Optimizing electron transfer mediators based on arylimidazoles by ring fusion: Synthesis, electrochemistry, and computational analysis of 2-aryl-1-methylphenanthro[9,10-d ]imidazoles

Article abstract of DOI:10.1021/ja410865z

A significant improvement of the properties of redox catalysts based on the triarylimidazole framework can be achieved with a simple structural modification. By linking the ortho-carbons of the aromatics positioned at C-4 and C-5, a fused framework is generated, removing the distortion from planarity and enhancing the influence of the substituents on the redox properties. This modification leads not only to a much broader range of available redox potentials for the resulting phenanthro[9,10-d]imidazoles but also to improved stability of the corresponding radical cation. These concepts were verified with eight new phenanthro[9,10-d]imidazole derivatives, using cyclic voltammetry and DFT calculations. For this purpose, an optimized and general synthetic route to the desired compounds was developed. An excellent linear correlation of the calculated effective ionization potentials with the experimental oxidation potentials was obtained, allowing for an accurate prediction of oxidation potentials of derivatives yet to be synthesized. Moreover, high catalytic activity was found for electro-oxidative C-H activation reactions.

Full text of DOI:10.1021/ja410865z

Article
pubs.acs.org/JACS
Optimizing Electron Transfer Mediators Based on Arylimidazoles by
Ring Fusion: Synthesis, Electrochemistry, and Computational
Analysis of 2‑Aryl-1-methylphenanthro[9,10‑d]imidazoles
Robert Francke and R. Daniel Little*
Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106-9510, United States
S
* Supporting Information
ABSTRACT: A significant improvement of the properties of redox catalysts based on
the triarylimidazole framework can be achieved with a simple structural modification. By
linking the ortho-carbons of the aromatics positioned at C-4 and C-5, a fused framework
is generated, removing the distortion from planarity and enhancing the influence of the
substituents on the redox properties. This modification leads not only to a much broader
range of available redox potentials for the resulting phenanthro[9,10-d]imidazoles but
also to improved stability of the corresponding radical cation. These concepts were
verified with eight new phenanthro[9,10-d]imidazole derivatives, using cyclic
voltammetry and DFT calculations. For this purpose, an optimized and general
synthetic route to the desired compounds was developed. An excellent linear correlation
of the calculated effective ionization potentials with the experimental oxidation potentials was obtained, allowing for an accurate
prediction of oxidation potentials of derivatives yet to be synthesized. Moreover, high catalytic activity was found for electro-
oxidative C−H activation reactions.
electron transfer mediation currently attract attention in other
fields. For instance, given the similarities between photo-
sensitized and electrochemically mediated transformations, it is
reasonable to anticipate that catalysts originally designed for
indirect electrosynthesis will likely prove useful as excited state
electron transfer agents in photoinitiated processes.17 Another
application of such compounds lies in the field of battery
research, where they have been used as redox shuttles for
overcharge protection.18
The most popular oxidative mediator system is represented
by the triarylamines, with numerous applications and a broad
range of accessible potentials.10,15,16,19−24 However, they tend
to undergo intermolecular oxidative coupling reactions, thereby
reducing the reversibility of the redox couple and making a para
substitution of the aryl rings necessary.25,26 Another drawback
to their use in some instances is that a high molar percentage of
the triarylamine must be employed in order to achieve a
complete conversion.10,16 To circumvent many of these issues,
we recently developed a new class of metal-free, easy to
synthesize redox catalysts based on the triarylimidazole
framework (1) (see Scheme 1, left).27 We have demonstrated
that their oxidation potential can be tuned within a wide range
by modification of the substitution pattern on the aromatics.28
They have already proven to be useful mediators for the
activation of benzylic C−H bonds under mild conditions (see
Scheme 1, right). While the range of accessible potentials for
triarylimidazoles is good, their significant distortion from
planarity presents a problem because the effectiveness of
INTRODUCTION
■
Organic electrosynthesis is recognized as an environmentally
compatible methodology since toxic and dangerous oxidizing or
reducing reagents can be replaced and the overall energy
consumption can be reduced.1,2 Moreover, unstable or
hazardous reagents can be produced in situ and reactions
carried out under very mild conditions.3−5 One approach to
increase the efficiency of electro-organic synthesis is to use
electrochemically generated redox agents. Since the electron
transfer step is shifted from a heterogeneous to a homogeneous
process (an indirect electrolysis), the kinetic inhibition which is
usually associated with the electron transfer from electrode to
substrate can be eliminated. Typically, higher and/or totally
different selectivity is achieved.6−8 If one or more of the
subsequent steps is an irreversible chemical reaction, then the
electron transfer can occur even against a potential gradient.8,9
Therefore, much higher or totally different selectivity can be
achieved with lower energy consumption. In some difficult
cases, electron transfer mediators help to avoid overoxidation of
the substrate or avoid electrode passivation that may result by
the formation of a polymer film on the electrode surface.8,10
Ideally, the mediator engages in a reversible redox couple, one
that is initiated at the electrode and is followed by a reaction of
interest, thereby allowing for catalytic employment of the
electron transfer mediator, in order to avoid reagent waste and
difficult separation procedures.
