The role of arene-arene interactions in the folding of ortho-phenylenes

Article abstract of DOI:10.1021/ja4026006

The ortho-phenylenes are a simple class of helical oligomers and representative of the broader class of sterically congested polyphenylenes. Recent work has shown that o-phenylenes fold into well-defined helical conformations (in solution and, typically, in the solid state); however, the specific causes of this folding behavior have not been determined. Here, we report the effect of substituents on the conformational distributions of a series of o-phenylene hexamers. These experiments are complemented by dispersion-corrected DFT calculations on model oligomers (B97-D/TZV(2d,2p)). The results are consistent with a deterministic role for offset arene-arene stacking interactions on the folding behavior. On the basis of the experimental and computational results, we propose a model for o-phenylene folding with two simple rules. (1) Conformers are forbidden if they include a particular sequence of biaryl torsional states that causes excessive steric strain. These ABA states correspond to consecutive dihedral angles of -55 /+130 /-55 (or +55 /-130 /+55). (2) The stability of the remaining conformers is determined by offset arene-arene stacking interactions that are easily estimated as an additive function of the number of well-folded torsional states (±55) along the backbone. For the parent, unsubstituted poly(o-phenylene), each interaction contributes roughly 0.5 kcal/mol to the helix stability (in chloroform), although their strength is sensitive to substituent effects. The behavior of the o-phenylenes as a class is discussed in the context of this model. They are analogous to α-helices, with axial aromatic stacking interactions in place of hydrogen bonding. The model predicts that the overall folding propensity should be quite sensitive to relatively small changes in the strength of the arene-arene stacking. In a broader sense, these results demonstrate that polyphenylenes may exhibit folding behavior that is amenable to simple models, and validate the use of diffusion-corrected DFT methods in predicting their three-dimensional structures.

Full text of DOI:10.1021/ja4026006

Article
pubs.acs.org/JACS
The Role of Arene−Arene Interactions in the Folding
of ortho-Phenylenes
Sanyo M. Mathew,† James T. Engle,‡ Christopher J. Ziegler,‡ and C. Scott Hartley*,†
†Department of Chemistry & Biochemistry, Miami University, Oxford, Ohio 45056, United States
‡Department of Chemistry, University of Akron, Akron, Ohio 44325, United States
S
* Supporting Information
ABSTRACT: The ortho-phenylenes are a simple class of helical
oligomers and representative of the broader class of sterically
congested polyphenylenes. Recent work has shown that
o-phenylenes fold into well-defined helical conformations
(in solution and, typically, in the solid state); however, the specific
causes of this folding behavior have not been determined.
Here, we report the effect of substituents on the conforma-
tional distributions of a series of o-phenylene hexamers. These
experiments are complemented by dispersion-corrected DFT
calculations on model oligomers (B97-D/TZV(2d,2p)). The results are consistent with a deterministic role for offset arene−
arene stacking interactions on the folding behavior. On the basis of the experimental and computational results, we propose a
model for o-phenylene folding with two simple rules. (1) Conformers are forbidden if they include a particular sequence of
biaryl torsional states that causes excessive steric strain. These “ABA” states correspond to consecutive dihedral angles
of −55°/+130°/−55° (or +55°/−130°/+55). (2) The stability of the remaining conformers is determined by offset arene−arene
stacking interactions that are easily estimated as an additive function of the number of well-folded torsional states ( 55°) along
the backbone. For the parent, unsubstituted poly(o-phenylene), each interaction contributes roughly 0.5 kcal/mol to the helix
stability (in chloroform), although their strength is sensitive to substituent effects. The behavior of the o-phenylenes as a class is
discussed in the context of this model. They are analogous to α-helices, with axial aromatic stacking interactions in place of
hydrogen bonding. The model predicts that the overall folding propensity should be quite sensitive to relatively small changes in
the strength of the arene−arene stacking. In a broader sense, these results demonstrate that polyphenylenes may exhibit folding
behavior that is amenable to simple models, and validate the use of diffusion-corrected DFT methods in predicting their three-
dimensional structures.
shown that short o-phenylene oligomers are strongly predis-
posed to folding into well-defined helical conformations in
solution, although, in general, they exhibit conformational
disorder at their ends.24 Likewise, Fukushima and Aida have
reported a remarkable helical o-phenylene oligomer that
spontaneously resolves in the solid state and undergoes a large
decrease of its racemization rate on oxidation.25 More recently,
they have demonstrated a surprising solvent effect whereby
acetonitrile appears to uniquely promote the folding of their
oligomers into “perfect” helices (i.e., without end-group
disorder).26 This phenomenon has been attributed to a steric
effect in a general sense, but to our knowledge, its precise origin is
currently undetermined.
Thus, while not necessarily useful as conjugated polymers in
the conventional sense, the o-phenylenes represent a new class of
helical polymers27,28 and foldamers.29−31 Interest in other,
related sterically congested polyphenylene architectures also
appears to be increasing; for example, Dichtel32 and Manabe33
have recently reported examples of alternating o/p and o/o/p
INTRODUCTION
■
The polyphenylenes,1−3 architectures based on directly con-
nected aromatic rings, are inherently conjugated and chemically
robust;4 consequently, they are an important class of compounds
for applications in materials science and nanotechnology.
The best-known examples of polyphenylenes are para-phenylene
polymers and oligomers,5 which have been used in organic
electronics6 (particularly as emissive materials)7 and as single
molecule wires.8 While p-phenylenes are essentially rigid rods
(and can be very useful as structural units9), the incorporation of
ortho and meta linkages, possibly in combination with branching,
gives polyphenylenes with more complex three-dimensional
structures. Thus, m-phenylenes have been used as helical
polymers10,11 and oligomers12 with applications in molecular
recognition.13,14 Similarly, highly branched polyphenylene
dendrimers15 have been used as platforms for controlled energy
transfer.16,17
The o-phenylenes, one of the most fundamental types of poly-
phenylene, have recently received increasing attention. Building
on the few reports of o-phenylenes prior to 2010,18−20 our own
work has shown that o-phenylenes exhibit weak delocalization
that is quite sensitive to substituent effects.21−23 We have also
Received: March 13, 2013
Published: April 4, 2013
© 2013 American Chemical Society
6714
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
polyphenylenes, respectively. In both cases, twisting from the
ortho linkages is implicated in the adoption of compact
conformations and there is at least the potential for interesting
secondary structures. However, experimental determination of the
precise folding behavior of these compounds is challenging, and
it remains to be determined whether these sorts of motifs adopt
(or can be made to adopt) well-defined three-dimensional structures.
