Investigating the core moiety of banana-shaped liquid crystals using 2H NMR coupled with quantum simulations

Article abstract of DOI:10.1016/j.cplett.2004.12.051

Bent-core or 'banana-shaped' molecules have displayed an array of novel chiral liquid crystal (LC) phases. However, the descriptions of these molecules as having well defined bend angles, and possibly exhibiting shape-chirality, have limited analytical su

Full text of DOI:10.1016/j.cplett.2004.12.051

Chemical Physics Letters 402 (2005) 318–323
www.elsevier.com/locate/cplett
Investigating the core moiety of banana-shaped liquid crystals using
2H NMR coupled with quantum simulations
b
b,
a
V. Domenici a,b, L.A. Madsen , E.J. Choi b,c, E.T. Samulski *, C.A. Veracini
a
´
Dipartimento di Chimica e Chimica Industriale, Universita di Pisa, Via Risorgimento 35, 56126 Pisa, Italy
b
Department of Chemistry, Campus Box 3290, Venable and Kenan Laboratories, University of North Carolina at Chapel Hill,
Chapel Hill, NC 27599-3290, USA
c
Department of Polymer Science and Engineering, 188 Shinpyung-Dong, Kumi, Kyungbuk 730-701, Korea
Received 30 August 2004; in final form 3 December 2004
Available online 5 January 2005
Abstract
Bent-core or Ôbanana-shapedÕ molecules have displayed an array of novel chiral liquid crystal (LC) phases. However, the descrip-
tions of these molecules as having well defined bend angles, and possibly exhibiting shape-chirality, have limited analytical support.
Here, we present a step toward a more detailed understanding of the core moiety in banana molecules in LC phases, with the goal of
guiding further molecular design and characterization efforts. We present 2H NMR studies on two of the popular banana core moi-
eties, analyzed using ab initio structure calculations and the steric Ôinertial frameÕ model.
Ó 2004 Elsevier B.V. All rights reserved.
1. Introduction
rect evidence from X-ray diffraction and other analyt-
ical techniques [6–8].
A rapidly expanding area of liquid crystal (LC)
research is that of Ôbanana-shapedÕ or bent-core mole-
cules. Since the discovery of these Ôbanana phasesÕ in
1996 [1], approximately 50 000 new LC-forming Ômes-
ogenÕ structures [2] have been synthesized. These mes-
ogens produce a rich variety of novel mesophases
possessing chiral or achiral symmetries, and these
symmetry properties give rise to various electro-optical
and other physical properties, such as ferroelectricity
[3–5]. These materials promise to provide new degrees
of freedom in electro-optic display design as well as in
other LC and Ôsoft materialsÕ applications. Researchers
attribute many of these novel behaviors to the meso-
gen core geometries, as these molecules appear to be
bent when one analyzes them using simple chemical
structure arguments, computational analysis, and indi-
Recently, some have questioned the commonly
accepted structures of banana molecules, namely planar
and bent, in favor of more Ôrod-likeÕ [7], ÔtwistedÕ [6,8,9]
or Ônot totally planarÕ [10] structures. Some have claimed
using 13C NMR evidence [11] that the mesogens take on
chiral average conformations due to twist in the core
backbone, which in turn dictates the mesophase chirality
observed in this class of achiral molecules. While molec-
ular chirality is consonant with these observations, pla-
nar achiral conformations are also consistent with these
data. In short, there have been many assumptions made
when discussing the structures of banana molecules, but
little analytical data to support these assumptions.
We have begun a more detailed look at core confor-
mations in LC phases, with the goal of improving
molecular design and characterization strategies. Here,
we describe NMR studies coupled with quantum
structural calculations on two of the popular core moi-
eties. We have synthesized the selectively deuterium-la-
beled core molecules shown in Fig. 1, 1,3-benzenediol
*
Corresponding author. Fax: +1 919 962 2388.
E-mail address: et@unc.edu (E.T. Samulski).