Over the past decade, many intriguing new developments
have been observed in this field, including indirect flash
electrochemistry, mediated reductive dehalogenation using o-
carboranes, or housane rearrangements triggered by redox
catalysis.11−16 Furthermore, compounds originally designed for
Received: October 23, 2013
Published: December 13, 2013
© 2013 American Chemical Society
427
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
Scheme 1. Triarylimidazoles (1) and Phenanthro[9,10-d]imidazoles (2) as Mediators for Electro-organic Synthesis (Left) and
Application for Electro-oxidative C−H Activation at Benzylic Positions (Right)
Table 1. Substitution Patterns, Sum of the σ+ Values, and Oxidation Potentials Eox,1 (vs Ag/AgNO3) of the Synthesized
Phenanthroimidazoles 2 and the Corresponding Triarylimidazoles 1
substituents to influence the potential is impaired.28 Moreover,
the limited stability of the radical cation formed upon electro-
RESULTS AND DISCUSSION
■
Synthesis. Eight 2-aryl-1-methylphenanthro[9,10-d]-
oxidation leads to a quasi- rather than a fully reversible
oxidation process.27 In the present work, we demonstrate how
these problems can be overcome by linkage of the ortho-
carbons of the aromatics positioned at C-4 and C-5 of the
imidazole subunit to generate the fused framework 2. Toward
this end, a series of substituted phenanthro[9,10-d]imidazoles
was synthesized, an optimized and general procedure was
developed, and their characteristics were studied using cyclic
voltammetry and bulk electrolysis. Using DFT calculations, the
redox properties of these systems could be explained and
oxidation potentials of derivatives yet to be synthesized become
predictable. We note with interest that our systems may be of
use in the field of optoelectronics. This realization stems from
recent reports that the photoemissive properties of phenanthro-
[9,10-d]imidazoles combined with their high thermal stability
and easy modifiability make them useful building blocks for the
assembly of functional layers in organic light-emitting diodes
(OLEDs) or for photosensitizers in dye-sensitized solar
cells.29−33
imidazoles (2) bearing methoxy, methyl, bromo, and fluoro
moieties in three different positions were synthesized (for
particular substitution patterns, see Table 1).34 Initially, we
attempted a direct condensation of phenanthrene-9,10-quinone
3 with benzaldehyde 4, methylamine, and ammonium
acetate.31,32 Unfortunately, the method primarily afforded a
mixture of oxazoles and several structures where the imidazole
ring failed to form.
In further attempts, 3, 4, NH4OAc, and N-methylamine were
allowed to react in the presence of NaH2PO4 in a manner
analogous to a procedure that was previously used to synthesize
1-methyltriarylimidazoles, 1.37 Once again, however, mostly
oxazole formation was observed. We then explored a two-step
sequence calling first for the condensation of 3 and 4, in the
presence of 2 equiv of NH4OAc, leading to the formation of the
1H-imidazole, followed by N-methylation. For the first step, we
tried a procedure known to lead to unsubstituted 2-aryl-1H-
phenanthro[9,10-d]imidazoles.38 Although this approach
turned out to be more successful in terms of imidazole ring
formation, the 1H-imidazole was accompanied by substantial
amounts of side products, thereby leading to a tedious
428
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
Scheme 2. Synthesis of Substituted Phenanthro[9,10-d]imidazoles 2
Scheme 3. Synthesis of Substituted Phenanthrene-9,10-quinones as Precursors for Phenanthro[9,10-d]imidazoles 235,36
purification process. We eventually discovered that a clean
conversion to the 1H-imidazole framework could be achieved
in the presence of NaH2PO4 with ethanol as solvent at 150 °C
(sealed tube; see Scheme 2), rendering pure 1H-phenanthro-
imidazoles in 83−94% yield after simple trituration of the crude
product. The subsequent N-methylation proceeds smoothly at
room temperature in DMF after deprotonation of the 1H-
imidazole with NaH and addition of methyl iodide (yield after
recrystallization = 91−95%).39
Bromo- and methoxy-substituted phenanthrene-9,10-qui-
nones 3b and 3c were obtained using a sequence elaborated
by MacLachlan et al. (see Scheme 3),35 starting with the
bromination of 3a using Br2 in nitrobenzene and dibenzoyl
peroxide as a radical initiator.36 Since a direct methoxylation of
3b failed,35 it was first reduced and then converted to the
bisenol ether 5 prior to aryl methoxylation with NaOMe in the
presence of catalytic amounts of CuBr. The resulting
intermediate, 3,6,9,10-tetramethoxyphenanthrene, was then
oxidized using ceric ammonium nitrate to obtain phenanthro-
imidazole precursor 3c.
Electrochemical Characterization. The electrochemical
properties of compounds 2 were studied with cyclic
voltammetry, using an electrolyte consisting of 0.2 M LiClO4
in CH3CN/CH2Cl2 (4:1 by volume), a glassy carbon working
electrode, and a platinum wire as the counter electrode. A silver
wire in 0.01 M AgNO3, separated from the analyte by a Vycor
frit, was used as the reference. To allow comparison of the
potentials recorded herein with other common reference
electrodes, one simply adds +87 mV to convert to the
ferrocene redox couple, +298 mV for SCE and +542 mV for
NHE.40 Each of the phenanthroimidazoles exhibited between
two and four peaks in the anodic scan and no activity in the
negative potential range (see Supporting Information Table
S1). At 100 mV s−1, the first redox couple is chemically and
electrochemically reversible (ΔEp ≈ 60 mV indicative of
electrochemical reversibility, and ip,ox/ip,red ≈ 1 demonstrating
chemical reversibility), indicating that the heterogeneous
electron transfer is fast and that the species formed during
the oxidation process is stable on the voltammetry time scale.41
Since the peak separation ΔEp between anodic and cathodic
peaks points toward a single electron transfer, it can be assumed
that the oxidized species is radical cation 2•+. Its role in
electrocatalytic processes is discussed later and is supported by
computational results.