Ultimately, linear (unbranched) polyphenylenes with various
sequences of ortho, meta, and para linkages may, by analogy with
biomolecules, be made to fold into complex tertiary structures.
Further, understanding the conformational preferences of
polyphenylenes may be important to understanding the complex
factors influencing their use34−36 as precursors to nanographenes
via the Scholl reaction. However, our understanding of the
interactions involved in polyphenylene folding is in its infancy:
o-phenylene helices have been demonstrated but not explained,
and the behavior of other types of repetitive linear oligo-
(phenylene) sequences is much less clear. The conformational
analysis of these simple oligomers is deceptively complex,
involving, in principle, both the inherent torsional preferences of
each biaryl bond and a number of different possible face-to-face
and edge-to-face arene−arene interactions. Unfortunately, com-
putational modeling of polyphenylene segments is made more
difficult by the failure of commonly used computational methods
to correctly predict their relative conformational stabilities.22
Here, we report a systematic investigation of substituent
effects on the conformational behavior of o-phenylenes.
Hexamers oP6(X)2 were chosen as targets because they are
sufficiently long that conformational exchange is slow on the
NMR time scale,22 but also sufficiently short that their NMR
spectra can be completely analyzed to provide quantitative and
complete conformational distributions. The substituents (NMe2,
Me, H, F, CF3, CN, NO2) were varied at the oligomer ends both
for synthetic convenience and because these positions were
expected to be more sensitive to substituent effects.23 The
oligomers do indeed exhibit distinct conformational distribu-
tions, which can be rationalized in the context of model systems
for the study of arene−arene interactions. The experimental
results also validate the use of the diffusion-corrected B97-D
DFT method for the prediction of relative conformer stabilities.
By combining the experimental and computational results, we
propose a very simple, semiquantitative model for the folding of
o-phenylenes that clarifies the role of arene−arene interactions in
determining their behavior and rationalizes several of their
reported properties. The model provides guidelines for the
design of new o-phenylenes with high folding propensities. These
results should also be applicable to the analysis of congested
polyphenylenes more generally.
o-phenylenes21,22 and is shown in Scheme 1. Briefly, Suzuki−Miyaura
coupling of known boroxarene 137 and 2-(2-bromophenyl)phenol
a
Scheme 1. Synthesis of oP6(X)2
a
Reagents and conditions: (a) Pd(PPh3)4, Na2CO3(aq), THF/H2O
(4:1), Δ; (b) Tf2O, pyridine, CH2Cl2; (c) arylboronic acid, Pd(OAc)2,
SPhos, K3PO4, THF/H2O (4:1), Δ.
238 gave symmetrical o-phenylene tetramer diol 3, which was
converted to ditriflate 4 in good yield. The target series of
hexamers was then prepared by Suzuki−Miyaura coupling with
the appropriate phenylboronic acid derivatives in good yields.
Full details are given in the Supporting Information. As is
common for o-phenylenes, the NMR spectra of the final
compounds are complex because of slow conformational
exchange; however, they were fully assigned to the proposed
structures as described below. X-ray quality single-crystals were
grown of four of the compounds, oP6(NMe2)2 (DCM/hexanes),
oP6(Me)2 (CDCl3/hexanes), oP6(NO2)2 (EtOH/DCM), and
oP6(CF3)2 (DCM/EtOH). The resulting crystal structures are
shown in Figure 1. All four oligomers crystallize into very similar
Figure 1. ORTEP representations (50% ellipsoid probability) of the
crystal structures of oP6(NO2)2, oP6(CF3)2, oP6(NMe2)2, and
oP6(Me2)2. Hydrogen atoms have been omitted for clarity.
RESULTS AND DISCUSSION
■
Synthesis. Our synthesis of the oP6(X)2 series is a straight-
forward modification of our previously developed approach to
quasi-C2-symmetric helical conformers, as has been observed for
some (but not all) other o-phenylene hexamers.25,26,39 The unit
6715
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
cells of all of the compounds comprise both enantiomeric helices
in all cases.
Conformational Analysis. For simple o-phenylene hexam-
ers, such as oP6(X)2, the backbone conformation is dictated by
the internal dihedral angles ϕ2, ϕ3, and ϕ4, shown in Figure 2.
at room temperature is the compact 31 AAA helix. Some disorder
occurs at the ends of the oligomers; thus, we believe that the
behavior of o-phenylenes can be characterized, in general, as well-
defined helical cores with disorder occurring primarily at the
ends. These conclusions are also consistent with solid-state
structures reported by us21 (see above) and the results of
Fukushima and Aida.25,26
Unfortunately, the specific interactions responsible for the
relative stabilities of the conformers in Figure 3 (or the many
analogous conformers for longer o-phenylenes) are not obvious.
Offset stacking interactions between every third arylene
presumably stabilize the “folded” AAA conformer, but the
dependence of such stacking interactions and various other (e.g.,
edge-to-face) interactions on the backbone geometry is not clear.
The challenge of analyzing the folding properties of o-phenylenes
as a class is exacerbated by the exponential increase in the
number of possible conformers with increasing length. Further,
although DFT geometry optimizations of the oligomers are
straightforward, standard DFT methods (e.g., B3LYP) do not
correctly predict the relative stabilities of the conformations,
incorrectly favoring the B states. This failure is not surprising, as
these methods are not expected to predict the strength of arene−
arene interactions (and other dispersion-based interac-
tions),41−43 which, regardless of the particular conformer
under consideration, are likely to be very important.
Figure 2. Key dihedral angles in the folding of oP6(X)2.
