Dragan B Ninković1, Jelena P Blagojević Filipović1, Michael B Hall2, Edward N Brothers3, Snežana D Zarić4,3. 1. Innovation Center of the Faculty of Chemistry in Belgrade, Studentski trg 12-16, Belgrade 11001, Serbia. 2. Department of Chemistry, Texas A&M University, College Station, Texas 77843-3255, United States. 3. Department of Chemistry, Texas A&M University at Qatar, P.O. Box 23874, Doha, Qatar. 4. Faculty of Chemistry, University of Belgrade, Studentski trg 12-16, Belgrade 11000, Serbia.
Abstract
High-level ab initio calculations show that the most stable stacking for benzene-cyclohexane is 17% stronger than that for benzene-benzene. However, as these systems are displaced horizontally the benzene-benzene attraction retains its strength. At a displacement of 5.0 Å, the benzene-benzene attraction is still ∼70% of its maximum strength, while benzene-cyclohexane attraction has fallen to ∼40% of its maximum strength. Alternatively, the radius of attraction (>2.0 kcal/mol) for benzene-benzene is 250% larger than that for benzene-cyclohexane. Thus, at relatively large distances aromatic rings can recognize each other, a phenomenon that helps explain their importance in protein folding and supramolecular structures.
High-level ab initio calculations show that the most stable stacking for benzene-cyclohexane is 17% stronger than that for benzene-benzene. However, as these systems are displaced horizontally the benzene-benzene attraction retains its strength. At a displacement of 5.0 Å, the benzene-benzene attraction is still ∼70% of its maximum strength, while benzene-cyclohexane attraction has fallen to ∼40% of its maximum strength. Alternatively, the radius of attraction (>2.0 kcal/mol) for benzene-benzene is 250% larger than that for benzene-cyclohexane. Thus, at relatively large distances aromatic rings can recognize each other, a phenomenon that helps explain their importance in protein folding and supramolecular structures.
Aromatic–aromatic
interactions[1−49] have been invoked as key features of a number of molecular phenomena:
protein folding,[34−37] crystal engineering,[38−41] catalysis,[42−46] and drug design.[1,47−49] Explanations
have suggested that there is something special about these interactions.[15,17,20,21] However, it has been clearly demonstrated that the aromaticity is
not the key as nonaromatic, planar 6π electron systems have
stacking energies similar to those of benzene.[17] Here, we will use the term aromatic–aromatic interactions
as most observations fall into that category, but our conclusions
will apply in other cases.Unexpectedly, the calculated interaction
energies for the stacking
of cyclohexane dimers are nearly as large as that for benzene dimers.[18] Furthermore, the stacking interaction between
benzene and cyclohexane is somewhat stronger than either homodimer.[19] In all of these interactions, electrostatic
and dispersion play important roles, but dispersion is larger in benzene
dimer.[18,20] As the molecules become larger, more favorable
dispersion and less repulsion in large aromatic systems, with more
than 10–15 carbon atoms, contributes to stronger aromatic stacking
interactions in comparison to aliphatic.[15,21]Numerous computational studies of the interaction energy between
two benzene molecules[22−30] have established that the most stable benzene dimer has the tilted
T-shape (edge-to-face), with a CCSD(T)/CBS interaction energy of −2.84
kcal/mol,[29] while the most stable stacking
benzene–benzene interaction has a geometry with a parallel
displacement (offset) of 1.5 Å, and an interaction energy of
−2.79 kcal/mol.[30] Recent work has
shown that substantial interaction energies of around −2.0
kcal/mol are predicted for larger offsets of 4.0–5.0 Å.[30] Furthermore, statistical analysis of the data
from the Cambridge Structural Database (CSD) showed that the preferred
stacking (parallel) benzene–benzene interactions in crystal
structures are at large horizontal displacements (3.5–5.0 Å),
and not at the 1.5 Å, where the energy minimum lies in the gas-phase
dimer. Similar large displacements are also found for interactions
between aromatic rings of phenylalanine in proteins,[31] and pyridine–pyridine interactions in crystal structures
from the CSD.[32]In this work, we
compared potential energy curves for stacking
benzene–benzene vs stacking benzene–cyclohexene interactions,
including large horizontal displacements. We also analyzed the nature
of these interactions by performing SAPT analyses.Calculations
at high level, including the coupled-cluster/complete-basis-set
limit, CCSD(T)/CBS, show that benzene–benzene dimers have a
much large radius of attraction compare to benzene–cyclohexane
dimers (Figure ) in
spite of the benzene–cyclohexane dimers having a larger attraction
energy at their corresponding minimum energy structures. Details of
these calculations and a SAPT analysis of contributing energy components
are described in this paper.