Since the potential difference between the first and the
second oxidation peak (Eox,1 and Eox,2) is between 350 and 480
mV, depending on the derivative (see Supporting Information
Table S1 and Figures S3−S5), 2•+ can be formed exclusively by
operating at the less positive potential. Presumably, the second
irreversible oxidation peak corresponds to the oxidation of 2•+
to dication 22+. This dication seems to undergo a fast chemical
reaction (probably deprotonation), and the following peaks
could correspond to the oxidation of the products of the follow-
up reaction(s). A large difference between the potentials Eox,1
and Eox,2 is essential for application in electrosynthesis in order
to avoid side reactions and loss of catalytic activity. Another
important attribute of the systems described herein is that they
are electrochemically inert in the negative potential range, out
to the negative stability limit of the electrolyte (−2.4 V vs Ag/
AgNO3). The practical implication of this observation is that an
undivided electrolysis cell may be considered for possible
applications, as long as the chemical follow-up reaction is fast
enough to avoid diffusion of 2•+ to the cathode where it would
be discharged.
The first oxidation peak potentials of 2 (Eox,1) are
summarized in Table 1 and compared to the corresponding
triarylimidazoles 1.28 A full list of all peak potentials is provided
in the Supporting Information (see Table S1). The values are in
the range between 0.59 V (2h, R1 = OMe, R2 = OMe) and 1.09
V (2a, R1 = Br, R2 = F) versus Ag/AgNO3 (0.89−1.39 V vs
SCE and 1.13−1.63 vs NHE). With such relatively high
oxidation potentials, 2•+ can be considered to be strong
oxidizing agents, situated in the same range of common
oxidizers such as Ce3+/Ce4+ (1.61 V vs NHE) or the tris-para-
bromophenylamine radical cation (1.30 V vs NHE).8 The
potentials follow a clear trend with the values increasing as the
number of electron-withdrawing substituents (Br, F) increases
and the number of electron-donating substituents (OMe, Me)
diminishes. With the compounds synthesized thus far, a
potential range of 0.5 V is accessible, which means an increase
of about 40% compared to the equally substituted triarylimi-
dazoles 1 (potential range for 1: 0.36 V). The influence of the
429
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
substituents on the electronic character of the aromatic ring
system is therefore significantly enhanced.
refers to the oxidation potential of the unsubstituted compound
of the series.43,44 For both systems 1 and 2, a linear relationship
is observed (see Figure 2), with a coefficient of determination
(R2) of 0.963 for 2 (see eq 1) and 0.973 for 1 (see eq 2).
As previously noted, an important characteristic of an
electron transfer mediator is the reversibility of its redox
couple. Low reversibility usually means that the mediator is
prone to side reactions, leading to lower catalytic turnover
numbers. As noted above, the first redox process for 2 is
reversible at a scan rate of 100 mV s−1. Furthermore, the ratio
between the anodic and cathodic peak current is not
significantly affected by the substitution pattern (see Figure
S1). In order to compare the reversibility of systems 2 and 1,
the voltammograms were recorded at a lower scan rate; the
results for 1e and 2e are plotted in Figure 1 (R1 = H, R2 =
Eox,1(2) = 1.015 + 0.192
Eox,1(1) = 0.949 + 0.134
σ+ (R2 = 0.963)
σ+ (R2 = 0.973)
∑
∑
(1)
(2)
Figure 2. Plot of the observed oxidation potential Eox,1 versus the sum
of Hammett σ+ values, Σσ+.
Thus, in both cases, the correlation of Eox,1 with Σσ+ provides
a convenient way to estimate the oxidation potential of a
compound not yet synthesized. Furthermore, we note that a
comparison of the slopes for the triarylimidazole framework 1
with that of the phenanthroimidazole structures 2 reveals that
the influence of the substitution pattern is much more
pronounced for the latter because the slope increases by 43%,
from 0.134 to 0.192.
A reasonable explanation of the considerably higher
Hammett slope is revealed when one considers the dihedral
angles (ϕ, ϕ′, and ϕ″, see Figure 3, top) of the computed
structures of the unsubstituted systems 1c and 2c (Figure 3,
second row from top). Each dihedral reflects the degree to
which the aromatic rings deviate from the planarity of the
imidazole core. The geometry-optimized structures were
obtained using the B3LYP hybrid functional together with
the 6-31+G* basis set.45 Solvation effects were accounted for
using the polarized continuum model SMD with acetonitrile as
solvent.46 The largest torsion in framework 1 is expected for ϕ
since the aryl group appended to C-5 is adjacent to a second
aryl moiety and a methyl group. Indeed, for 1c, a very
significant twist of the aryl ring out of the plane is observed (ϕ
= 58°). In comparison, the distortion for the phenanthro-
imidazole 2c at the equivalent location is very small (ϕ = 3°).
Also, for 1c, the aryl rings positioned at C-4 and C-2 exhibit
strong torsions with ϕ′ = 34° and ϕ″ = 40°. In contrast, for 2c,
the dihedral angle is slightly larger at position 2 (ϕ″ = 45°),
while dihedral angle ϕ′ is 0.
