Chiral Self-Sorting and the Realization of Ferroelectricity in the Columnar Liquid Crystal Phase of an Optically Inactive N,N'-Diphenylurea Derivative Possessing Six (±)-Citronellyl Groups.

An axially polar-ferroelectric columnar liquid crystal (AP-FCLC) phase that exhibits both switching and maintenance of the macro-polarity in the column axis direction has been achieved in an N,N'-bis(3,4,5-trialkoxyphenyl)urea compound (rac-1) prepared from (±)-citronellyl bromide. Although it had been thought that chirality is necessary to achieve the AP-FCLC phase from our previous study, the optically inactive compound which is a mixture of 21 stereoisomers, generated an AP-FCLC phase. We confirmed its ferroelectricity and investigated the mechanism for realizing the AP-FCLC phase using optoelectronic experiments, X-ray diffraction, and circular dichroism spectroscopy. As a result, it was suggested that chiral self-sorting occurs in the columnar liquid crystal phase, in which molecules with a similar stereochemistry form a one-handed helical column, and columns with the same helicity gather together to form a chiral domain. Accordingly, we conclude that the optically inactive compound rac-1 also indicates ferroelectricity similar to that of an optically pure urea compound because of chiral self-sorting.

Introduction

Recently, axially polar ferroelectric columnar liquid crystal (AP-FCLC) phases (Figure ), in which the polarity along the column axis is induced by applying an electric field and is maintained after removal of the electric field, have been increasingly studied.[1−7] In the columnar liquid crystal (LC) phase, the polar directions of the molecules (Figure a) are fixed to generate a polar column (Figure b) by intermolecular steric interactions[8−14] or intermolecular hydrogen bonding between the groups such as amide,[4,15−28] urea,[29−31] triazole,[32] and vinylidene fluoride oligomer (VDFO)[33] groups. The polar columns that are self-organized into a nonpolar state (Figure c) are polarized by applying an electric field, and the induced columnar polarities are switched (Figure d,e) by changing the directions of the applied electric field. The AP-FCLC phase has the potential to realize writable, rewritable, and nonvolatile nano-sized memory devices if one-bit of information can be recorded by a nano-sized electrode.[2,4,31] However, it is still challenging to maintain the induced polarization after removal of the electric field, and we believe that polarization maintenance is most important to realize nonvolatile memory devices.

Figure 1

Behaviors of the AP-FCLC phase: (a) polar molecules, (b) polar column, (c) nonpolar state, (d) polar state (upward polarity), and (e) polar state (downward polarity).

In 2012, two examples of time scanning of the residual polarization after removal of the applied voltage were reported. Fitié et al.[34] reported that the directions of the three amide carbonyl groups in benzenetrisamides (BTA) were controlled by an applied electric field in an LC cell with a 5 μm cell gap and that the induced polarization was maintained for a long time (1–1000 s). Miyajima et al. reported an AP-FCLC phase of a benzene derivative, possessing two cyano and two amide groups forming core–shell columnar structures,[4] and indicated that the polarity induced by corona poling was maintained at 80% even after 1000 min.

More recently, it was reported that LC compounds possessing a large coercive electric field (Ec), an electric field necessary for switching, show ferroelectricity by sandwiching a submicron thin sample film between two metal electrodes. FCLCs with a large Ec are advantageous in maintaining the induced polar state because their depolarization processes are suppressed by the large energy barrier. In 2016, Urbanaviciute et al. reported ferroelectricity in LC BTA films (thickness: 350–400 nm),[21,35] and in 2019, Casellas et al. reported ferroelectricity in an LC benzotrithiophenetrisamide (BTTTA) film (thickness: 300–800 nm).[20] These ferroelectric LC C3 symmetric trisamides require a large Ec (coercive fields of BTA and BTTTA: 25–225 and 150–300 V μm–1). García-Iglesias et al. in 2016[33] and Gorbunov et al. in 2017[36] reported that VDFO-functionalized perylenebisimides and phthalocyanines showed ferroelectricity, and that the Ec values measured in the 1 μm thin films were also large (100 and 25 V μm–1, respectively).

In 2020, our group also reported ferroelectricity in a rather thick layer (5 μm) of a chiral urea compound, N,N′-bis(3,4,5-tri(S)-citronellyloxyphenyl)urea [(S)-1][31] (Scheme ), which achieved both a low Ec (4.12 V μm–1) and highly stable polarization (the polarity quantitatively remained even after 8 h). From the comparison of (S)-1 with non-chiral ureas [e.g., Urea-10 (R = (CH2)9CH3)],[29,31] we concluded that the helical structure generated by the chirality played an important role in generating its unique ferroelectricity. This result was because the one-handed, tightly wound helix kept the intermolecular distances between the substituent groups short, which effectively strengthened the attractive intermolecular interactions, such as alkyl–alkyl, phenyl–phenyl, and dipole–dipole interactions, in each column.

