient imide
rings (Figure S15). X-ray crystallographic analysis of a [Pd(II)−
−
decreasing Lewis basicity of anions: F− > AcO− > H2PO4 ,
as they show Ka’s of 1230, 70, and 40 M−1, respectively
Table 2. Stoichiometry, Ka, and ΔG of Interactions between
NDIs 1−7 and F− in ODCB (298 K) Derived from ITC
Analysis
NDI
NDI/F− ratio
Ka (M−1
)
ΔG (kcal/mol)
NDI 1
NDI 2
NDI 3
NDI 4
NDI 5
1:1
1:1
1:1
1:1
1:1
1230
1080
923
−4.24
−4.15
−4.07
−3.91
−3.74
708
528
(ODCB, 298 K). Because of the presence of 2,6-iodo and
fluoro groups at N-aryl rings, NDIs 1, 2, and 3 cannot form
CH···anion H-bonds via the N-aryl substituents. Corroborating
NMR and UV/vis data, stronger π-acidic NDIs 1−3 display
Figure 11. Comparison between F1s XPS peak profiles of (a) 1:1 F−/DPNDI mixture and (b) NDI 2 containing covalent C−F bonds indicates no
C−F bond formation in (a).
13688
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
DPNDI]n zigzag coordination polymer28 reveals that oxygen
lone-pair electrons of TfO− counterions and THF molecules
(solvent of crystallization) indeed interact with the electron-
deficient imide rings of DPNDI ligands, albeit without
participating in ET events. Calculated energy of F−/DPNDI
interaction in the gas phase is ca. 45 kcal/mol, which is
comparable to a F−H···F− H-bond (ca. 40 kcal/mol),11a but
significantly weaker than a C−F covalent bond (ca. 115 kcal/
mol). Furthermore, the shortest distance between a F− ion and
the carbonyl C-atom of DPNDI is 1.65 Å (Figure S15a), which
is considerably longer than a typical C−F covalent bond (ca.
1.30 Å).
act as a better electron donor than strongly Lewis basic anions.
If this zwitterion, its byproduct, or a carbanion generated via
deprotonation of solvent (although unlikely) served as the
common electron source instead of the original anions (X−),
then irrespective of anions, all TBAX salts should have
produced the same species, i.e., either NDI•− radical anion or
NDI2− dianion from all NDIs. Undermining this proposition, in
aprotic solvents, TBA+ and TEA+ salts of stronger Lewis basic
anions, e.g., F− and OH−, generate NDI•− and NDI2− in two
−
steps, while weaker basic anions, e.g., AcO−, H2PO4 , and Cl−,
generate only NDI•− radical anion from stronger π-acidic NDIs,
and I− can only produce DCNDI•− radical anion from the most
π-acidic DCNDI. Finally, it was suggested that F− could
deprotonate H2O of TBAF·3H2O to generate OH−, which
would then be responsible for the reduction of NDIs.
According to the principles of acid−base chemistry, deproto-
nation of H2O with F− (F− + H2O ⇆ OH− + HF) is unlikely,
since the equilibrium is favored toward the weaker conjugate
acid, H2O (pKa in DMSO = 32), over the stronger conjugate
acid HF (pKa in DMSO = 15).23 Some OH− could exist in the
equilibrium, and TBAOH itself is known to produce NDI•−
and NDI2− species. However, the disappearance of F− NMR
signal in the presence of NDIs indicates F• formation due to
ET, and ITC and ESIMS data show a 1:1 F−·NDI complex
formation, providing a clear indication that it is the F− ion of
TBAF, not OH−, that reduces NDIs. Furthermore, NOBF4-
mediated oxidation of OH−, F−, Cl−, Br−, and I− ions in solid
phase and in MeCN demonstrates their electron-donating
properties.26 Thus, all experimental results corroborate that the
electronegativity of elements does not accurately reflect the
electron-donating ability of the corresponding anions in aprotic
solvents. Effective size, electron-density, solvation, and
electronic reorganization energy of anions obviously play
major roles in their electron-donating abilities.