Figure 1. Cyclic voltammograms of compounds 1e (left) and 2e
(right). Scan rate: 100 mV s−1 (solid line) and 10 mV s−1 (dashed
line).
OMe; for additional examples, see Figure S2). For ease of
comparison, the observed current is normalized to the peak
current ip,ox; thus the ordinate is illustrated as i/ip,ox. Whereas at
10 mV s−1 for 1e the current drops substantially during the
back-scan, only a slight decrease of the cathodic current is
visible for 2e. These results impressively demonstrate how
much more stabilized the fused framework 2•+ is compared to
1•+. Compared to the triarylamines, for which quasi-reversibility
can only be achieved by blocking each of the para-positions on
the aryl rings,23,26 the stability of the radical cation 2•+ is even
more pronounced. Oxidative couplings do not seem to play an
important role since reversibility is achieved with partially
blocked derivatives 2b and 2e as well as with unsubstituted
compound 2c. It should be noted that phenanthroimidazoles
without N-substitution undergo an irreversible oxidation
process (see Figure S6), most likely due to deprotonation
after formation of the radical cation. Hence, N-alkylation seems
to be crucial for application as an electrocatalyst.
Computational Analysis and Empirical Correlations.
As noted earlier, the potential range accessible to compounds
2a−2h is 0.5 V. Compared to the identically substituted
triarylimidazoles 1a−1h whose range is 360 mV, this amounts
to an increase of about 40%. A mathematical description of the
influence of the substituents on the oxidation potential Eox,1 was
obtained by plotting the observed Eox,1 versus the sum of the
Hammett σ+ constants of the substituents appended to the aryl
rings.42,43 The slope provides a measure of the influence of the
substituents upon the observed potential, while the intercept
Obviously, the sum of the dihedral angles is significantly
smaller for 2 than for 1. This implies that there should be
increased delocalization with a concomitant improvement in
the ability of substituents to communicate their electronic
character. Consequently, each substituent that can interact with
the π-system by resonance should have a greater influence on
430
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
the plane with a decrease of ϕ″ from 45 to 33°. Again, the sum
of the dihedral angles is significantly smaller for 2c•+, indicating
that compared to 1c•+ an improved stabilization of the radical
and the positive charge is achieved by ring fusion. This
conclusion is consistent with the higher reversibility of the
oxidation of 2 compared to 1 (see Figure 1). In order to
confirm and generalize the results obtained for the unsub-
stituted systems 1c and 2c, the dihedral angles ϕ, ϕ′, and ϕ″ for
all computed compounds (1a−1h, 1a•+−1h•+, 2a−2h, and
2a•+−2h•+) were determined and summarized in Table S7; in
each case, the same trend is observed.
Apart from the increased Hammett slope and the improved
reversibility, a further advantage inherent to the phenanthro-
imidazoles could arise from the planarity of the radical cation.
In contrast to 1•+, the nearly flat radical cation 2•+ provides a
more accessible platform especially when the substrate to be
oxidized contains an aromatic electrophore that might engage
in cation−π interactions.
Moreover, the computational results reveal that for both 2c•+
and 1c•+ the spin density (Figure 3, fourth row from top) and
the positive charge (Figure 3, bottom row) are mainly located
within the imidazole subunit. It can therefore be assumed that
the imidazole ring represents the reactive center of the activated
form of the mediator. Since this center is more exposed in 2c•+
than in 1c•+, we anticipate improved interaction between the
SOMO (singly occupied molecular orbital) of the oxidized
form of the mediator and the HOMO of the substrate. This
notion that greater accessibility can be correlated with
oxidizability has been examined by Kochi et al. for the
photoinduced electron transfer reaction between excited state
quinones and a series of arene donors.47 Their study indicated
how the electron transfer can be changed from an inner-sphere
to an outer-sphere mechanism by controlling the steric
periphery of the arene donors, thereby influencing the ability
of the reacting partners to approach one another. Whether
there is a similar influence of the planarity of a mediator system
2/2•+ on the electron transfer rate remains to be explored.
Compounds 2a−h were examined to determine whether
there is an empirical relationship between the first oxidation
potential Eox,1 and the effective ionization potential IP*
(calculation method 1) or the HOMO energy εH (calculation
method 2), and these quantities were determined using DFT-
based methods. Should they exist, such correlations would
provide a convenient tool that could be used to predict Eox,1 for
a structure yet to be synthesized and allow one to tailor a
mediator to a specific redox reaction whose oxidation potential
is known. We were also interested in determining the minimum
level of theory necessary for an accurate prediction since the
use of larger basis sets, diffuse orbitals, and the inclusion of
solvent models increases computation times. For the
calculation of the effective ionization potentials (IP*), we
used the B3LYP hybrid functional starting with a 6-31G* basis
set (vacuum) followed by inclusion of diffuse functions (6-
31+G*, vacuum). Finally, acetonitrile was incorporated because
it was the solvent used in the CV experiments,48 by applying
the 6-31+G* basis set in combination with the polarized
continuum model SMD.46 The values for IP* were obtained
according to eq 3 by subtraction of the SCF energy εSCF of
radical cation 2•+ from that of the neutral form 2 (see Tables
S2−S4).