0009-2614/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2004.12.051

V. Domenici et al. / Chemical Physics Letters 402 (2005) 318–323
319
y
at room temperature. After 5 min (30 min, ClBOB),
white precipitate appeared and another 5 mL dichlo-
romethane was added (none added for ClBOB). After
stirring 10 h at room temperature, and heating at
80 °C for 1 h (10 h for ClBOB), the mixture was cooled
to room temperature. After diluting with 50 mL dichlo-
romethane (60 mL, ClBOB) and washing with 50 mL
distilled water (200 mL, ClBOB), the dichloromethane
soluble fraction was collected, and dichloromethane
was removed using a rotatory evaporator. This product
was washed with distilled water and recrystallized from
a mixture of ethanol/water (1:1, BOB; 1:2, ClBOB).
Yield: 73% (83%, ClBOB).
Z
(A)
x
Y
H
H
z
X
H
X
d
L
d
R
O
O
D
b
D
O
O
d
2'
c
d
d
1'
1
d
a
D
2
D
x'
z'
D
D
D
D
D
D
We formed homogeneous solutions of these probe
molecules in two LC solvents for NMR investigations.
Merck and Aldrich, respectively, supplied the nematic
mixture Phase V (I) and the nematic 40-pentyl-4-biphe-
nyl carbonitrile (5 CB) (II). For NMR experiments, we
prepared 2.0–2.3 wt% solutions by weighing solvent
and solute amounts into 5 mm NMR tubes, heating into
the isotropic phase, and then mixing using a Vortex
apparatus.
y'
(B)
O
O
O
O
2
We acquired H NMR spectra at room temperature
Fig. 1. (A) Molecular structure of the two probe molecules (X = 1H,
BOB; X = Cl, ClBOB). The reference frames used in the analyses are
shown superimposed. (B) A planar achiral conformation of BOB that
would exhibit NMR chemical shift differences between the two outer
phenyls.
on the probe molecules in both solvents at 55.28 MHz
on a Bruker DMX-360 spectrometer using a broadband
10 mm rf coil probe and a simple pulse-acquire sequence
with a pulse of 2 ls (<20°) and a spectral width of
100 KHz. The numbers of scans were 20 000 and 4000
for BOB and ClBOB in solvent I, and 100 000 in solvent
II.
dibenzoate-d10 and 4-chloro 1,3-benzenediol dibenzo-
ate-d10 and refer to them as BOB and ClBOB (derived
from the common name benzoyloxy benzene). Dissolv-
ing these probe molecules into traditional rodlike
nematics and employing 2H NMR spectroscopy pro-
vides a sensitive probe of the conformational and ori-
entational order of these molecules. Deuterium
labeling on the outer phenyl rings allows for character-
ization of the relative conformations of these rings. We
have analyzed this data with the aid of ab initio struc-
ture calculations in conjunction with the Steric Ôinertial
frameÕ (IF) model [12]. This study takes a step toward
understanding the conformational preference of these
structures in real mesophases.
3. Quantum-mechanical structure calculations
In order to determine both the preferred conforma-
tions and the flexibility of these core molecules, we
employed the restricted Hartree–Fock (RHF) method
to build potential energy surfaces (PESs) for the internal
motions in both BOB and ClBOB. Of paramount inter-
est, especially for analysis of NMR data, was to under-
stand what conformations are favored for the lateral
wings (outer phenyl rings and carboxyls) relative to each
other and to the central phenyl ring of each probe mol-
ecule. Both BOB and ClBOB yielded somewhat surpris-
ing results in terms of their conformational preferences
and flexibilities.