Within a single o-phenylene strand, each of these can assume one
of two possible values: ϕi ≈ −55° or ϕi ≈ +130°. We refer to
these as the “A” and “B” states, respectively. Thus, the AAA
conformer, with ϕ2 ≈ ϕ3 ≈ ϕ4 ≈ −55°,40 is a compact,
left-handed helix with every third ring stacked. All of the solid-
state structures shown in Figure 1 correspond to the AAA
conformer. There are six, and only six, possible backbone
configurations for an o-phenylene hexamer: four that are
C2-symmetric (AAA, BAB, BBB, ABA) and two that are
unsymmetric (AAB and ABB), as shown in Figure 3 for
The conformational behavior of the oP6(X)2 series in CDCl3
was probed by NMR spectroscopy at 277 K using our previously
reported strategy.24 As expected, the 1D 1H NMR spectra of all
of the hexamers are complex, comprising contributions from
three slowly exchanging conformational states we label I, II, and
III in order of decreasing signal intensity. The 1H NMR
spectrum of oP6(NO2)2 is shown in Figure 4, with the others
Figure 3. Backbone conformers available to oP6(NO2)2. The
geometries were optimized at the B97-D/TZV(2d,2p) level.
oP6(NO2)2. Of course, there is a degenerate, enantiomeric
counterpart for each of these states with A′ = +55° and
B′ = −130°. In this study, none of the compounds have been resolved
and thus all observed conformers are racemates. Note, however, that
the two enantiomeric pools are distinct:24 a single molecule can have
A/B states or A′/B′ states but no A/A′ and B/B′ mixtures.
Conformational exchange between different backbone conformers
(e.g., AAA ⇄ AAB ⇄ etc.), considered in this study, is faster
than, and can be considered independently of, racemization
(AAA/AAB/etc. ⇄ A′A′A′/A′A′B′/etc.). A more thorough
discussion of these issues was given in an earlier paper.24
We have previously shown that, for o-phenylene oligomers
longer than the pentamer, the backbone conformational states
are in slow exchange on the NMR time scale at room
temperature and below. Thus, deconvolution and integration
Figure 4. 1D 1H NMR (top) and EXSY spectra of oP6(NO2)2 (CDCl3,
277 K).
1
of H NMR spectra can be used to determine conformational
distributions. Simple o-phenylenes, such as the parent series,22
provided in the Supporting Information. For each of the
compounds, COSY, HMQC, HMBC, and EXSY spectra were
show a preference for the A states: the most populated conformer
6716
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
obtained. The EXSY spectra, also shown in Figure 4 for
oP6(NO2)2, confirm, in all cases, that the purity of the
compounds is quite high: the minor signals show cross-peaks
with corresponding signals from conformer I, indicating that they
represent conformers in slow exchange, not impurities. Chemical
shift assignments (1H and 13C) for conformer I were made using
the COSY, HMQC, and HMBC spectra. Chemical shift
assignments for the minor conformers, II and III, were then
obtained using the EXSY (and COSY) spectra to map the
assignments from I. Chemical shift assignments for the complete
oP6(X)2 series in all three observed conformational states are
given in the Supporting Information. On the basis of the number
of signals observed in the spectra, it is readily determined that
conformers I and III represent 2-fold-symmetric geometries,
whereas conformer II is unsymmetrical.
absence of stabilizing effects such as arene−arene stacking,
ΔG°fold would be +0.38 kcal/mol (−RT ln(1/2)) because of the
reduced symmetry of the AAB conformer (i.e., both ends of the
oligomer are equivalent; therefore, statistically, a 2:1 AAB:AAA
ratio would be expected). Thus, in all cases, the AAA conformer
is stabilized by 0.3−1.1 kcal/mol (at 277 K) relative to a
hypothetical unbiased freely mobile oligomer.
The ΔG°fold values show a good linear dependence on the
Hammett constants σm for the terminal substituents,50 as shown
in Figure 5, with r = −0.96, representing a strong, statistically
To assign specific geometries to I, II, and III, we compared the
experimental chemical shift assignments to those predicted for all
six backbone configurations possible for o-phenylene hexamers
(i.e., Figure 2). For the geometry optimizations, we used the
B97-D/TZV(2d,2p) method,44 a dispersion-corrected DFT
method that is now commonly used to quantitatively investigate
arene−arene interactions, including substituent effects.45−47
Gratifyingly, unlike the other DFT methods we have previously
employed, the B97-D method correctly predicts the order of
stability of the previously observed conformers of oP6(H)2 (AAA
> AAB > BAB). Relative stabilities of the conformers for each of
the oP6(X)2 series are given in Figure S1. 1H chemical shifts were
calculated for each geometry at the PCM/WP04/6-31G(d)
level.48,49 All three conformers for all members of the oP6(X)2
series could be unambiguously matched to one of the optimized
geometries: in all cases, the match had the lowest RMS error
when compared to the experimental chemical shift data, and the
RMS errors were always uniquely below our previously
determined standard of 0.15 ppm for o-phenylenes using this
method. Consistent with our previous results, in all cases
conformers I correspond to the AAA geometries, conformers II
correspond to the AAB geometries, and conformers III
correspond to the BAB geometries. Relative conformer
populations were then obtained from the integration of the 1H
NMR spectra through deconvolution of relatively uncluttered
regions, and are compiled in Table 1.
Figure 5. ΔG°fold plotted against σm for oP6(X)2 (CDCl3, 277 K).
significant correlation (p = 0.0005). As discussed in more detail
below, the behavior of these short oligomers is reminiscent of
molecular torsion balances used to study quasi-intermolecular
interactions (including arene−arene interactions).51,52 The
quality of the correlation between ΔG°fold and σm is typical of
similar systems which were designed explicitly to quantify
arene−arene interactions, including face-to-face stacking
measured in 1,8-diarylnaphthalenes53 or chemical double
mutant cycles54,55 and related systems examining edge-to-face
interactions.56
The experimental ΔG°fold data was then compared to the
B97-D/TZV(2d,2p) energy differences ΔEfold between the AAB and
AAA geometries, as shown in Figure 6. Of course, our goal here is
Table 1. Equilibrium Conformer Populations and ΔG°fold
Values for Hexamers oP6(X)2 in CDCl3 at 277 K
a
X
AAA
AAB
BAB
ΔG°fold (kcal/mol)
NO2
CN
CF3
F
74.3
74.4
74.6
56.6
49.0
41.0
44.1
20.6
18.2
21.0
33.9
41.7
49.0
47.4
5.2
7.4
4.4
9.5
9.3
8.9
8.4
−0.71 0.04
−0.77 0.04
−0.70 0.04
−0.28 0.04
−0.09 0.04
+0.10 0.04
+0.04 0.04
H
Me
N(Me)2
a
Errors calculated assuming a 5% error on integration of the 1H NMR
spectra.