Figure 3. Dihedral angles ϕ, ϕ′, and ϕ″ of 1c and 2c (top); computed
geometries (second row from top) and HOMO maps (third row from
top) for 1c (left) and 2c (right); spin density (fourth row from top)
and electrostatic potential (bottom row; unit for color definition: kcal
mol−1 e−1) for radical cations 1c•+ (left) and 2c•+ (right); structures
and surfaces calculated on the B3LYP/6-31+G* level (solvent:
acetonitrile).
the oxidation potential in the phenanthroimidazoles 2. As
noted in Figure 3 (third row from top), the HOMO of 2 is
spread throughout the phenanthro and imidazole subunits, and
the aryl ring appended to C-5 in 2c contributes more
significantly than the analogous ring in 1c.
In the computed structures of the corresponding radical
cations 2c•+ and 1c•+, the dihedral angles ϕ, ϕ′, and ϕ″ alter
significantly compared to their neutral forms. For 1c•+ each
angle decreases between 5 and 10°, allowing for a better
conjugation and therefore better stabilization of the radical and
the positive charge. For 2c, a slight increase of ϕ and ϕ′ is
observed, a shift from 3 to 5° and from 0 to 1°, respectively.
Compared to the neutral form 2, the aryl ring at C-2 twists into
IP* = εSCF(2•+) − εSCF(2)
(3)
431
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
Both energies εSCF corresponded to the geometry-optimized
structures. By adopting this approach, the differences in time
scale between a voltammetry experiment and vertical excitation
are taken into consideration. In particular, the radical cation has
sufficient time to relax to an energy minimum due to the
significantly longer voltammetry time scale. A similar analysis
has been used by Fry et al. in conjunction with substituted
benzalacetophenones and triarylamines.43,44 Because this
approach does not satisfy the definition of the ionization
potential, we prefer to use the term effective ionization
potential and use the symbol IP*.
the linear regressions are summarized, and in Figure 4, the plots
of Eox,1 versus IP* and Eox,1 versus εH are depicted.52 Excellent
linear correlations with R2 close to unity can be obtained
applying method 1, using both B3LYP/6-31G* and B3LYP/6-
31+G*, with slightly better results when the solvent model is
applied. Apparently, inaccuracies are canceled out when the
energy difference between the neutral form and radical cation is
considered. Thus, the improvement obtained by using a more
sophisticated computational method is rather small. All three
outperform the Hammett approach using ∑σ+ (see eq 1)
though the simplicity of the latter is most appealing.
According to Koopman’s theorem, the ionization potential IP
can be derived from the HOMO energy εH using the following
relationship.49
In contrast, with a coefficient of determination R2 of 0.939,
the employment of method 2 (B3LYP/6-31G*/vacuum) is
rather imprecise. However, R2 increases dramatically as the
sophistication of the method increases; calculations using
B3LYP with a 6-31+G* basis set and inclusion of solvent
afforded an excellent linear correlation of Eox,1 versus εH (with
R2 = 0.993). We conclude that choosing an appropriate level of
theory seems to be essential when method 2 is used. The key to
obtaining the highest accuracy (but also with the highest
computational cost) is to use method 1 with B3LYP/6-31+G*
in combination with the solvent model (R2 = 0.997). The fact
that both correlation methods lead to linear relationships might
stem from a linear dependence of the relaxation energy for
radical cations 2•+.
IP = −εH
(4)
In contrast to Hartree−Fock theory, this relationship becomes
accurate when DFT-based methods are used to calculate
εH.50,51 These IP values have a different physical significance
than IP* derived from the subtraction approach described above
(vertical ionization potential vs energy difference between 2
and 2•+, eq 3). In this context, we calculated εH of 2 for each
level of theory (see Table S5) to determine whether there is an
empirical relationship between Eox,1 and εH for this series of
compounds (calculation method 2).
For both method 1 and method 2, a linear relationship is
observed at each level of theory. In Table 2, the parameters for
To expand the oxidation potential of system 2 farther into
the positive range, the introduction of moieties with stronger
electron-withdrawing character is necessary. In Table 3, the
Table 2. Intercepts, Slopes, and Coefficients of
Determination (R2) for Linear Regressions of the Plots of
Eox,1 versus Calculated IP and εH, Respectively
Table 3. Projection of the Oxidation Potentials Eox,1 (vs Ag/
AgNO3) of Nitro-Substituted Phenanthroimidazoles 2i−2k
Using Calculation Method 1 (B3LYP/6-31G*, vacuum)
method 1 (Ip*)
method 2 (εH)
no.
R1
R2
IP* (eV)
Eox,1 (V)
level of
theory
intercept slope
R2
intercept
slope
R2
2i
2j
2k
H
NO2
H
6.87
7.34
7.70
1.18
1.44
1.64
B3LYP/6-
31G*
−2.619
−2.864
−3.509
0.553
0.570
0.856
0.982
−2.054
−2.339
−3.732
−0.569
0.939
NO2
NO2
NO2
(vac.)
B3LYP/6-
31+G*
(vac.)
0.986
0.997
−0.592
−0.846
0.956
0.993
predicted oxidation potentials of three conceivable nitro-
substituted derivatives (2i−2k) are summarized. To obtain
this information, the approach involving IP* (calculation
method 1, B3LYP/6-31G*, vacuum) was used. The results
suggest that the present limit of 1.09 V versus Ag/AgNO3 (2h)
B3LYP/6-
31+G*
(CH3CN)
Figure 4. Correlations of Eox,1 with IP* (left) and Eox,1 with εH (right) for series 2 (IP* and Eox,1 were calculated using B3LYP/6-31+G* and the PCM
solvent model).