2. Synthesis, sample preparation, and NMR
measurements
Initially, we performed geometry optimizations using
the GAUSSIAN 03 package [13] (RHF 6-31G* basis set)
without imposing symmetry restrictions, starting from
the molecular geometry optimised with the AM1 meth-
od. Both molecules optimize to planar conformations
with the torsional angles dR and dL, shown in Fig. 1,
measuring 0° and 0°, in BOB, and 0° and 180°, in
ClBOB, respectively. Thus, the two carbonyl groups
are on the same side of the central phenyl ring in
BOB, and on the opposite side in ClBOB.
Benzoic acid-d5 (11.8 mmol, BOB and ClBOB) was
refluxed in 25 mL of SOCl2 for 4 h. Excess SOCl2 was
removed by distillation under reduced pressure. The
benzoyl chloride-d5 product was dissolved in 10 mL of
dichloromethane. To this solution was added slowly
with vigorously stirring
a solution of resorcinol
(5.90 mmol, BOB), or 4-chlororesorcinol (5.90 mmol,
ClBOB) dissolved in a mixture (12 mL) of pyridine
and dichloromethane (1:10 v/v, BOB, 1:5 v/v, ClBOB)

320
V. Domenici et al. / Chemical Physics Letters 402 (2005) 318–323
Starting from these two stabilised geometries, several
low resolution scans were done for both molecules, by
ergy conformation {dR,dL} = {0°,0°} by only 1.44 kcal/
mol. The PES is quite flat over the entire ranges of these
two angles. Thus, BOB is a very flexible molecule in
terms of dR and dL (and ClBOB similarly in its dR an-
gle), and it is reasonable to assume that its conforma-
tions may be influenced by quite subtle intermolecular
(e.g., solvent–solute) interactions. Twisted chiral confor-
mations may arise in either molecule, but only under the
influence of symmetry-breaking intermolecular driving
forces.
0
varying the six dihedral angles dR and dL, d1 and d1 ,
0
d2 and d2 (see Fig. 1). According to these calculations
and to reference [6], the outer phenyl rings and carbonyl
groups are strongly favored to lie in the same plane,
allowing us to search only the two torsional angles dR
and dL to build PESs. Single point energy scans (with
geometry re-optimization) performed by varying dR
and dL in steps of 18°, from d0R to dR0 þ 180ꢀ and from
d0L to d0L þ 180ꢀ, clearly revealed a r symmetry with re-
spect to the central ring plane, and Fig. 2 show the opti-
mized PESs (100 conformations for each molecule).
Using this study, and contrary to the conventional
pictures propagated in the banana LC literature, one no-
tices the following structural properties of BOB and
ClBOB. For ClBOB the PES is not symmetric with
respect to the angles dR and dL in that the conformation
{dR,dL} = {0°, 0°}, in which the chlorine is closer to the
carbonyl oxygen atom, is less favored than the most sta-
ble conformation {dR,dL} = {0°,180°}, by 30.75 kcal/
mol. Furthermore, the {dR,dL} = {180°,180°} conforma-
tion is only 1.77 kcal/mol above the most stable confor-
mation. For BOB, the two conformations {dR,dL} =
{0°,180°} and {dR,dL} = {180°,0°} clearly have the same
energy, and they are less favored than the minimum en-
4. Steric inertial frame model of mesogen conformation
The computation of second-rank properties of non-
rigid molecules in uniaxial phases through equilibrium
statistical procedures [12,14] has a long history and here,
we summarize the main features, focusing on our NMR
analysis needs. We detail these calculations for quadru-
pole spin interactions, but they are easily extended to
other NMR observables [15].
The ordering matrix element S, which describes the
average orientation of a cartesian frame fixed to the
molecule and relative to the nematic director, can be de-
fined as
ꢀ
ꢁ
Sab ¼ 1=2 3lalb ꢁ dab
;
ð1Þ
where the number of independent elements in S depends
on the shape anisotropy of the solute [16], which
depends on the configuration. For qualitative modeling,
we adopt the coarse approximation that S{/} is
assumed diagonal in the molecular frame (X, Y, Z) that
diagonalizes the moment of inertia tensor I{/}, where
the masses are replaced with the van der Waals radii
[17,18], giving
ꢂ
ꢃ
X
Iab
¼
Ri ri ꢂ ri ꢂ dab ꢁ riarib
;
ð2Þ
i
where Ri and ri are the radius and the position of the
i-atom, respectively.