Figure 1
Areas in which attractive interaction energy
is larger than −2.0
kcal/mol for benzene–benzene (left) and benzene–cyclohexane
(right).
Areas in which attractive interaction energy
is larger than −2.0
kcal/mol for benzene–benzene (left) and benzene–cyclohexane
(right).
Results and Discussion
Important
insight about the specificity of aromatic–aromatic
interactions can be obtained by comparing the calculated potential
energy curves for cyclohexane–benzene,[19] and benzene–benzene stacking interactions,[30] following the geometric parameters shown in Figure .
Figure 2
Geometric parameters
used in calculations of benzene–benzene
and cyclohexane–benzene interactions. Ωa and
Ωb denote benzene or cyclohexane ring centers. R
denotes distance between parallel mean planes of the rings, while
Ωb′ are benzene and cyclohexane ring centers
projections onto the benzene rings planes; r denotes displacement
of the second ring projection on the benzene.
Geometric parameters
used in calculations of benzene–benzene
and cyclohexane–benzene interactions. Ωa and
Ωb denote benzene or cyclohexane ring centers. R
denotes distance between parallel mean planes of the rings, while
Ωb′ are benzene and cyclohexane ring centers
projections onto the benzene rings planes; r denotes displacement
of the second ring projection on the benzene.Accurate potential energy curves were calculated by high-level
quantum chemical methods,[19,30] which are in good agreement
with CCSD(T)/CBS,[33,50] with Gaussian09[51] (version D.01, the details of the calculations are given
in the SI).The data in Figure show the variation
of the average interaction energy as one ring
is displaced from the other ring (r) and maintained
at the minimum distance apart (R). The average is
done with respect to three possible rotational orientations of the
rings and displacements in both the left and right direction (see Figures S1 and S2 for details of the orientations
and all six interaction energy curves). The largest difference in
average energies is at small offsets (<1.0 Å), where the cyclohexane–benzene
dimer is significantly more stable than the benzene–benzene
dimer (Figure , Tables and 2). At offset 0.0 Å (sandwich or face-to-face geometry)
the CCSD(T)/CBS average interaction energy for cyclohexane–benzene
is −3.15 kcal/mol, while the average benzene–benzene
interaction energy is −1.75 kcal/mol. Benzene–benzene
has its strongest interaction energy, −2.78 kcal/mol, at an
offset about 1.5 Å, where the average cyclohexane–benzene
interaction energy is nearly the same value.
Figure 3
Calculated interaction
energies averaged over the three possible
orientations and two directions for cyclohexane–benzene[19] and benzene–benzene[30] (Figures S1 and S2) plotted
as a function of the displacement (r) (Figure ).
Table 1
Results of SAPT Analysis and CCSD(T)/CBS
Calculations for Benzene–Benzene Interactions at Several Offset
Values (Figure )a
offset
(r)
r = 0.0
r = 1.5
r = 4.0
r = 5.0
normal distance (R)
3.90
3.50
3.20
2.67
electrostatics
0.09
–1.50
–1.30
–1.31
exchange
3.28
6.58
3.10
2.67
induction
–0.22
–0.70
–0.35
–0.30
dispersion
–5.00
–7.21
–3.47
–2.95
net dispersionb
–1.72
–0.62
–0.38
–0.28
total SAPT2+3
–1.85
–2.83
–2.03
–1.89
CCSD(T)/CBS
–1.75
–2.78
–2.02
–1.85
% of the most stable
interactionc
63
100
72
69
Offset
and normal distance values
are given in Å. All interaction energies and energy components
are average for the three potential curves (Tables S1–S3) and given in kcal/mol.
Net dispersion is sum of dispersion
and exchange terms.
The
most stable interaction is for
minimum at Abb curve with energy of −2.79 kcal/mol (Figure S2).
Table 2
Results of SAPT Analysis and CCSD(T)/CBS
Calculations for Cyclohexane–Benzene Interactions at Several
Offset Values (Figure )a
offset
(r)
r = 0.0
r = 1.5
r = 4.0
r = 5.0
normal distance (R)
4.20
4.00
3.65
3.03
electrostatics
–1.75
–1.92
–0.65
–0.55
exchange
4.61
5.36
2.30
2.05
induction
–0.57
–0.59
–0.25
–0.21
dispersion
–5.49
–5.89
–2.99
–2.66
net dispersionb
–0.88
–0.54
–0.69
–0.61
total SAPT2+3
–3.20
–3.02
–1.59
–1.38
CCSD(T)/CBS
–3.15
–2.88
–1.55
–1.35
% of the most stable
interactionc
96
88
47
41
Offset
and normal distance values
are given in Å. All interaction energies and energy components
are average for the three potential curves (Tables S4–S6) and given in kcal/mol.