432
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
could be steadily increased up to 1.64 V (2k) by stepwise
introduction of NO2 groups. This extension would result in a
potential range of 1.05 V for compounds 2a−2k. With respect
to indirect electro-organic synthesis, the oxidation potentials of
most common functional groups would be covered by mediator
system 2.53 Considering the fact that for a mediated electron
transfer the redox potential of the mediator can be up to 0.5 V
below the potential of the substrate, this range ought to be even
broader.7,8,28,54
Application of 2 as Electrocatalyst. To establish that 2
can serve as a useful mediator, we selected two substrates, viz.
p-anisyl alcohol (6) with an oxidation potential of 1.21 V versus
Ag/AgNO3 and p-anisyl-n-octyl ether 7 (Eox = 1.33 V vs Ag/
AgNO3). From series 2, we selected mediators 2c and 2a whose
oxidation potentials are 1.00 and 1.09 V, respectively (note
Table 1). For the oxidation of 6, mediator 2c appeared to be a
suitable candidate, the potential difference between the two
being 0.21 V. For 7, mediator 2a was a promising candidate
with ΔEox = 0.24 V. The results of the cyclic voltammetry
studies are shown in Figure 5 (the currents are normalized to
increase by a factor of 9 for 2c/6 and by a factor of 12 in the
case of 2a/7 (green dashed lines). It was also confirmed that
this increase of the current does not stem from a potential shift
of the oxidation potential of the substrate upon addition of base
(see Figure S7). It should be noted that the catalytic currents of
mediators 2a and 2c are much more pronounced compared to
the electro-oxidation of 6 and 7 using triarylimidazole catalysts
1a and 1c under the same conditions (see Figures S8 and S9).
With a base concentration of 50 mM, the current at the
mediator peak potential is increased by a factor of 2 to 4.
A potentiostatic bulk electrolysis was conducted at 0.98 V
versus Ag/AgNO3, using mediator 2a in combination with
substrate 6 in a divided cell (experimental details; see
Supporting Information). The oxidation product p-anisalde-
hyde was obtained in a 75% isolated yield (95% with respect to
recovered starting material). The cyclic voltammetry data and
the product selectivity suggest that the reaction mechanism in
Scheme 1 (right side) is also applicable for this case.
CONCLUSION
■
Using our generalized synthetic procedure, substituted
phenanthroimidazoles 2 are easy to synthesize. With the
derivatives prepared so far, an oxidation potential range of 0.5 V
is accessible, which is wide enough to oxidize many common
functional groups, especially when it is taken into account that
for a mediated electron transfer the redox potential of the
mediator can be up to 0.5 V below the potential of the
substrate, sometimes more.7,8,28,54
In comparison to the triarylimidazoles 1, the influence of the
aryl substituents on the first oxidation event is dramatically
increased, and their redox behavior is far more reversible. Since
the phenanthroimidazole framework provides an opportunity
for tuning the redox potential (two different types of
substituents in the para position can be introduced in one
step), the potential range could be extended beyond that we
have explored thus far by using substituents with stronger
electron-donating or electron-withdrawing properties, such as
amino or nitro groups.
With the calculation of IP* and εH and the correlation to Eox,1
of a set of derivatives, we are able to predict the oxidation
potentials of phenanthroimidazoles not yet synthesized.
Consequently, tedious trial-and-error synthesis in order to
obtain a system with the desired oxidation potential can be
avoided. Furthermore, the level of accuracy can be adjusted to
fit the problem. Very precise values (R2 = 0.997) can be
obtained by calculating IP* but at the expense of extended
computation times (B3LYP/6-31+G* and solvent model). A
rapid estimation can be accomplished with lower accuracy by
simply determining ∑σ+ (R2 = 0.969) or εH (R2 = 0.939 using
B3LYP/6-31G* in vacuum). The application of these
prediction methods suggests that the potential range can be
extended to 1.05 V by introducing nitro groups.
Figure 5. Cyclic voltammograms of mediators 2c (left) and 2a (right):
black lines, 1 mM 2; red dotted line, 1 mM 2 + 20 mM substrate; blue
dash-dotted line, 1 mM 2 + 20 mM substrate +5 mM 2,6-lutidine;
green dashed line, 1 mM 2 + 20 mM substrate + 50 mM 2,6-lutidine;
substrates, p-anisyl alcohol (6; left) and p-anisyl-n-octyl ether (7;
right).
the peak current of the mediator). In both cases, addition of 20
mM substrate to the electrolyte containing 1 mM mediator
leads to low catalytic currents (compare black line to red dotted
line). In addition, oxidation of the mediator becomes
irreversible due to the existence of a chemical reaction that
serves to drain the initial electron transfer equilibrium process.
Assuming that the oxidation pathway illustrated in Scheme 1 is
applicable, the addition of a base should facilitate the chemical
follow-up reaction. In fact, in a previous study, we have shown
that for the oxidation of benzylic substrates with mediators 1
the addition of base is essential; 2,6-lutidine proved to be an
excellent choice.27 In the case of mediators 2, this also proved
to be the case. Upon the addition of 5 mM 2,6-lutidine to
mediator 2c and substrate 6, the current at the peak potential
Finally, by using simple model substrates, we have
demonstrated that 2 can be applied as mediator in indirect
electro-organic synthesis. In future studies, we intend to explore
the substrate spectrum for this promising new class of
mediators and to expand the potential range by synthesis of
new phenanthroimidazole derivatives.