According to this model, the three diagonal compo-
nents of S can be calculated for each conformation {/}
from the equivalent ellipsoid semiaxes Ac that are related
to the principal moments of inertia of said configuration
{/} [19]. Once Sab{/} is known, the quadrupolar split-
ting can be calculated using the general expression
X;Y ;Z
X X
2
Dmi ¼
Pf/g ꢂ qiabf/g ꢂ Sabf/g;
ð3Þ
3
a;b
f/g
where qiabf/g is the quadrupole interaction tensor
defined in the molecular frame for the i–CD bond, and
P{/} = Zꢁ1exp[ꢁE{/}/kT] is the Boltzmann probabil-
ity that the molecule has configuration {/}. The IF
model assumes the energy of a single molecule E{/} de-
pends only on the configuration {/}, where we used the
Fig. 2. Potential energy surfaces for BOB (A) and ClBOB (B),
obtained by varying dihedral angles dR and dL.

V. Domenici et al. / Chemical Physics Letters 402 (2005) 318–323
321
energy evaluated for each conformation {dR, dL} in the
ab initio study described above.
Since it is more convenient to describe the quadru-
pole interaction tensor in the CD bond local frame (x0,
y0, z0), where it is diagonal and independent of the con-
formation {/}, we used
x0;y0;z0
x0;y0;z0
D
E
X
X
X
2
3
2
3
Dmi ¼
qiab
Pf/g ꢂ siabf/g ¼
qiab siab
;
a;b
a;b
f/g
ð4Þ
where hsiabi is the ordering matrix element defined in the
CD bond frame and averaged over all configurations.
A simple way to include contributions to the energy
due to the anisotropic environment of a locally uniaxial
mesophase consists of assuming that the molecule with
conformation {/} is contained in a hypothetical cylin-
der of radius rc, whose axis is coincident with the Z-axis,
which is the direction of the minor principal moment of
inertia [12,20]. In this coarse model, the configuration
{/} is accepted if all atoms are contained in the cylinder,
otherwise it is discarded. This procedure is repeated for
each value of the radius rc, calculating all properties,
such as the quadrupole splittings Dmi (rc), and orienting
matrices hsiabðrcÞi, in different frames, e.g., for different
rings or CD bonds.
Fig. 3. 2H NMR spectra for BOB (a) and (b) and for ClBOB (c) and
(d).
5. NMR results and analysis
2
Figs. 3a, b show the H NMR spectra of BOB in the
two nematic solvents. In all spectra reported here, we
show only half of the frequency axis, since the spectra
are symmetric about zero frequency. These spectra are
characterized by 3 quadrupolar splittings, with the larg-
est splitting ascribable to the para CD bonds of the outer
phenyl rings, and the inner splittings to the meta and
ortho CD bonds. Due to the equivalence of the quadru-
pole splittings between the 2 rings to within our resolu-
tion (peak widths), these rings must be orientationally
and conformationally equivalent. Thus, there is no evi-
dence that these molecules exhibit average chiral confor-
mations in nematic solvents. Prior to conducting this
study, we also dissolved BOB (undeuterated) into con-
ventional liquid solvents (CDCl3) and into chiral liquid
solvents ((ꢁ)-acetoxy-p-menthane). In these cases,
BOB exhibited no outer ring inequivalence as observed
(dR ¼ dL). We cannot comment from this analysis di-
rectly on the chirality of this molecule.