Scheme 1

Molecular Structures of rac-1, (S)-1, (R)-1, and Urea-10

To verify the effect of chirality on the generation of the FCLC phase of (S)-1, we prepared urea compound rac-1 from (±)-citronellyl bromide. Since compound rac-1 is optically inactive, it was proposed that rac-1 would not show ferroelectricity. However, surprisingly, the columnar phase of rac-1 shows a perfect ferroelectricity similar to that of (S)-1, possessing six (S)-citronellyl groups. Because either of the (S)- or (R)-citronellyl groups is introduced to each of its six substituent groups in the rac-1 molecule with the same probability, compound rac-1 should be a mixture of 21 stereoisomers and optically inactive.

In this study, the ferroelectric properties of rac-1 are investigated, and the mechanism for the generation of the ferroelectricity in the columnar LC phase is discussed. As a result, we conclude that the key to the generation of the ferroelectricity of rac-1 is chiral self-sorting in the columnar LC phase.

Results and Discussion

Synthesis of Diphenylureas 1

Compounds rac-1, (S)-1, and (R)-1, as shown in Scheme , were prepared as follows (the synthetic route and procedures are shown in Figure S1 and Supporting Information). Pyrogallol was alkylated with (±)-, (S)-, and (R)-citronellyl bromide to provide the corresponding 1,2,3-trialkoxybenzenes. Nitration of the trialkoxybenzenes with NaNO2/HNO3 followed by reduction with iron powder in ethanol/water in the presence of ammonium chloride produced 3,4,5-trialkoxyanilines. The aniline derivatives were reacted with 1,1′-carbonyldiimidazole to produce diphenylureas 1.

Identification of the Phase Transition Behaviors

The phase transition behaviors (Table ) were investigated by polarized light optical microscopy (POM). Upon cooling, rac-1 showed an isotropic (Iso) to hexagonal columnar (Colh) phase transition. As shown in Figure a, both planarly and homeotropically aligned domains are observed. Further cooling caused the phase transition from the Colh phase to the rectangular columnar (Colr) phase, and the dark texture of the Colh phase changed to a bright texture (Figure b) of the Colr phase at the transition. No change in the texture of the Colr phase was observed down to room temperature. Then, the phase transition temperatures were investigated by differential scanning calorimetry (DSC). The rac-1, (S)-1, and (R)-1 compounds had a Colh-Colr phase transition. In addition, (S)-1[31] and (R)-1 showed a phase between the Colh and Iso phases, and this phase was identified to be a nematic columnar (Ncol) phase by X-ray diffraction (XRD) measurements, although it was not observable in the POM results. A Ncol phase is a more disordered columnar phase, in which the columns are roughly parallel to each other and there is no long-range order.

Table 1

Phase Transition Temperatures of rac-1, (S)-1, and (R)-1a

compound	phase transition behaviors
rac-1
(S)-1
(R)-1

Colr: rectangular columnar, Colh: hexagonal columnar, Ncol: nematic columnar, and Iso: isotropic phases. The numbers above and below the arrows are the transition temperatures (°C), and the numbers in parentheses are the transition enthalpy changes (kJ mol–1), determined by DSC (temperature rate: 2 °C min–1) on the second heating and cooling cycle. The DSC measurements of these LCs were performed above room temperature, and no melting point was observed in the measurement range.

Figure 2

Microphotographs of rac-1 at (a) 151 °C (Colh phase) and (b) 148 °C (Colr phase) upon cooling. The arrows indicate the directions of the polarizers.

Investigation of the Molecular Packing Structure

The molecular packing structure of rac-1 was investigated by XRD. Upon heating, the XRD pattern of rac-1 at 144 °C (Figure a) showed d(200), d(110), d(400), d(600), and d(420) peaks, indicating a Colr phase (centered lattice) with lattice constants = 49.6 Å, b = 18.3 Å, c = 4.7 Å, and Z = 2.3. Upon further heating to 154 °C (Figure b), only one peak at 20.1 Å was observed. By considering its dendritic texture with sixfold symmetry, as shown in Figures a and S2, it was assigned to be a Colh phase (primitive lattice) with lattice constants = 23.2 Å, c = 4.7 Å, and Z = 1.2. The broad halo at around the d(100) peak indicates partially collapsed hexagonal order.