Figure 6. Experimental ΔG°fold plotted against computational ΔEfold
.
The populations of each conformer are highly substituent-
not quantitative prediction of the experimental results.
Considering the relatively small substituent effect (ΔΔG°fold
0.9 kcal/mol), and that solvation and ΔS° are not at all treated by
these (gas-phase) calculations, the calculated stabilities do a good
job of accounting for the substituent effects in the series, with a
correlation coefficient of r = 0.94 (p = 0.0016). The roughly
1.5−2 kcal/mol difference between ΔEfold and ΔG°fold is
dependent, with, in general, an increasing preference for the AAA
conformer when the oligomers are functionalized with electron-
withdrawing substituents (NO2, CN, CF3). We chose to focus on
the folding reaction AAB ⇄ AAA with a free energy change
ΔG°fold, as the populations of these two conformers were
typically large enough to allow quantitative integration. In the
<
6717
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
consistent with previous efforts to compare B97-D calculations
to experimental models of aromatic stacking in chloroform.47,57
The difference likely comprises 0.4 kcal/mol for the entropic bias
toward AAB (see above) and 1−1.5 kcal/mol for changes in
solvation (and other factors). Since the AAB geometry should
expose more molecular surface to the solvent, a change in
solvation energy of 1−1.5 kcal/mol would appear to be
reasonable given that the free energy of solvation of benzene in
chloroform is −4.6 kcal/mol.58 Unfortunately, the limited
solubility of the oP6(X)2 compounds in polar solvents precluded
the investigation of their conformational behavior in other media.
However, we did examine the behavior of oP6(H)2 in toluene-d8;
the conformational distribution is the same to that in CDCl3
within experimental error (53:40:8 AAA:AAB:BAB). The most
striking difference between the calculated energies and the
observed experimental free energies of folding is that the
calculations predict that all substituents should have a stabilizing
effect relative to the unsubstituted case oP6(H)2, whereas the
electron donating substituents (Me, NMe2) were observed
experimentally to decrease the population of the helical
conformer. This discrepancy is, however, too small for us to
say whether it derives from a genuine failure by the calcula-
tions or an unaccounted-for substituent effect (on solvation,
for example).
(ΔG°fold(oP6(CN)2) − ΔG°fold(oP6(Me)2)) is in very good
agreement with the magnitude of substituent effects in closely
related systems.54,55,60 Again, as these other systems were
explicitly designed to examine individual arene−arene stacking
forces, this similarity suggests that the difference between the
AAB and AAA conformers is determined by a single such
interaction.
Third, further evidence for the deterministic role of arene
stacking in these systems comes from the B97-D DFT
calculations. Superficially, the simple observation that this
method correctly predicts the relative stabilities of the con-
formers, whereas more common methods such as B3LYP do not,
implicates an important role for dispersion interactions in these
systems. The influence of the substituents on rotation about the
biaryl bonds, another possible effect, does not explain the
observed substituent effect, as substituents should have very little
effect on the preferred dihedral angles ϕ2/ϕ4 (see Supporting
Information, Figure S2). More significantly, the ΔEfold values for
the AAB ⇄ AAA equilibrium are correlated to a single geometric
variable from just the AAA conformer, the centroid−centroid
stacking distance (RCn−Cn) corresponding to the offset stacking
interaction at the oligomer terminus, as shown in Figure 7
Arene−Arene Interactions and o-Phenylene Folding.
On the basis of these results, we propose that the determining
factor controlling conformational behavior in the o-phenylenes is
offset aromatic interactions directed along the helical axis.
In other words, o-phenylenes can be considered as analogues of
α-helices, with hydrogen bonding replaced by axial arene−arene
stacking. This assertion does not necessarily mean that other
interactions are insignificant (e.g., edge-to-face interactions in the
BAB conformer); rather, we believe that they are not the
deciding factor in determining the conformational distribution.
o-Phenylenes are therefore similar to related foldamers exploiting
aromatic stacking, such as Iverson’s aedamers.59 This assertion is
supported by several pieces of experimental and computational
evidence.
Figure 7. ΔEfold plotted against the centroid−centroid distance RCn−Cn
,
First, the experimental ΔG°fold values for the oP6(X)2 series
are consistent with a single offset aromatic stacking interaction as
the differentiating factor between the AAA and AAB conformers.
Accounting for symmetry, the AAA conformer is stabilized
0.3−1.1 kcal/mol relative to AAB, depending on the substituent.
These values are in good agreement with the strength of a single
aromatic stacking interaction in chloroform, as measured by
Hunter using chemical double mutant cycles,54,55 Gung using
molecular torsion balances,60 and Diederich using the self-
assembly of adenine-based receptors.61 Interestingly, the overall
strength of the arene−arene interactions is also directly
comparable to the contribution of a phenylalanine−phenyl-
alanine aromatic interaction to the stability of α-helices in
peptides.62
depicted as the dashed line in the model of oP6(NO2)2.
(r = 0.97). Put another way, the overall conformational stability
of the perfectly folded AAA conformer is highly correlated to the
separation between the rings at the oligomer ends, which should
reflect the strength of the arene−arene stacking interaction.