As a first analysis of these NMR data, and with the
goal of understanding the orientational and conforma-
tional structures of these molecules, we performed com-
putations using the Steric IF model of the quadrupolar
splittings as a function of the cylinder radius rc from 3
˚
to 10 A by introducing our calculated ab initio energies
E{/} into Eq. (4) (see Fig. 4). We note that in this sim-
plistic model of the nematic environment, there is not a
large difference in volume of the cylinder of radius rc gi-
ven by different conformations of the molecules so that
˚
for BOB, rc is restricted. That is, for rc < 4.54 A, no con-
˚
formations are accepted, while for rc > 5.16 A, all 100
initial conformations are accepted. For ClBOB, this
1
in H and 13C chemical shifts or J-couplings.
range is 4.16 A < rc < 6.36 A, and this extension of the
sensitive range of rc is clearly due to the chlorine atom,
which dramatically influences the molecular shape.
˚
˚
As anticipated from our quantum-structural study
above, the deuterium spectra of ClBOB showed quite
different behavior in the two solvents, as shown in Figs.
3c,d. The appearance of six quadrupolar splittings, the
outermost two splittings ascribable to the para CD
bonds of the two outer phenyl rings and the four inner
splittings to the meta and ortho CD bonds, shows that
these rings are not conformationally equivalent
2
These computations qualitatively coincide with the H
spectra, where BOB exhibits equivalent splittings for
the outer phenyl rings, while ClBOB does not.
To quantitatively fit the spectra to extract values of rc
shown in Fig. 4, we used the ratios of the quadrupole
splittings since the absolute values of the calculated

322
V. Domenici et al. / Chemical Physics Letters 402 (2005) 318–323
molecules [24]. And, a cursory look at the conformer
shapes depicted in Fig. 2 show that BOB does not corre-
spond to simple elongated molecules thereby rendering
the utility of the IF model less reliable for this class of
molecules. Nevertheless, the IF model can serve as a
zero-order approximate description of the conformer
preferences exhibited by these important constituents
of banana liquid crystals.
In sum, these observations are a first step to bringing
a more discerning eye to the study of banana mesogens
in mesophases. In the uniaxial solvents employed herein,
there is no apparent propensity for chiral conformations
of the central core moiety of banana mesogens. Apply-
ing more complex and discerning models of excluded-
volume interactions (than the crude shape selection of
the IF model), e.g., bent-core models [25–27], as well
as implementing different 2H-labeling schemes holds
promise for providing insights into the complicated
structure-property correlations in banana phases.
Acknowledgments
The authors thank the P.R.I.N. 2003 (Italian MIUR)
and the NSF (DMR-9971143) for financial support.
References
Fig. 4. Plots of scaled quadrupole splittings vs. rc for BOB (a) and
ClBOB (b). Vertical lines represent best fits to rc.
[1] T. Niori, T. Sekine, J. Watanabe, T. Furukawa, H. Takezoe, J.
Mater. Chem. 7 (1997) 1307.
[2] N.A. Clark, in: Plenary Lecture ÔPL2Õ at the International Liquid
Crystal Conference, Ljubljana, Slovenja, 2004.
splittings will be scaled down due to LC averaging. Due
2
to the well defined geometry of H nuclei on the phenyl
[3] D.R. Link, G. Natale, R. Shao, J.E. Maclennan, N.A. Clark, E.
Korblova, D.M. Walba, Science 278 (1997) 1924.
[4] J. Ortega, J.A. Gallastegui, C.L. Folcia, J. Etxebarria, N. Gimeno,
M.B. Ros, Liq. Cryst. 31 (2004) 579.
rings, we have a fully defined system (in terms of split-
ting ratios) with which to fit rc. We reproduced the
experimental spectra exactly for ClBOB in both solvents
[5] D.A. Olson, M. Veum, A. Cady, M.V. DÕAgostino, P.M.
Johnson, H.T. Nguyen, L.C. Chien, C.C. Huang, Phys. Rev. E
63 (2000) 041702.
˚
(I and II) where rc was 4.66 and 4.62 A for the two sol-
vents with 17 and 18 conformers rejected, respectively,
[6] T. Imase, S. Kawauchi, J. Watanabe, J. Mol. Struct. 560 (2001)
275.