2.11. The Central Dogma. Consistent with the Lewis
−
basicity trend of anions (OH− > F− > CN− > AcO− > H2PO4
> Cl− > Br− > I−),23 all experimental results show that, in
aprotic solvents, where smaller anions are not highly solvated,
strongly Lewis basic anions (OH−, F−, CN−, etc.) reduce
DCNDI, DPNDI, and other NDIs to corresponding radical
anion and dianion in two steps, whereas weaker Lewis basic
anions (AcO−, H2PO4 , Cl−, etc.) generate only NDI•− radical
−
anion from stronger π-acidic NDIs (e.g., DCNDI, NDIs 1 and
2), and even less Lewis basic I− can only reduce the most π-
acidic DCNDI to DCNDI•−. However, these observations
seemed counterintuitive to some critics because of a
misconception that higher electronegativity of F, O, and N
than I should render F−, OH−, and CN− ions weaker electron
donors than I− in all solvents. It was insisted that better
reducing ability of I− over other anions in H2O should be
preserved universally in all solvents. While large, less solvated I−
acts as a good electron donor in H2O,12a our studies clearly
demonstrate that it is misleading to extrapolate this trend in
aprotic organic solvents, in which Lewis basicity of less solvated
F−, OH−, and CN− ions is significantly higher. In aprotic
solvents (ODCB, MeCN, DMF, and DMSO), highly Lewis
basic F− acts as a strong electron donor that can reduce NDIs
to NDI•− and NDI2−, whereas in protic solvents, it becomes
highly solvated and stabilized via H-bonding and loses its ability
to reduce NDIs via ET. It is needless to say that electro-
negativity is an elemental property, which should not be
misconstrued for the anion’s electron-donating ability. Instead,
it is Lewis basicity that reflects anion’s electron-donating ability
in aprotic solvents, in which stabilization of anions through
solvation is minimum. Since all experimental evidence confirms
the reduction of NDIs with strongly Lewis basic anions and
rules out Meisenheimer complex formation and deprotonation
of NDIs, the above-mentioned misconception led critics to
offer alternative hypotheses for NDI reductions that would
avoid ET from Lewis basic anions. To draw a clear picture and
dispel any confusion, herein, we discuss these views and debunk
them all with evidence and logic.
3. CONCLUSIONS
Powerful evidence obtained from UV/vis, NMR, and EPR
spectroscopies, ITC, ESIMS, and XPS experiments all converge
to show that, depending on the Lewis basicity of anions and π-
acidity of NDIs, the nature of anion/NDI interactions ranges
from extremely weak non-chromogenic anion−π interactions to
anion-induced chromogenic charge-transfer, photoinduced
electron-transfer, and thermal electron-transfer interactions. In
spite of our conscious attempts to look for any evidence that
could indicate the formation of covalent anion−NDI
Meisenheimer complex or CH···anion H-bonded complex,
none was found to support these alternative possibilities. When
anions and NDIs are strong electron donors and acceptors,
respectively, positioning the HOMO of an anion above the
LUMO of an NDI, a thermal anion-to-NDI ET pathway is
turned ON. When the HOMO of a weakly Lewis basic anion
falls below the LUMO of an NDI but still lies above its
HOMO, the thermal ET is turned OFF, but light can activate
First, it was proposed that TBAX (X− = OH−, F−, CN−,
AcO−, Cl−, etc. but ≠ I−) could undergo β-elimination to form
Bu3N (Bu4N+X− → Bu3N + HX + CH2CHCH2CH3↑),25
which would then reduce NDIs to corresponding NDI•− and
NDI2− species. In control experiments, direct addition of Bu3N
and Et3N to DPNDI did not produce any DPNDI•− or
DPNDI2− species, dismissing this hypothesis. An alternative
hypothesis was that Lewis basic anions (X− = F−, OH−, AcO−)
in TBAX salts could abstract an α-H, forming a
1
PET from the anion’s HOMO to the photogenerated *NDI’s
SOMO−1. Both pathways generate the same NDI•− radical
anions, as evident from identical spectroscopic signals. Anion/
NDI CT interactions take place when anions are poor Lewis
bases and cannot trigger ET or PET. In aprotic solvents, where
stabilization of anions via solvation is minimum, electron-
donating abilities of anions are dictated by their Lewis basicity.