Figure 3

XRD profiles of rac-1 at (a) 144 °C (Colr phase) and (b) 154 °C (Colh phase). The lattice constant c (=4.7 Å) reported in our previous study[31] was used. The italic letters , b, and c are the lattice parameters. Z indicates the number of molecules in one unit lattice.

Electro-optical Characterization of rac-1

Compound rac-1 was introduced into an indium tin oxide (ITO) glass cell (cell gap: 5 μm, ITO area size: 1 cm × 1 cm, ITO was coated with a polyimide film) by the capillary action in its Iso phase (Figure a). Then, a rectangular wave voltage (200 Vpp with 1.0 Hz) was applied to the sample at 144 °C (Colr phase), and the texture of the sample on the ITO area immediately changed to a mosaic texture due to the change from a planar to homeotropic alignment (Figure b,c).

Figure 4

Electro-optic experiment of rac-1 at 144 °C (Colr phase): (a) experimental apparatus for observing polarity switching (cell gap: 5 μm, ITO area size: 1 cm × 1 cm, ITO was coated with a polyimide film); and microphotographs of the textures (b) before and (c) after applying a rectangular wave voltage for 5 min (frequency: 1.0 Hz, voltage: 200 Vpp). The white dashed line is the boundary between the ITO and non-ITO areas.

Investigation of the Macroscopic Polarity of Homeotropically Aligned rac-1

To investigate the macroscopic polarity of rac-1 in the Colr phase, second harmonic generation (SHG) measurement was performed, which is known as an established method to investigate the polarity in materials.[4,37] An overview of the SHG measurement system is shown in Figure a,b. By irradiation of an infrared laser beam [Nd:YAG (10 Hz), λ = 1064 nm] to the sample in the LC cell with macroscopic polarity, an SH wave (λ = 532 nm) from the sample is generated. The result of the SHG measurement is shown in Figure c. By applying a DC voltage (+100 VDC) for 1 min, the SHG intensity immediately increased. After removal of the voltage, the SHG intensity was maintained even after 6 h. Before applying the voltage, rac-1 showed planar textures in the POM results, indicating that the columns were planarly aligned to the substrate. After applying the voltage, each urea-carbonyl group of rac-1 responded to the voltage, the columns were aligned in parallel to the electric field, and the textures on the ITO area changed to the abovementioned mosaic texture because of the vertical alignment of the columns to the substrate. However, the switching current peak under applying a triangle wave volage (frequency: 1.0 Hz, voltage: 200 Vpp) was not detected because the switching response was slow.

Figure 5

SHG experiments of rac-1: (a) setup for SHG observation, (b) LC cell with the name of each part, and (c) plots of the SHG intensity of rac-1 at 144 °C (Colr phase). The voltage was controlled as follows: 0 V (3 min) → +100 VDC (1 min) → 0 V (360 min). (d) Setup for the SHG interferometry observation and (e) plots of SHG intensities of rac-1 at 144 °C (Colr phase) measured after applying +100 VDC (red filled circle) and −100 VDC (blue open square) for 1 min followed by removal of the voltage.

Subsequently, the direction of the macroscopic polarization in the sample was investigated by SHG interference experiments. In the SHG interference experiments, a quartz plate is placed next to the laser source to generate a standard SH wave, and a glass plate was set between the quartz plate and the sample to control the light path length of the laser and SHG lights (Figure d). The SH waves from the quartz and the polarized sample interfered with each other. Since the light path lengths of the laser and standard SH lights were changed by rotating the glass plate, plots of the SHG intensity against the glass plate angle formed an interferent curve depending on the polar direction of the sample. At 144 °C (Colr phase), after applying +100 VDC for 1 min followed by removal of voltage, the SHG intensity was measured with a detector by rotating the glass plate from 0 to 40° (Figure e, red filled circle). Then, at this temperature, after −100 VDC application for 1 min followed by removal of the voltage, the SHG intensity was measured using the same procedure (Figure e, blue open square). The two obtained interference curves showed line symmetry, which indicated that macroscopic polar directions of the sample in the LC cell were inversed by applying a DC voltage for 1 min.

By these two SHG experiments of rac-1 in the Colr phase, it was confirmed that the macroscopic polarity induced by applying the voltage was maintained even after removal of the voltage, and the polar direction could be switched by applying an external voltage. Although compound rac-1 is not optically active, it is clear that true ferroelectricity is realized in the Colr phase.