E
ox,1 increases by a factor of 4 compared to the peak current in
ASSOCIATED CONTENT
* Supporting Information
Detailed synthetic procedures, spectral data, description of the
the absence of substrate and base (by a factor of 6 in the case of
2a/7, see blue dash-dotted lines). The catalytic current is
clearly dependent on the concentration of the base because
increasing the concentration of lutidine to 50 mM led to an
■
S
electrochemical methods, supporting figures, computational
433
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
(29) Kwon, J. E.; Park, S.; Park, S. Y. J. Am. Chem. Soc. 2013, 135,
11239−11246.
parameters. This material is available free of charge via the
Internet at http://pubs.acs.org.
(30) Mao, M.; Wang, J.-B.; Xiao, Z.-F.; Dai, S.-Y.; Song, Q.-H. Dyes
Pigm. 2012, 94, 224−232.
AUTHOR INFORMATION
Corresponding Author
little@chem.ucsb.edu
■
(31) Wang, Z.; Lu, P.; Chen, S.; Gao, Z.; Shen, F.; Zhang, W.; Xu, Y.;
Kwok, H. S.; Ma, Y. J. Mater. Chem. 2011, 21, 5451.
(32) Yuan, Y.; Li, D.; Zhang, X.; Zhao, X.; Liu, Y.; Zhang, J.; Wang, Y.
New J. Chem. 2011, 35, 1534.
Notes
(33) Park, S.; Kwon, J. E.; Kim, S. H.; Seo, J.; Chung, K.; Park, S.-Y.;
Jang, D.-J.; Medina, B. M.; Gierschner, J.; Park, S. Y. J. Am. Chem. Soc.
2009, 131, 14043−14049.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
(34) Compound 2b has been reported before: Nielsen, C. B.;
Pittelkow, M.; Sørensen, H. O. Acta Crystallogr. E 2005, 61, 955−956.
However, NMR and MS data have not been published to date.
(35) Boden, B. N.; Jardine, K. J.; Leung, A. C. W.; MacLachlan, M. J.
Org. Lett. 2006, 8, 1855−1858.
We acknowledge partial support from the Center for the
Sustainable Use of Renewable Feedstocks (CenSURF), an NSF
Center for Chemical Innovation (CHE-1240194). R.F. is
particularly grateful to the Alexander von Humboldt
Foundation for granting a Feodor Lynen Fellowship. We are
pleased to acknowledge the contributions of Professor Dean
Tantillo (University of California, Davis, CA) to our initial
application of quantum calculations to the present problem.
(36) Brunner, K.; van Dijken, A.; Borner, H.; Bastiaansen, J. A. M.;
̈
Kiggen, N. M. M.; Langeveld, B. M. W. J. Am. Chem. Soc. 2004, 126,
6035−6042.
(37) Karimi-Jaberi, Z.; Barekat, M. Chin. Chem. Lett. 2010, 21, 1183−
1186.
(38) Steck, E. A.; Day, A. R. J. Am. Chem. Soc. 1943, 65, 452−456.
(39) The fact that 1H-phenanthroimidazoles are easily alkylated at
position 1 might prove useful in other regards. For instance, position 1
could serve as a linking site for modification of an electrode surface
with 2 serving as heterogeneous catalyst.
(40) Pavlishchuk, V. V.; Addison, A. W. Inorg. Chim. Acta 2000, 298,
97−102.
(41) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
Fundamentals and Applications; Wiley: New York, 2001.
(42) Zuman, P. Substituent Effects in Organic Polarography; Plenum
Press: New York, 1967.
REFERENCES
■
(1) Frontana-Uribe, B. A.; Little, R. D.; Ibanez, J. G.; Palma, A.;
Vasquez-Medrano, R. Green Chem. 2010, 12, 2099−2119.
(2) Schafer, H. J.; Harenbrock, M.; Klocke, E.; Plate, M.; Weiper-
̈
Idelmann, A. Pure Appl. Chem. 2007, 79, 2047−2057.
(3) Sperry, J. B.; Wright, D. L. Chem. Soc. Rev. 2006, 35, 605.
(4) Utley, J. H. P.; Folmer Nielsen, M. Electrogenerated Bases. In
Organic Electrochemistry, 4th ed.; Lund, H., Hammerich, O., Eds.; M.
Dekker: New York, 2001; pp 1227−1258.
(5) Moeller, K. D. Tetrahedron 2000, 56, 9527−9554.
(6) Saveant, J.-M. Chem. Rev. 2008, 108, 2348−2378.
́
(7) Steckhan, E. Top. Curr. Chem. 1987, 142, 1−69.
(8) Steckhan, E. Angew. Chem., Int. Ed. Engl. 1986, 25, 683−701.
(9) Simonet, J.; Pilard, J.-F. Electrogenerated Reagents. In Organic
Electrochemistry, 4th ed.; Lund, H., Hammerich, O., Eds.; M. Dekker:
New York, 2001; pp 1163−1225.
(10) Platen, M.; Steckhan, E. Chem. Ber. 1984, 117, 1679−1694.
(11) Saito, K.; Ueoka, K.; Matsumoto, K.; Suga, S.; Nokami, T.;
Yoshida, J.-I. Angew. Chem., Int. Ed. 2011, 50, 5153−5156.