Thus, these comprehensive analyses not only depict a clear
picture of how anions interact with NDIs under different
conditions but also add a new paradigm to anion-recognition
+
CH3CH2CH2CH− NBu3 zwitterion (ylide) or other by-
products, which could then act as a common reducing agent for
NDIs. Even if this zwitterion is formed, which is unlikely and
unknown, it is implausible that such a stable zwitterion would
13689
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
RSC Publishing: London, 2006. (c) Anion Coordination Chemistry;
Bowman-James, K., Bianchi, A., Garcia-Espana, E., Eds.; Wiley-
Blackwell: New York, 2011. (d) Wenzel, M.; Hiscock, J. R.; Gale, P.
A. Chem. Soc. Rev. 2012, 41, 480−520.
chemistry, namely, anion-induced ET interactions. This knowl-
edge could be exploited in various practical applications,
ranging from anion sensing11a and fingerprinting to radical
polymerization, NDI-based battery technology, conducting
polymers, as well as molecular electronics and magnetism.
Anion-induced ET to NDIs, generating NDI•−, could have
undesired consequences during the synthesis of NDI-based
conjugated polymers29 and in other reactions involving NDI
compounds. On the other hand, carboxylate-induced ET to
strongly π-acidic NDIs could be exploited for biomimetic
synthesis of neurotransmitters from α-amino acids via radical
decarboxylation and initiation of radical polymerization, both of
which are currently under study in our laboratory.
(5) (a) Lehn, J.-M.; Sonveaux, E.; Willard, A. K. J. Am. Chem. Soc.
1978, 100, 4914−4916. (b) Best, M. D.; Tobey, S. L.; Anslyn, E. V.
Coord. Chem. Rev. 2003, 240, 3−15. (c) Cho, E. J.; Moon, J. W.; Ko, S.
W.; Lee, J. Y.; Kim, S. K.; Yoon, J.; Nam, K. C. J. Am. Chem. Soc. 2003,
125, 12376−12377. (d) Custelcean, R.; Delmau, L. H.; Moyar, B. A.;
Sessler, J. L.; Cho, W.-S.; Gross, D.; Bates, G. W.; Brooks, S. L.; Light,
M. E.; Gale, P. A. Angew. Chem., Int. Ed. 2005, 44, 2537−2542.
(e) Curiel, D.; Cowley, A.; Beer, P. D. Chem. Commun. 2005, 236−
238. (f) Kang, S. O.; Begum, R. A.; Bowman-James, K. Angew. Chem.,
Int. Ed. 2006, 45, 7882−7894. (g) Bhosale, S. V.; Bhosale, S. V.;
Kalyankar, M. B.; Langford, S. J. Org. Lett. 2009, 11, 5418−5421.
(6) Wade, C. R.; Broomsgrove, A. E. J.; Aldridge, S.; Gabbaï, F. P.
Chem. Rev. 2010, 110, 3958−3984.
ASSOCIATED CONTENT
* Supporting Information
■
S
(7) (a) Ma, J. C.; Dougherty, D. A. Chem. Rev. 1997, 97, 1303−1324.
Experimental section, and additional experimental and
computational results. This material is available free of charge
via the Internet at http://pubs.acs.org
(b) Garau, C.; Frontera, A.; Quinonero, D.; Ballester, P.; Costa, A.;
̃
Deya, P. M. J. Phys. Chem. A 2004, 108, 9423−9427.
(8) (a) Rosokha, Y. S.; Lindeman, S. V.; Rosokha, S. V.; Kochi, J. K.