Generalization of these results from the oligomer termini to
the o-phenylenes as a class may be done by examining the
complete conformational space for the oligomers. Thus, the
relative energies ΔErel of all six possible conformers for oP6(H)2
(i.e., Figure 3) were compared, taking the most stable AAA state
as the reference. Of the six, the ABA conformer is exceptionally
unstable (17.5 kcal/mol less stable than AAA): this particular
sequence of dihedrals forces the two terminal rings into each
other, imparting significant steric strain. The relative energies of
the remaining conformers (AAA, AAB, BAB, ABB, BBB) have a
simple relationship to the backbone geometry. As shown in
Figure 8, ΔErel shows a good linear dependence on the number of
B states in each oligomer (r = 0.97), which is easily rationalized as
each B state defect would be expected to eliminate exactly one
stacking interaction. The slope of the plot, 2.1 0.2 kcal/mol per
B state, confirms that each B state breaks one stacking
interaction, as it is a good match for the reference value of
2.75 kcal/mol for the breaking of a parallel-displaced benzene
dimer at the B97-D/TZV(2d,2p) level.44 The difference likely
Second, the variation of of ΔG°fold with σm is consistent with a
single aromatic stacking interaction at the oligomer ends. The
overall trend, increasing interaction (more negative ΔG°fold
)
with increasing electron-withdrawing power of the substituent, is
exactly as expected for aromatic stacking interactions on the
basis of the popular Hunter−Sanders63 and Cozzi−Siegel53,64,65
models and is broadly consistent with gas-phase computa-
tional studies.45,66−68 This trend has been observed in
many model systems designed explicitly to probe these
effects.54,55,60,69 Moreover, the overall substituent effect of
roughly 1 kcal/mol for the substituents considered here
6718
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
results from geometric constraints placed on the interaction by
the o-phenylene backbone.
This analysis is readily extended to longer o-phenylenes. In
Figure 8, we also show the relative energies of the conformers of
confirmed in oligomers up to the dodecamer;24 similar behavior
has been reported by Fukushima and Aida.26 The reason for this
“frayed ends” behavior of the o-phenylenes is easily rationalized.
The two rules described above require that the second-most
stable conformation of an o-phenylene have a single B state
(maximizing stacking interactions via rule 2) and that this B
configuration be located at the end, since internal isolated B
defects require ABA sequences which are forbidden (rule 1).
The overall conformer population will favor well-defined
helical folding so long as the number of backbone conformers
with two or more B states does not entropically swamp the
enthalpic preference for helical folding. To examine the overall
folding behavior of the o-phenylenes in more detail, we estimated
the equilibrium fraction of “well-folded” oligomers as a function
of oligomer length (n) and arene−arene interaction energy
assuming the simple model above. We define “well-folded” to
include the perfectly folded conformer (e.g., AAAAA) as well as
those with defects only at the very ends (e.g., AAAAB and
BAAAB). Thus, there are three “well-folded” conformers for each
oligomer regardless of n (n ≥ 6). Every possible backbone
configuration for oligomers up to the [30]-mer was considered
explicitly using a simple computer algorithm (see Supporting
Information). The results are shown in Figure 9.
Figure 8. Relative conformer energies ΔErel for oP6(H)2 (black, left)
and oP8(H)2 (blue, right) plotted against the number of B states. The
ΔErel scales are identical, but the data for oP6(H)2 are offset by +4 kcal/
mol for clarity.
oP8(H)2. All possible conformers that do not have ABA
sequences are included. A good linear dependence on the
number of B states is again observed. The linear fit is parallel to
that for the oP6(H)2 conformers, at 2.2 0.4 kcal/mol per B
state, again suggesting that each B state defect breaks one
stacking interaction relative to the AAAAA conformer. Of course,
on the basis of the previous comparison of ΔG°fold and ΔEfold, it is
clear that these gas-phase enthalpy calculations overestimate the
actual preference for the helical conformer, presumably because
they omit the effect of solvent.
Taken together, these observations suggest that the conforma-
tional behavior of o-phenylenes can be understood using a very
simple model with two rules. (1) ABA sequences are prohibited
as they require the backbone to contort into a structure with high
steric strain. (2) The relative energies of the remaining
conformers are determined by offset arene−arene stacking
interactions that are a simple additive function of the number of
A states along the backbone.
Our experimental ΔG°fold values suggest that these inter-
actions are stabilizing by roughly 0.5 kcal/mol for the parent
(unsubstituted) o-phenylenes but are sensitive to substituent
effects. This model reproduces the observed complete conforma-
tional distribution of oP6(H)2 very well: a stacking interaction of
−0.47 kcal/mol predicts a conformer distribution at 277 K of
49:42:9 for AAA:AAB:BAB, a perfect match for the exper-
imentally observed population.70
While this simple model accounts for the preference for helical
conformations in the o-phenylenes in a general sense, it can be
used to treat specific aspects of their folding behavior. For
example, we have previously shown that the conformational
population of long o-phenylenes favors the compact helix and a
minor “frayed ends” state in which one of the termini is flipped
from the helical path. Put another way, for an o-phenylene
[n]-mer, the An−1 conformer (AA...AA) is the major conformer
and the An−2B conformer (AA...AB) is always the next most
populated. For one series of oligomers, this behavior has been
Figure 9. Equilibrium fraction of total population in well-folded
conformations as a function of oligomer length up to the [30]-mer,
shown for different values of arene−arene stacking stabilization (0.0, 0.5,
1.0, 1.5, and 2.0 kcal/mol).
This model is, of course, idealized, but makes some useful
predictions with respect to o-phenylene folding. In the absence of
any stabilization by arene−arene interactions (black curve in
Figure 9), the occurrence of well-folded conformers is a function
of simple statistics, and since the number of backbone
configurations increases exponentially with n, their fraction of
the total population decreases rapidly with increasing length.
However, interactions with strengths typical of arene−arene
stacking are expected to have a significant impact on the overall
distribution. Arene−arene stabilization of 0.5 kcal/mol per
interaction (blue curve), as expected for the parent series of
oligomers, shifts the population toward the well-folded states,
although the folded population does still decrease significantly
with length. Therefore, the parent poly(o-phenylene) should
exist as short helical segments interrupted by many defect states.
An increase in the stabilization energy to 1.0 kcal/mol per
interaction (red curve), accessible by strengthening the aromatic
interactions through simple substitution with electron-
withdrawing groups (Figure 5), should have a substantial effect
on the conformational population and give oligomers with much
better folding properties. A further increase in stabilization to
6719
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
1.5−2.0 kcal/mol per interaction, attainable through non-
covalent interactions such as hydrogen bonding, would yield
oligomers and polymers with a very high fraction of perfect
folding. This sensitivity of the conformational behavior may also,
in part, explain the observation by Fukushima and Aida of
solvent-dependent “perfect” folding for their oligomers,26 since
significant changes in solvation energy could have a dramatic
effect on the folding propensity.