Net dispersion is sum of dispersion
and exchange terms.
The
most stable interaction is for
minimum at Abc curve with energy of −3.27 kcal/mol (Figure S2).
Calculated interaction
energies averaged over the three possible
orientations and two directions for cyclohexane–benzene[19] and benzene–benzene[30] (Figures S1 and S2) plotted
as a function of the displacement (r) (Figure ).Offset
and normal distance values
are given in Å. All interaction energies and energy components
are average for the three potential curves (Tables S1–S3) and given in kcal/mol.Net dispersion is sum of dispersion
and exchange terms.The
most stable interaction is for
minimum at Abb curve with energy of −2.79 kcal/mol (Figure S2).Offset
and normal distance values
are given in Å. All interaction energies and energy components
are average for the three potential curves (Tables S4–S6) and given in kcal/mol.Net dispersion is sum of dispersion
and exchange terms.The
most stable interaction is for
minimum at Abc curve with energy of −3.27 kcal/mol (Figure S2).An important difference between the average cyclohexane–benzene
and benzene–benzene stacking energies also occurs at large
offsets, where that for benzene–benzene is stronger. At large
offsets (4.0–5.0 Å) the cyclohexane–benzene energies
are only 41–47% of their minimum energy, while the benzene–benzene
energies are 69–72% of their minimum energy (Tables and 2).In addition to relatively strong interactions (Figure , Table ) the interactions at large
offsets leave
faces of benzene molecules available for additional interactions with
surrounding species in supramolecular structures. One example is shown
in Figure a where
every benzene ring has additional interactions on both sides of benzene
(CH/π and stacking) which is not possible in benzene stacking
interactions with smaller offset values.
Figure 4
(a) Parallel benzene–benzene
interaction with a large offset
(r = 4.81 Å) in the crystal structure of EREYUV.[52] Both benzene molecules, involved in parallel
interaction, also form CH/π and stacking interactions on both
sides of the rings with molecules from the environment. (b) Histogram
of the offset values r for phenyl–cyclohexyl[19] (blue, left bars) and benzene–benzene
(orange, right bars)[30] interactions. N is the number of interactions.
(a) Parallel benzene–benzene
interaction with a large offset
(r = 4.81 Å) in the crystal structure of EREYUV.[52] Both benzene molecules, involved in parallel
interaction, also form CH/π and stacking interactions on both
sides of the rings with molecules from the environment. (b) Histogram
of the offset values r for phenyl–cyclohexyl[19] (blue, left bars) and benzene–benzene
(orange, right bars)[30] interactions. N is the number of interactions.The specific behavior of the benzene–benzene interactions
is illustrated by the benzene–benzene and benzene–cyclohexane
interactions in crystal structures. The data in Figure b show quite different offset distributions
for stacked benzene–benzene[30] and
benzene–cyclohexane[19] in crystal
structures. Specifically, most benzene–benzene interactions
(orange, right bars, Figure b) were observed for large offsets, from 4.5 to 5.5 Å,
with a very small number of the interactions at small offsets. Such
a tendency is not so pronounced for phenyl–cyclohexyl contacts
(blue, left bars, Figure b). Here again, this is due to benzene interactions at large
offsets, since most of the maximum possible interaction energy is
preserved at large offsets (Table , Figure ). As was mentioned above, an additional advantage of non-negligible
interactions at large offsets in supramolecular structures is the
possibility of forming simultaneous interactions (Figure a).The SAPT method[53] provides insight into
the nature of the cyclohexane–benzene and benzene–benzene
stacking interactions, including interactions at large horizontal
displacements. We used a SAPT method with a density-fitting approximation
(DF-SAPT2+3)[54] and the def2-tzvppd basis
set, since using this basis set gave results in good agreement with
accurate CCSD(T)/CBS energies (Tables and 2, Tables S1–S6). The calculations were performed using
the PSI4 program.[55] The data show that
the most important contribution to the total energy is dispersion
at all offsets for both systems (Tables and 2). The second
important attractive contribution is electrostatics, with the exception
of the benzene–benzene sandwich geometry (r = 0.0 Å), where the electrostatic term is repulsive. If we
add the attractive dispersion term to the exchange, the resulting
net dispersion is less attractive than −1.0 kcal/mol, except
for the benzene–benzene sandwich geometry.In the cyclohexane–benzene
dimer, the electrostatic term
at small offsets (0.0 and 1.5 Å) is more attractive than the
net dispersion, while at larger offsets (4.0 and 5.0 Å) the net
dispersion term is more attractive than the electrostatic term. The
benzene–benzene interaction is different: the net dispersion
term dominates at the offset of 0.0 Å, while the electrostatic
term dominates at other offsets and decreases only slightly with increasing
offset.These differences are illustrated in Figure which shows the key energy
terms as a function
of the displacement (r) for both cyclohexane–benzene
and benzene–benzene. Hence, the significantly more attractive
cyclohexane–benzene interaction, at 0.0 Å offset, is mostly
a consequence of attractive electrostatic energy. At 1.5 Å offset
electrostatics is the most dominant in both systems; it is still somewhat
stronger in benzene–cyclohexane, causing slightly stronger
interaction. At larger displacements the benzene–benzene interaction
is stronger than the benzene–cyclohexane interaction, mostly
due to more favorable electrostatic contribution in benzene–benzene
dimer, despite the fact that net dispersion is slightly more favorable
in cyclohexane–benzene than in benzene–benzene dimer.