Angew. Chem., Int. Ed. 2004, 43, 4650 −4652. (b) Wang, D.-X.; Zheng,
Q.-Y.; Wang, Q.-Q.; Wang, M.-X. Angew. Chem., Int. Ed. 2008, 47,
7485−7488. (c) Chifotides, H. T.; Schottel, B. L.; Dunbar, K. R.
Angew. Chem., Int. Ed. 2010, 49, 7202−7207. (d) Dawson, R. E.;
Henning, A.; Weimann, D. P.; Emery, D.; Ravikumar, V.; Montenegro,
J.; Takeuchi, T.; Gabutti, S.; Mayor, M.; Mareda, J.; Schalley, C. A.;
Matile, S. Nat. Chem. 2010, 2, 533−538.
AUTHOR INFORMATION
Corresponding Author
saha@chem.fsu.edu
■
Notes
The authors declare no competing financial interest.
(9) (a) Mulliken, R. S. J. Am. Chem. Soc. 1952, 74, 811−824.
(b) Mulliken, R. S.; Person, W. B. Molecular Complexes; Wiley: New
York, 1969.
(10) (a) Lokey, R. S.; Iverson, B. L. Nature 1995, 375, 303−305.
(b) 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.Eur. J. 2004, 10, 6375−6392. (c) Mukhopadhyay, P.; Iwashita,
Y.; Shirakawa, M.; Kawano, S.-I.; Fujita, N.; Shinkai, S. Angew. Chem.,
Int. Ed. 2006, 45, 1592−1595. (d) Schneider, H.-J. Angew. Chem., Int.
Ed. 2009, 48, 3924−3977.
(11) (a) Guha, S.; Saha, S. J. Am. Chem. Soc. 2010, 132, 17674−
17677. (b) Guha, S.; Goodson, F. S.; Roy, S.; Corson, L. J.;
Gravenmier, C.; Saha, S. J. Am. Chem. Soc. 2011, 133, 15256−15259.
(12) (a) Melby, L. R.; Harder, R. J.; Hertler, W. R.; Mahler, W.;
Benson, R. E.; Mochel, W. E. J. Am. Chem. Soc. 1962, 84, 3374−3387.
(b) Sun, D.; Rosokha, S. V.; Kochi, J. K. J. Phys. Chem. B 2007, 111,
6655−6666.
ACKNOWLEDGMENTS
■
This research was supported by an ACS-PRF DNI grant (PRF
no. 51737-DNI4) and FSU startup fund. We acknowledge Drs.
Igor Alabugin, Naresh Dalal, and Vasant Ramachandran (FSU)
for their assistance with computational analysis and EPR
studies.
REFERENCES
■
(1) (a) Quinonero, D.; Garau, C.; Rotger, C.; Frontera, A.; Ballester,
̃
P.; Costa, A.; Deya, P. M. Angew. Chem., Int. Ed. 2002, 41, 3389−3392.
(b) Mascal, M.; Armstrong, A.; Bartberger, M. D. J. Am. Chem. Soc.
2002, 124, 6274−6276. (c) Alkorta, I.; Rozas, I.; Elguero, J. J. Am.
Chem. Soc. 2002, 124, 8593−8598. (d) Garau, C.; Frontera, A.;
Quinonero, D.; Ballester, P.; Costa, A.; Deya, P. M. ChemPhysChem
2003, 4, 1344−1348. (e) Berryman, O. B.; Bryantsev, V. S.; Stay, D. P.;
Johnson, D. W.; Hay, B. P. J. Am. Chem. Soc. 2007, 129, 48−58.
(13) Njus, D.; Jalukar, V.; Zu, J.; Kelley, P. M. Am. J. Clin. Nutr. 1991,
54, 1179S−1183S.
(14) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B.
Angew. Chem., Int. Ed. 2002, 41, 2596−2599.
(f) Albrecht, M.; Wessel, C.; de Groot, M.; Rissanen, K.; Luchow, A. J.