Furthermore, to investigate the coercive electric field (Ec) of rac-1 in the Colr phase, the SHG interferometry experiment was performed at 134 °C with setting the glass angle at 0o on applying a triangular voltage (voltage: 100 Vpp, frequency: 20 mHz, cell gap: 5 μm), and the SHG intensity was plotted against the applied voltage to give a hysteresis loop (Figure S3). The SHG intensity considerably changed at ±18 V, from which the Ec was calculated to be 3.6 V μm–1. This Ec value was smaller than that of (S)-1 (Ec = 4.12 V μm–1),[31] and the small Ec is a desirable feature for achieving ferroelectrics with low energy consumption. Achievement of the low Ec value suggests that the helical column of rac-1 includes several stereoisomers, which weakens the intracolumnar interactions.

Circular Dichroism Experiments of rac-1

To investigate the chiral induction in rac-1, the circular dichroism (CD) spectra of rac-1 were obtained. A chloroform solution of rac-1 (concentration: 3.5 mΜ) was added dropwise onto a quartz plate and spin-coated to produce a thin film on the quartz plate (film thickness: approximately 4 μm) at room temperature. In each CD measurement, 12 CD spectra were collected by rotating the sample by 30° around the center of the incident light, and then after turning over, 12 more CD spectra were collected in the same way. The resulting 24 CD signals obtained were averaged to minimize the effects of linear birefringence and linear dichroism. The CD spectrum of the thin film, measured at room temperature using incident light through an 8 mm diameter circular slit (Figure a), showed a very weak CD signal. In contrast, when a 1 mm diameter circular slit was used instead of the 8 mm diameter slit, as shown in Figure b, the intensity and sign of the CD signal measured varied from region to region. A comparison of Figure a,b showed that the CD signal obtained with an 8 mm diameter slit was the result of the many CD signals of smaller areas canceling each other out. This result strongly suggests that stereoisomers in the mixture (rac-1) were separated into opposite chiral domains by self-sorting because a similar CD observation in chiral self-sorting in hexagonal crystal phases was reported by Roche et al.[38] To demonstrate this hypothesis, we prepared a mixture of (S)-1 and (R)-1 at every 10 ratio from 0:100 to 100:0, placed their thin films (thickness: approximately 3 μm) on a quartz plate, and measured their CD spectra. As shown in Figure c,d, the intensity of the CD signals (slit diameter: 8 mm) changed in proportion to the enantiomeric excess. This result indicated that the majority rule[39] did not occur when forming helical columns because the CD signal intensity increased linearly with the ratio. In other words, this result meant that the (S)-1 and (R)-1 molecules were self-sorted and separately formed opposite-handedness helical columns,[40] and the columns with the same helicity were organized in a small domain. The DSC traces of (S)-1, (R)-1, (S)-1/(R)-1 mixture, and rac-1 are shown in Figure S4. Since the NCol phase was not observed in the DSC charts of the (S)-1/(R)-1 mixture, it is assumed that (S)-1 and (R)-1 are slightly miscible with each other. In addition, it is noteworthy that rac-1 showed a simple DSC chart with two sharp transition peaks, although it is a mixture of many stereoisomers.

Figure 6

CD measurements of the thin films of Colr phases: (a) CD spectrum of a rac-1 thin film using an 8 mm diameter slit, (b) CD spectra of several positions of a rac-1 thin film using a 1 mm diameter slit, (c) CD spectra of thin films of (S)-1/(R)-1 mixtures using an 8 mm diameter slit (red and blue arrows indicate the change in the CD signal with an increasing ratio of (S)-1 and (R)-1, respectively), (d) plot of the CD signal intensity at 255 nm against ee % of the (S)-1/(R)-1 mixture, (e) CD spectra of the thin film of (S)-1/Urea-10 mixture, using an 8 mm diameter slit (red arrow indicates the change in the CD signal with an increasing ratio of (S)-1), and (f) plot of the CD signal intensity at 255 nm against the molar ratio of (S)-1/Urea-10 mixture.

The results of the above CD experiments suggest that the chiral stereoisomers in rac-1 are self-sorted to form either left-handed [minus (M)] or right-handed [plus (P)] helical columns separately, and columns with the same helicity assembled into a chiral domain. However, it is necessary to explain the behavior of molecules with no chirality (or weak chirality), such as molecules with three (S)- and three (R)-citronellyl groups.