[7] H. Hartung, J. Mol. Struct. 526 (2000) 31.
˚
and for BOB in 5CB (II), where rc was 5.06 A (with 11
conformers rejected), indicating that these probe mole-
cules experience a calamitic mean field due to the host
solvents. Indeed, these values of rc found for BOB and
ClBOB, are in good agreement with the IF literature
on less rigid molecules [21,22].
[8] J. Thisayukta, H. Niwano, H. Takezoe, J. Watanabe, J. Am.
Chem. Soc. 124 (2002) 3354.
[9] H. Kurosu, M. Kawasaki, M. Hirose, M. Yamada, S. Kang, J.
Thisayukta, M. Sone, H. Takezoe, J. Watanabe, J. Phys. Chem.
A 108 (2004) 4674.
The IF model fails to generate a fit for any value of rc
for the solute BOB in Phase V (I). The conformational
space for BOB is inherently larger than that of ClBOB
because of the intramolecular steric interactions in the
latter. In BOB, rotations about the dR and dL angles
are virtually unimpeded (see Fig. 2). Hence, the orienta-
tional ordering of the solute BOB may be more suscep-
tible to specific (electrostatic) solute–solvent interactions
than ClBOB [23]. Additionally, it should be emphasized
that the IF model was introduced merely as a crude way
to identify the so called long-molecular-axis of elongated
[10] R.Y. Dong, K. Fodor-Csorba, J. Xu, V. Domenici, G. Pramp-
olini, C.A. Veracini, J. Phys. Chem. B 108 (2004) 7964.
[11] T. Sekine, T. Niori, M. Sone, J. Watanabe, S. Choi, Y. Takanishi,
H. Takezoe, Jpn. J. Appl. Phys. 36 (1997) 6455.
[12] E.T. Samulski, Ferroelectrics 30 (1980) 83.
[13] M.J. Frisch et al., GAUSSIAN 03, Revision A.1, Gaussian, Inc.,
Pittsburgh, PA, 2003.
[14] J.W. Emsley, G.R. Luckhurst, Mol. Phys. 41 (1980) 19.
[15] C.A. Veracini, in: J.W. Emsley (Ed.), Nuclear Magnetic
Resonance of Liquid Crystals, Reidel, Dordrecht, vol. 141,
1985, p. 99.
[16] J.W. Emsley, J.C. Lindon, NMR Spectroscopy Using Liquid
Crystal Solvents, Pergamon Press, Oxford, 1975.

V. Domenici et al. / Chemical Physics Letters 402 (2005) 318–323
323
[17] D. Catalano, C. Forte, C.A. Veracini, C. Zannoni, Israel J.
Chem. 23 (1983) 283.
[23] T. Dingemans, D.J. Photinos, E.T. Samulski, A.F. Terzis, C.
Wutz, J. Chem. Phys. 118 (2003) 7046.
[18] A. Bondi, J. Phys. Chem. 68 (1964) 441.
[19] E.T. Samulski, R.Y. Dong, J. Chem. Phys. 77 (1982) 5090.
[20] J.W. Emsley, G.R. Luckhurst, C.P. Stockley, Mol. Phys. 38
(1979) 1687.
[24] E.T. Samulski, in: E.E. Burnell, C.A. de Lange (Eds.), NMR of
Ordered Liquids, Kluwer, 2003, p. 295 (Chapter 13).
[25] R. Memmer, Mol. Phys. 29 (2000) 483.
[26] T.C. Lubenski, L. Radzihovsky, Phys. Rev.
031704.
[27] A. Dewar, J.P. Camp, Phys. Rev. E 70 (2004) 011704.
E 66 (2002)
[21] E.T. Samulski, Israel J. Chem. 23 (1983) 329.
[22] E.T. Samulski, H. Toriumi, J. Phys. Chem. 79 (1983) 5194.