̈
Am. Chem. Soc. 2008, 130, 4600−4601. (g) White, N. G.; Kitchen, J.
A.; Brooker, S. Eur. J. Inorg. Chem. 2009, 1172−1180.
(15) (a) Berryman, O. B.; Bryantsev, V. S.; Stay, D. P.; Johnson, D.
W.; Hay, B. P. J. Am. Chem. Soc. 2007, 129, 48−58. (b) Schneider, H.;
Vogelhuber, K. M.; Schinle, F.; Weber, J. M. J. Am. Chem. Soc. 2007,
129, 13022−13026. (c) Hay, B. P.; Bryantsev, V. S. Chem. Commun.
2008, 2417−2428. (d) Albrecht, M.; Muller, M.; Mergel, O.; Rissanen,
K.; Valkonen, A. Chem.Eur. J. 2010, 16, 5062−5069.
(2) (a) Schneider, H.-J.; Werner, F.; Blatter, T. J. Phys. Org. Chem.
1993, 6, 590−594. (b) de Hoog, P.; Gamez, P.; Mutikainen, I.;
Turpeinen, U.; Reedijk, J. Angew. Chem., Int. Ed. 2004, 43, 5815−5817.
́
(c) Campos-Fernandez, C. S.; Schottel, B. L.; Chifotides, H. T.; Bera,
J. K.; Bacsa, J.; Koomen, J. M.; Russell, D. H.; Dunbar, K. R. J. Am.
Chem. Soc. 2005, 127, 12909−12923. (d) Gorteau, V.; Bollot, G.;
Mareda, J.; Perez-Velasco, A.; Matile, S. J. Am. Chem. Soc. 2006, 128,
14788−14789. (e) Berryman, O. B.; Hof, F.; Hynesc, M. J.; Johnson,
D. W. Chem. Commun. 2006, 506−508. (f) Mascal, M.; Yakovlev, I.;
Nikitin, E. B.; Fettinger, J. C. Angew. Chem., Int. Ed. 2007, 46, 8782−
8784. (g) Berryman, O. B.; Johnson, D. W. Chem. Commun. 2009,
3143−3153.
(3) (a) Gamez, P.; Mooibroek, T. J.; Teat, S. J.; Reedjik, J. Acc. Chem.
Res. 2007, 40, 435−444. (b) Schottel, B. L.; Chifotides, H. T.; Dunbar,
K. R. Chem. Soc. Rev. 2008, 37, 68−83. (c) Frontera, A.; Gamez, P.;
Mascal, M.; Mooibroek, T. J.; Reedjik, J. Angew. Chem., Int. Ed. 2011,
50, 9564−9583. (d) Ballester, P. Acc. Chem. Res. 2012, DOI: 10.1021/
ar300080f.
(16) (a) Mooibroek, T. J.; Black, C. A.; Gamez, P.; Reedijk, J. Cryst.
Growth Design 2008, 8, 1082−1093. (b) Hay, B. P.; Custelcean, R.
Cryst. Growth Design 2009, 9, 2539−2545.
(17) Li, Y.; Flood, A. H. Angew. Chem., Int. Ed. 2008, 47, 2649−2652.
(18) (a) Strauss, M. J. Chem. Rev. 1970, 70, 667−712. (b) Olson, E.
J.; Xiong, T. T.; Cramer, C. J.; Buhlmann, P. J. Am. Chem. Soc. 2011,
̈
133, 12858−12865.
(19) (a) Kim, D. Y.; Singh, N. J.; Lee, J. W.; Kim, K. S. J. Chem. Theo.
́
Comp. 2008, 4, 1162−1169. (b) Campo-Cacharron, A.; Cabaleiro-
Lago, E. M.; Rodríguez-Otero, J. ChemPhysChem 2012, 13, 570−577.
(20) (a) Bhosale, S. V.; Janiab, C. H.; Langford, S. J. Chem. Soc. Rev.