Figure 5
Results
of SAPT analysis and CCSD(T)/CBS calculations for (a) benzene–benzene
and (b) cyclohexane–benzene at several offset values. All interaction
energies and selected energy components are average (Tables and 2) and given in kcal/mol.
Results
of SAPT analysis and CCSD(T)/CBS calculations for (a) benzene–benzene
and (b) cyclohexane–benzene at several offset values. All interaction
energies and selected energy components are average (Tables and 2) and given in kcal/mol.The importance of the electrostatic term in benzene–benzene
stacking interaction at small offsets is well-known.[56,57] The electrostatic term remains quite favorable at large offset in
the benzene–benzene interaction because the local C–H
dipoles are in an antiparallel orientation, as illustrated in Figure . On the other hand,
in cyclohexane–benzene interaction the C–H dipoles are
at right angles; hence, the electrostatic attraction is relatively
small. One can also notice that hydrogen atoms in the cyclohexane
molecule have significantly smaller positive potentials (Figure ).
Figure 6
Overlaid electrostatic
potentials for benzene–benzene and
cyclohexane–benzene interactions at 5.0 Å. Electrostatic
potentials for cyclohexane and benzene were plotted on a contour with
an electron density of 0.001 e–/(a.u.).[3]
Overlaid electrostatic
potentials for benzene–benzene and
cyclohexane–benzene interactions at 5.0 Å. Electrostatic
potentials for cyclohexane and benzene were plotted on a contour with
an electron density of 0.001 e–/(a.u.).[3]
Conclusions
Based
on SAPT analysis one can conclude that the difference in
potential energy surfaces for cyclohexane–benzene and benzene–benzene
is a consequence of the ability of cyclohexane–benzene to form
strong electrostatic interactions at small offset, and the capability
of benzene–benzene to preserve relatively strong electrostatics
at large offsets.The potential curves in Figure indicate that the important advantage of
aromatic–aromatic
interactions is the larger range of the attractive interactions (Figure ), which arise from
the long-range nature of the electrostatic interactions (Figure a). The stronger
interaction at large offsets is the key feature that makes aromatic–aromatic
interactions special. As illustrated in Figure , strong benzene–benzene attraction
(>−2.0 kcal/mol) occurs up to ∼4.5 Å, which
gives
a 4.5 Å radius of attraction while that for cyclohexane–benzene
is only 2.75 Å. After subtracting the small area where benzene–benzene
interactions are weaker than −2.0 kcal/mol the area of strong
attraction is 61.8 Å2, while that for cyclohexane–benzene
is only 23.8 Å2 (Figure ). Therefore, the “region of attraction”
stronger than −2.0 kcal/mol is 2.5 times larger for benzene–benzene
than for cyclohexane–benzene. Thus, benzene–benzene
stacking interactions have a remarkable advantage since two benzenes
(phenyl groups) can recognize each other over a much greater range
of distances.
Authors: P Chandra Sekar; E Srinivasan; G Chandrasekhar; D Meshach Paul; G Sanjay; S Surya; N S Arun Raj Kumar; R Rajasekaran Journal: J Mol Model Date: 2022-04-24 Impact factor: 2.172
Authors: Milan R Milovanović; Ivana M Stanković; Jelena M Živković; Dragan B Ninković; Michael B Hall; Snežana D Zarić Journal: IUCrJ Date: 2022-08-05 Impact factor: 5.588