(12) Hosoi, K.; Inagi, S.; Kubo, T.; Fuchigami, T. Chem. Commun.
2011, 47, 8632−8634.
(43) Hicks, L. D.; Fry, A. J.; Kurzweil, V. C. Electrochim. Acta 2004,
50, 1039−1047.
(44) Wu, X.; Davis, A. P.; Lambert, P. C.; Kraig Steffen, L.; Toy, O.;
Fry, A. J. Tetrahedron 2009, 65, 2408−2414.
(45) The Gaussian ’09 package was used for the calculations: Frisch,
M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson,
G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov,
A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.;
Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda,
Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J.
E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.;
Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.;
Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega,
N.; Millam, N. J.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.;
Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.;
Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.;
Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.;
(13) Park, Y. S.; Little, R. D. Electrochim. Acta 2009, 54, 5077−5082.
(14) Park, Y. S.; Little, R. D. J. Org. Chem. 2008, 73, 6807−6815.
(15) Park, Y. S.; Wang, S. C.; Tantillo, D. J.; Little, R. D. J. Org. Chem.
2007, 72, 4351−4357.
(16) Gerken, J. B.; Wang, S. C.; Preciado, A. B.; Park, Y. S.;
Nishiguchi, G.; Tantillo, D. J.; Little, R. D. J. Org. Chem. 2005, 70,
4598−4608 and references cited therein.
(17) Konig, B., Ed. Chemical Photocatalysis; De Gruyter: Berlin, 2013.
̈
̈
Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.;
(18) Xu, K. Chem. Rev. 2004, 104, 4303−4418.
(19) Takahashi, K.; Furusawa, T.; Sawamura, T.; Kuribayashi, S.;
Inagi, S.; Fuchigami, T. Electrochim. Acta 2012, 77, 47−53.
(20) Halas, S. M.; Okyne, K.; Fry, A. J. Electrochim. Acta 2003, 48,
1837−1844.
Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09,
revision D.01; Gaussian, Inc.: Wallingford, CT, 2009.
(46) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B
2009, 113, 6378−6396.
(47) Hubig, S. M; Rathore, R.; Kochi, J. K. J. Am. Chem. Soc. 1999,
121, 617−626.
(48) Due to the low solubility of mediators 2a and 2b in acetonitrile,
the addition of dichloromethane was required (CH3CN/CH2Cl2 =
4:1). Since the oxidation potentials of 2c−2h do not change
significantly upon the addition of dichloromethane to the acetonitrile
solution, we are confident that our approach is still valid.
(49) Szabo, A.; Ostlund, N. S. Modern Quantum Chemistry; Dover
Publications: Mineola, NY, 1996.
(21) Fuchigami, T.; Tetsu, M.; Tajima, T.; Ishii, H. Synlett 2001,
1269−1271.
(22) Wend, R.; Steckhan, E. Electrochim. Acta 1997, 42, 2027−2039.
(23) Dapperheld, S.; Steckhan, E.; Brinkhaus, K. H. G.; Esch, T.
Chem. Ber. 1991, 124, 2557−2567.
(24) Schmidt, W.; Steckhan, E. Chem. Ber. 1980, 113, 577−585.
(25) Yurchenko, O.; Freytag, D.; Zur Borg, L.; Zentel, R.; Heinze, J.;
Ludwigs, S. J. Phys. Chem. B 2012, 116, 30−39.
(26) Seo, E. T.; Nelson, R. F.; Fritsch, J. M.; Marcoux, L. S.; Leedy,
D. W.; Adams, R. N. J. Am. Chem. Soc. 1966, 88, 3498−3503.
(27) Zeng, C.-C.; Zhang, N.-T.; Lam, C. M.; Little, R. D. Org. Lett.
2012, 14, 1314−1317.
(28) Zhang, N.-T.; Zeng, C.-C.; Lam, C. M.; Gbur, R. K.; Little, R. D.
J. Org. Chem. 2013, 78, 2104−2110.
(50) Perdew, J.; Levy, M. Phys. Rev. B 1997, 56, 16021−16028.
(51) Chong, D. P.; Gritsenko, O. V; Baerends, E. J. J. Chem. Phys.
2002, 116, 1760−1772.
(52) In contrast to the Hammett plots, the intercepts of the linear fits
in Figure 4 have no specific physical meaning other than the oxidation
434
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435

Journal of the American Chemical Society
Article
potential of a hypothetical compound with either equal energy of
neutral and charged form (for Eox,1 vs IP*) or with an εH on the
vacuum level (for Eox,1 vs εH). Interestingly, for both methods, the
absolute value of the slope is between 0.55 and 0.59 for the vacuum
calculation and around 0.85 for the solvent-corrected method. Ideally,
a slope of unity would be obtained in a plot of electronvolts versus
volts. Deviations might stem from solvation effects, which is in
agreement with the slope closer to unity when the solvent model is
used. A more detailed discussion can be found in Davis, A.; Fry, A. J. J.
Phys. Chem. A 2010, 114, 12299−12304.
(53) Weinberg, N. L.; Weinberg, H. R. Chem. Rev. 1968, 68, 449−
523.
(54) Miranda, J. A.; Wade, C. J.; Little, R. D. J. Org. Chem. 2005, 70,
8017−8026.
435
dx.doi.org/10.1021/ja410865z | J. Am. Chem. Soc. 2014, 136, 427−435