2008, 37, 331−342. (b) Lin, N.-T.; Jentzsch, A. V.; Guen
́
ee
́
, L.;
(4) (a) Beer, P. D.; Gale, P. A. Angew. Chem., Int. Ed. 2001, 40, 486−
516. (b) Sessler, J. L.; Gale, P. A.; Cho, W.-S. Anion Receptor Chemistry;
Neudorfl, J.-M.; Aziz, S.; Berkessel, A.; Orentas, E.; Sakai, N.; Matile, S.
̈
Chem. Sci. 2012, 3, 1121−1127.
13690
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691

Journal of the American Chemical Society
Article
(21) (a) Thalacker, C.; Roger, C.; Wurthner, F. J. Org. Chem. 2006,
̈
̈
71, 8098−8105. (b) Roger, C.; Wurthner, F. J. Org. Chem. 2007, 72,
̈
̈
8070−8075. (c) Sakai, N.; Mareda, J.; Vauthey, E.; Matile, S. Chem.
Commun. 2010, 46, 4225−4237. (d) Misek, J.; Jentzsch, A. V.; Sakurai,
S.-I.; Emery, D.; Mareda, J.; Matile, S. Angew. Chem., Int. Ed. 2010, 49,
7680−7683.
(22) (a) Gosztola, D.; Niemczyk, M. P.; Svec, W.; Lukas, A. S.;
Wasielewski, M. R. J. Phys. Chem. A 2000, 104, 6545−6551. (b) Andric,
G.; Boas, J. F.; Bond, A. M.; Fallon, G. D.; Ghiggino, K. P.; Hogan, C.
F.; Hutchison, J. A.; Lee, M. A.-P.; Langford, S. J.; Pilbrow, J. R.;
Troup, G. J.; Woodward, C. P. Aust. J. Chem. 2004, 57, 1011−1019.
(23) (a) Huheey, J. E.; Keiter, E. A.; Keiter, R. L. Inorganic Chemistry
Principles of Structure and Reactivity, 4th ed.; Harper Collins: New
York, 1993; pp 332−333. (b) Anslyn, E. V.; Dougherty, D. A. Modern
Physical Organic Chemistry; University Science Book: Sausalito, CA,
2006.
(24) Cametti, M.; Rissanen, K. Chem. Commun. 2009, 2809−2829.
(25) Sun, H.; DiMagno, S. G. J. Am. Chem. Soc. 2005, 127, 2050−
2051. While anhydrous TBAF may decompose through Hofmann
elimination, TBAF·3H2O does not. If TBAF·3H2O generated a
flammable 1-butene gas and Bu3N via β-elimination, it would not be
available in plastic containers from industrial manufacturers.
(26) Connelly, N. G.; Geiger, W. E. Chem. Rev. 1996, 96, 877−910.
NOBF4 (ERed = +1.25 V vs SCE in MeCN) oxidizes colorless TBAX
salts (X− = OH−, F−, Cl−, Br−, I−) in MeCN. NOBF4 generates yellow
and green gases from TBAF and TBACl, respectively, and turns
colorless TBABr and TBAI solutions orange and brown, respectively,
indicating the formation of the corresponding halogen molecules
(Figure S11). Reduction of NO+ with these anions should generate
NO, which can be oxidized by O2 to brown NO2 gas. Emission of
brown NO2 fume is clearly visible with OH−, when no other colored
species is formed. Due to relatively low solubility of NDIs (mM) and
an immediate formation of intensely colorful NDI•− and NDI2−,
formation of X2 by NDI could not be observed by the naked eye.
(27) Morgan, W. E.; Van Wazer, J. R.; Stec, W. J. J. Am. Chem. Soc.
1973, 95, 751−755.
(28) Guha, S.; Goodson, F. S.; Clark, J. R.; Saha, S. CrystEngComm
2012, 14, 1213−1215.
(29) Alvey, P. M.; Iverson, B. L. Org. Lett. 2012, 14, 2706−2709.
13691
dx.doi.org/10.1021/ja303173n | J. Am. Chem. Soc. 2012, 134, 13679−13691