To determine the behavior of the achiral (or weakly chiral) molecules in the chiral molecules, achiral compound Urea-10 was mixed with (S)-1, and DSC and CD measurements of the mixtures were performed. In the DSC experiments (Figure S5), the phase transition temperatures of the mixtures were significantly different from those of pure (S)-1 and Urea-10. This result indicated that the two compounds were highly miscible. Figure e,f shows the CD signals of the 0:100–100:0 mixture of (S)-1/Urea-10 and the plot of their intensity at 255 nm, respectively. The CD signal intensities of the (S)-1/Urea-10 mixtures at greater than 50% were almost the same as that of pure (S)-1. This result suggests that the (M)-helical structures generated by the chiral molecules [(S)-1] were mostly maintained despite the introduction of a small amount of the achiral molecules (Urea-10), and these Urea-10 molecules were incorporated into the (M)-helical columns. This result showed contrast to the plot of the CD signal intensities of the (S)-1/(R)-1 mixtures, as shown in Figure d, in which the CD intensities were changed linearly against the ee %.

Estimation of the Packing of Helical Columns from the Electron Density Map

The resolution of the (M)- and (P)-helical columns to the chiral domains are explained by the chiral selective packing of the columns in the Colr phase of rac-1. Although the XRD profile of rac-1 in the Colr phase did not show any peaks based on the helical structure due to the rather low symmetry of the helicity, our previous study clarified that (S)-1 molecules stack in 30° increments to form an (M)-helical column structure.[31] Accordingly, it is assumed that (R)-1 molecules stack to form the mirror image, (P)-helical column structure. Figure a shows the molecular model of (S)-1. Figure b–d shows the top, oblique, and side views of the (M)-helical column, and the alkyl chain parts directing the same direction are colorized in the same color (red, orange, yellow, green, light blue, or dark blue) for clarity. As shown in Figure a,b, in the Colr phase, it is suggested that the alkyl parts of the neighboring two columns interdigitate in the b-axis direction. As shown in Figure c, the four (M)-helical column models are set on the electron density map of rac-1, and the models fit on the electron density map. Furthermore, since a slight interdigitation occurs even in the -axis direction, it is expected that packing of the columns with the same helicity is prioritized while synchronizing the helical periodicity because of steric interactions. Therefore, the (M)-helical columns aggregate to form a chiral domain consisting of only (M)-helical columns, while the (P)-helical columns form a chiral domain consisting of only (P)-helical columns. Furthermore, the non-chiral (or weak chiral) molecules are inserted in either the (P)- or (M)-helical columns. Thus, only one type of XRD pattern is observed in the Colr phase of rac-1.

Figure 7

Schematic representation of the (M)-helical column of 1: (a) molecule 1, (b) top, (c) oblique, and (d) side views of the column. The alkyl chain parts directing the same direction are colorized in the same color (red, orange, yellow, green, light blue, or dark blue).

Figure 8

Models for adjacent (M)-helical columns in the Colr phase of (S)-1 and the electron density map of rac-1: (a) top and (b) side views of the adjacent (M)-helical columns (dashed red circles show the intercolumnar interdigitation between alkyl parts in the b-axis direction), and (c) the four (M)-helical column models set on the electron density map of rac-1 (the color bar indicates the relative electron density obtained from its XRD peak intensities). The arrows and b indicate the -axis and b-axis directions of the unit lattice.

In the case of (S)-1, the formation of a helical column played an important role in enhancing the intracolumnar interactions, which were assumed to be enough to maintain the polar columnar structure.[31] As the chiral domains generated in rac-1 are similar to those of (S)-1 or (R)-1, rac-1 showed an AP-FCLC phase as well as enantiomerically pure (S)-1.

Conclusions

In this study, it was clarified that compound rac-1, which is a mixture of stereoisomers, is self-sorted to form (M)- and (P)-helical columns, and the columns with the same helicity were adjacent to each other as the chiral domain grew. In this way, rac-1 achieved the same level of ferroelectricity as chiral (S)-1 and (R)-1. The chiral self-sorting occurred in these simple molecules was surprising, in that 21 kinds of diastereomeric molecules recognized their structural similarities to each other, and that the molecules with similar stereochemistry gathered to generate a chiral column. Furthermore, it was suggested that achiral molecules were included in the chiral helical columns. Thus, it was also important that the stereochemistry of each achiral molecule was not strictly distinguished for the column formation in the LC phase.

Since racemic starting materials are cheaper and more easily available than chiral starting materials, the large-scale synthesis of racemic ureas is much easier than that of chiral ureas, allowing research for a variety of future applications.