This model has implications for the mechanism of o-phenylene
racemization, a topic of interest given the remarkable behavior of
the resolved o-phenylenes reported by Fukushima and Aida.25,26
We assume here that racemization occurs via the B states (for
which a single, facile torsional motion converts to the inverted B′
state). Since isolated B states in the center of the chain are
forbidden, defects may originate at the ends of the chain and
effectively nucleate the formation of additional defect states
along the backbone. Interestingly, Fukushima and Aida have
measured free energies of activation for racemization ΔG⧧rac for
relative to the BBB state. However, the perfectly folded AAA
conformer is substantially more stable for oP6(FA), suggesting
that this series is much more prone to well-defined folding at the
termini compared to other o-phenylenes. At first glance, this
stabilization of the AAA conformer would appear to contradict
our model of additive arene−arene stacking interactions.
However, the reason for the enhanced stability of the oP6(FA)
AAA state is the asymmetry of the terminal arenes: the most
stable conformer, shown in Figure 10, is that for which these
groups are rotated such that the methoxy groups are out of
alignment with those in the center of the oligomer. Inspection of
the electrostatic potential map for veratrole (Figure S3) does not
provide an obvious explanation for this preference in terms of the
electronic distribution of the π-system in the repeat units; it is
most easily rationalized in the context of direct repulsive
interactions between the methoxy groups,45 which are much
farther apart in the most stable conformer. While this result for
oP6(FA) is purely computational, it may explain, in a general
sense, the many exceptional properties Fukushima, Aida, and co-
workers have observed for various veratrole-derived o-phenyl-
enes: these heavily substituted oligomers may represent a
privileged class of o-phenylenes with exceptional folding
properties at their ends, which may be key sites for controlling
racemization mechanisms (see above). This result also under-
scores the potential role of substituent effects in modulating the
aromatic stacking interactions in polyphenylenes as a class and
o-phenylenes in particular.
one of their series of o-phenylenes. The ΔG⧧ values increase
rac
linearly with increasing oligomer length up to the octamer, with
an increment of 0.45 kcal/mol per arylene unit. This value is in
good agreement with the model, assuming that the transition
state for racemization of o-phenylene oligomers essentially
involves simultaneous breaking of all the stabilizing arene−arene
interactions.
Given the success of the B97-D calculations of reproducing the
conformational behavior of the oP6(X)2 series, we were
interested in comparing the parent o-phenylenes to those
examined by Fukushima and Aida,25,26 which typically have been
based on veratrole repeat units. Thus, we carried out geometry
optimizations of a representative hexamer oP6(FA).71 Because
rotation about the terminal biaryl bonds in this series is not
degenerate, all possible orientations of the end groups for the
AAA, AAB, BAB, ABB, and BBB conformers were optimized
explicitly (see the Supporting Information).72 Energies of the
individual backbone configurations were obtained by taking the
Boltzmann averages. The conformer stabilities were then
compared to those for the parent hexamer oP6(H)2, as shown
in Figure 10. For the purposes of this comparison, the BBB
CONCLUSIONS
■
In summary, a series of o-phenylene hexamers with varying
substitution at their ends has been synthesized. The substituent
effect on the conformational distributions of these oligomers
demonstrates that the key determinant of the folding properties
of o-phenylenes is offset arene−arene stacking interactions
between every third repeat unit. The B97-D/TZV(2d,2p) DFT
method accurately predicts these substituent effects. Analysis of
different model systems suggests that the relative stabilities of
different o-phenylene conformers are predicted by two simple
rules: (1) ABA sequences are disallowed because they contort
the backbone into a highly sterically strained conformation, and
(2) arene−arene stacking interactions are essentially an additive
function of the number of A states. For the parent oligomer, the
net stabilization of each stacking interaction is approximately
0.5 kcal/mol, a value derived from the observed conformational
population of the hexamer and in excellent agreement with
expectations for aromatic stacking interactions in chloroform.
This model rationalizes the observed behavior of current
o-phenylene oligomers, and predicts that o-phenylene folding
should be very sensitive to relatively small perturbations of the
strengths of the arene−arene interactions. If the o-phenylenes are
taken as representatives of polyphenylenes in general, this study
demonstrates that the folding behavior of polyphenylenes,
despite apparent complexity, may be amenable to very simple
Figure 10. Left: conformer energies of oP6(FA) compared to oP6(H)2,
relative to the BBB state. Right: geometry of the lowest-energy AAA
conformer of oP6(FA).
conformer (i.e., the least stable), was chosen as a convenient
reference state as it should be essentially free of stacking
interactions in either case.
The calculated energies reveal a substantial difference in
behavior between the two prototypical o-phenylenes. Most of the
conformations (ABB, BAB, AAB) have very similar stabilities
6720
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
(18) Wittig, G.; Lehmann, G. Chem. Ber. 1957, 90, 875−892.
models. Further, modern, diffusion-corrected DFT methods,
such as B97-D/TZV(2d,2p), do an excellent job of predicting the
relative stability of polyphenylene conformers. Both of these
results may be of use in designing polyphenylenes with functional
tertiary structures.
(19) Ozasa, S.; Fujioka, Y.; Fujiwara, M.; Ibuki, E. Chem. Pharm. Bull.
1980, 28, 3210−3222.
(20) Kovacic, P.; Uchic, J. T.; Hsu, L.-C. J. Polym. Sci., Part A: Polym.
Chem. 1967, 5, 945−964.
(21) He, J.; Crase, J. L.; Wadumethrige, S. H.; Thakur, K.; Dai, L.; Zou,
S.; Rathore, R.; Hartley, C. S. J. Am. Chem. Soc. 2010, 132, 13848−
13857.
(22) Mathew, S. M.; Hartley, C. S. Macromolecules 2011, 44, 8425−
8432.
(23) He, J.; Mathew, S. M.; Cornett, S. D.; Grundy, S. C.; Hartley, C. S.
Org. Biomol. Chem. 2012, 10, 3398−3405.
ASSOCIATED CONTENT
* Supporting Information
■
S
Supplemental figures referred to in the text; crystal data for
oP6(Me)2, oP6(NMe2)2, oP6(NO2)2, and oP6(CF3)2; NMR
spectra at 277 K for oP6(X)2 with chemical shift assignments;
experimental procedures; computational energies and optimized
geometries. This material is available free of charge via the
Internet at http://pubs.acs.org.
(24) Hartley, C. S.; He, J. J. Org. Chem. 2010, 75, 8627−8636.
(25) Ohta, E.; Sato, H.; Ando, S.; Kosaka, A.; Fukushima, T.;
Hashizume, D.; Yamasaki, M.; Hasegawa, K.; Muraoka, A.; Ushiyama,
H.; Yamashita, K.; Aida, T. Nat. Chem. 2011, 3, 68−73.
(26) Ando, S.; Ohta, E.; Kosaka, A.; Hashizume, D.; Koshino, H.;
Fukushima, T.; Aida, T. J. Am. Chem. Soc. 2012, 134, 11084−11087.
(27) Nakano, T.; Okamoto, Y. Chem. Rev. 2001, 101, 4013−4038.
(28) Yashima, E.; Maeda, K.; Iida, H.; Furusho, Y.; Nagai, K. Chem. Rev.
2009, 109, 6102−6211.
AUTHOR INFORMATION
Corresponding Author
scott.hartley@miamioh.edu
■
Notes
(29) Gellman, S. H. Acc. Chem. Res. 1998, 31, 173−180.
(30) Hill, D. J.; Mio, M. J.; Prince, R. B.; Hughes, T. S.; Moore, J. S.
Chem. Rev. 2001, 101, 3893−4012.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
(31) Guichard, G.; Huc, I. Chem. Commun. 2011, 47, 5933−5941.
(32) Arslan, H.; Saathoff, J. D.; Bunck, D. N.; Clancy, P.; Dichtel, W. R.
Angew. Chem., Int. Ed. 2012, 51, 12051−12054.
C.S.H. and S.M.M. acknowledge support by the National Science
Foundation (CHE-0910477). Gaussian 09 was purchased with
the assistance of the Air Force Office of Scientific Research
(FA9550-10-1-0377). We thank Prof. Steven Wheeler for help
obtaining the TZV(2d,2p) basis set. C.J.Z. acknowledges the
National Science Foundation (CHE-0840446) for funds used to
purchase the Bruker ApexII Duo CCD X-ray diffractometer.
(33) Manabe, K.; Kimura, T. Org. Lett. 2013, 15, 374−377.
(34) Chen, L.; Hernandez, Y.; Feng, X.; Mullen, K. Angew. Chem., Int.
̈
Ed. 2012, 51, 7640−7654.
(35) Yan, X.; Li, L.-S. J. Mater. Chem. 2011, 21, 3295−3300.
(36) Pradhan, A.; Dechambenoit, P.; Bock, H.; Durola, F. J. Org. Chem.
2013, 78, 2266−2274.
(37) Zhou, Q. J.; Worm, K.; Dolle, R. E. J. Org. Chem. 2004, 69, 5147−
5149.
REFERENCES
■
(1) Berresheim, A. J.; Muller, M.; Mullen, K. Chem. Rev. 1999, 99,
̈
̈
(38) Shintani, R.; Maciver, E. E.; Tamakuni, F.; Hayashi, T. J. Am.
1747−1786.
Chem. Soc. 2012, 134, 16955−16958.
(2) Schmaltz, B.; Weil, T.; Mullen, K. Adv. Mater. 2009, 21, 1067−
̈
(39) Blake, A.; Cooke, P.; Doyle, K.; Gair, S.; Simpkins, N. Tetrahedron
Lett. 1998, 39, 9093−9096.
1078.
(3) Omachi, H.; Segawa, Y.; Itami, K. Acc. Chem. Res. 2012, 45, 1378−
1389.
(40) In previous publications, we assigned a value of 70° to the A
state. Here, we revise this value to 55° in light of the new
computational optimized geometries and crystal structures.
(4) Kandre, R.; Feldman, K.; Meijer, H. E. H.; Smith, P.; Schluter, A. D.
̈
Angew. Chem., Int. Ed. 2007, 46, 4956−4959.
(5) Banerjee, M.; Shukla, R.; Rathore, R. J. Am. Chem. Soc. 2009, 131,
1780−1786.
́
(41) Kristyan, S.; Pulay, P. Chem. Phys. Lett. 1994, 229, 175−180.
(42) Johnson, E. R.; Wolkow, R. A.; DiLabio, G. A. Chem. Phys. Lett.
2004, 394, 334−338.
(6) Ivory, D. M.; Miller, G. G.; Sowa, J. M.; Shacklette, L. W.; Chance,
R. R.; Baughman, R. H. J. Chem. Phys. 1979, 71, 1506−1507.
(7) Grimsdale, A. C.; Leok Chan, K.; Martin, R. E.; Jokisz, P. G.;
Holmes, A. B. Chem. Rev. 2009, 109, 897−1091.
̌
(43) Hobza, P.; Sponer, J.; Reschel, T. J. Comput. Chem. 1995, 16,
1315−1325.
(44) Grimme, S. J. Comput. Chem. 2006, 27, 1787−1799.
(45) Wheeler, S. E. J. Am. Chem. Soc. 2011, 133, 10262−10274.
(8) Weiss, E. A.; Ahrens, M. J.; Sinks, L. E.; Gusev, A. V.; Ratner, M. A.;
Wasielewski, M. R. J. Am. Chem. Soc. 2004, 126, 5577−5584.
́ ̀
(46) Vazquez-Mayagoitia, A.; Sherrill, C. D.; Apra, E.; Sumpter, B. G. J.
(9) Bhosale, S.; Sisson, A. L.; Talukdar, P.; Furstenberg, A.; Banerji, N.;
̈
Chem. Theory Comput. 2010, 6, 727−734.
Vauthey, E.; Bollot, G.; Mareda, J.; Roger, C.; Wurthner, F.; Sakai, N.;
̈
̈
(47) Grimme, S.; Antony, J.; Schwabe, T.; Muck-Lichtenfeld, C. Org.
̈
Matile, S. Science 2006, 313, 84−86.
(10) Reddinger, J. L.; Reynolds, J. R. Macromolecules 1997, 30, 479−
481.
Biomol. Chem. 2007, 5, 741−758.
(48) Wiitala, K. W.; Hoye, T. R.; Cramer, C. J. J. Chem. Theory Comput.
2006, 2, 1085−1092.
(11) Ben, T.; Goto, H.; Miwa, K.; Goto, H.; Morino, K.; Furusho, Y.;
Yashima, E. Macromolecules 2008, 41, 4506−4509.
(12) Goto, H.; Katagiri, H.; Furusho, Y.; Yashima, E. J. Am. Chem. Soc.
2006, 128, 7176−7178.
(49) Jain, R.; Bally, T.; Rablen, P. R. J. Org. Chem. 2009, 74, 4017−
4023.
(50) Exner, O. In Correlation Analysis in Chemistry: Recent Advances;
Chapman, N. B., Shorter, J., Eds.; Plenum Press: New York, 1978; pp
439−540.
(13) Goto, H.; Furusho, Y.; Yashima, E. J. Am. Chem. Soc. 2007, 129,
9168−9174.
(51) Salonen, L. M.; Ellermann, M.; Diederich, F. Angew. Chem., Int.
Ed. 2011, 50, 4808−4842.
(14) Miwa, K.; Furusho, Y.; Yashima, E. Nat. Chem. 2010, 2, 444−449.
(15) Bauer, R. E.; Grimsdale, A. C.; Mullen, K. Top. Curr. Chem. 2005,
̈
(52) Mati, I. K.; Cockroft, S. L. Chem. Soc. Rev. 2010, 39, 4195.
(53) Cozzi, F.; Cinquini, M.; Annuziata, R.; Siegel, J. S. J. Am. Chem.
Soc. 1993, 115, 5330−5331.
245, 253−286.
(16) Weil, T.; Reuther, E.; Mullen, K. Angew. Chem., Int. Ed. 2002, 41,
̈
1900−1904.
(54) Cockroft, S. L.; Hunter, C. A.; Lawson, K. R.; Perkins, J.; Urch, C.
J. J. Am. Chem. Soc. 2005, 127, 8594−8595.
(17) Nguyen, T.-T.-T.; Turp, D.; Wang, D.; Nolscher, B.; Laquai, F.;
Mullen, K. J. Am. Chem. Soc. 2011, 133, 11194−11204.
̈
̈
̈
6721
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722

Journal of the American Chemical Society
Article
(55) Cockroft, S. L.; Perkins, J.; Zonta, C.; Adams, H.; Spey, S. E.; Low,
C. M. R.; Vinter, J. G.; Lawson, K. R.; Urch, C. J.; Hunter, C. A. Org.
Biomol. Chem. 2007, 5, 1062−1080.
(56) Fischer, F. R.; Schweizer, W. B.; Diederich, F. Chem. Commun.
2008, 4031−4033.
(57) Graton, J.; Le Questel, J.-Y.; Legouin, B.; Uriac, P.; van de Weghe,
P.; Jacquemin, D. Chem. Phys. Lett. 2012, 522, 11−16.
(58) Hawkins, G. D.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B
1998, 102, 3257−3271.
(59) Lokey, R. S.; Iverson, B. L. Nature 1995, 375, 303−305.
(60) Gung, B. W.; Xue, X.; Reich, H. J. J. Org. Chem. 2005, 70, 3641−
3644.
(61) Faraoni, R.; Blanzat, M.; Kubicek, S.; Braun, C.; Schweizer, W. B.;
Gramlich, V.; Diederich, F. Org. Biomol. Chem. 2004, 2, 1962−1964.
(62) Butterfield, S. M.; Patel, P. R.; Waters, M. L. J. Am. Chem. Soc.
2002, 124, 9751−9755.
(63) Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990, 112,
5525−5534.
(64) Cozzi, F.; Cinquini, M.; Annunziata, R.; Dwyer, T.; Siegel, J. S. J.
Am. Chem. Soc. 1992, 114, 5729−5733.
(65) Cozzi, F.; Ponzini, F.; Annunziata, R.; Cinquini, M.; Siegel, J. S.
Angew. Chem., Int. Ed. Engl. 1995, 34, 1019−1020.
(66) Wheeler, S. E.; Houk, K. N. J. Am. Chem. Soc. 2008, 130, 10854−
10855.
(67) Watt, M.; Hardebeck, L. K. E.; Kirkpatrick, C. C.; Lewis, M. J. Am.
Chem. Soc. 2011, 133, 3854−3862.
(68) Sinnokrot, M. O.; Sherrill, C. D. J. Am. Chem. Soc. 2004, 126,
7690−7697.
(69) Cozzi, F.; Annunziata, R.; Benaglia, M.; Cinquini, M.; Raimondi,
L.; Baldridge, K. K.; Siegel, J. S. Org. Biomol. Chem. 2003, 1, 157−162.
(70) This population is derived from the following: ΔG°1 = −0.47 +
0.38 = −0.09 kcal/mol for the equilibrium AAB ⇄ AAA and ΔG°2 =
−0.47 − 0.38 = −0.95 kcal/mol for the equilibrium BAB ⇄ AAB at 277
K. The contributions of 0.38 kcal/mol arise from symmetry
considerations (−RT ln(1/2)). Of course, this is an idealized model:
we would expect to observe more of the ABB conformer, but evidently
its small deviation from the −0.47 kcal/mol assumed for the stacking
interaction reduces its population below the detection limit of the
experiment. This deviation is predicted by the B97-D calculations.
(71) Ormsby, J. L.; Black, T. D.; Hilton, C. L.; Bharat; King, B. T.
Tetrahedron 2008, 64, 11370−11378.
(72) The ABA conformers were ignored, for the reasons described
above.
6722
dx.doi.org/10.1021/ja4026006 | J. Am. Chem. Soc. 2013, 135, 6714−6722