*Article* **Effects of Tetrafluorocyclohexa-1,3-Diene Ring Position on Photoluminescence and Liquid-Crystalline Properties of Tricyclic** π**-Conjugated Molecules**

**Haruka Ohsato, Shigeyuki Yamada \*, Motohiro Yasui and Tsutomu Konno \***

Faculty of Molecular Chemistry and Engineering, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606-8585, Japan; m1673006@edu.kit.ac.jp (H.O.); myasui@kit.ac.jp (M.Y.) **\*** Correspondence: syamada@kit.ac.jp (S.Y.); konno@kit.ac.jp (T.K.)

**Abstract:** Tetrafluorocyclohexa-1,3-diene ring-containing tricyclic π-conjugated molecules are promising negative-dielectric-anisotropy guest species for vertical-alignment-type liquid-crystalline (LC) displays. Building on our previous work reporting the excellent photoluminescence (PL) properties of tricyclic π-conjugated molecules with central tetrafluorocyclohexa-1,3-diene rings, we herein synthesized four analogous molecules with terminal tetrafluorocyclohexa-1,3-diene rings from commercially available precursors and investigated the effects of substituent type and diene ring position on PL and LC properties using microscopic and spectroscopic methods. One of the prepared molecules exhibited a relatively planar molecular structure and formed herringbone-type aggregates via π/F and CH/π interactions instead of forming stacked aggregates via π/π stacking interactions, thus exhibiting relatively strong PL in solution and crystalline states. Moreover, the PL color of this compound depended on the electronic character of its terminal substituents along the long molecular axis. Of the four prepared species, two featured terminal ethyl groups and formed one or more LC phases. The PL properties of these phases indicated that the related phase transition induced changes in the aggregate structure, PL wavelength, and PL color. Our results expand the applicability of CF2CF2 moiety-containing tricyclic compounds as functional molecules for the fabrication of next-generation PL, LC, and PL-LC materials.

**Keywords:** fluorine; tetrafluorocyclohexa-1,3-diene; photoluminescence; liquid crystal; tricyclic molecule; aggregation

### **1. Introduction**

Fluorinated organic molecules have drawn much attention as the structural components of pharmaceuticals [1,2] and agrichemicals [3,4], as well as major constituents of liquid crystals [5–7] and optoelectronic materials [8,9]. This popularity is due to the unique properties of fluorine [10], namely, its highest electronegativity among all elements (4.0 on the Pauling scale), second smallest atomic radius (147 pm according to Bondi [11]), and the high dissociation energy of C–F bonds (105.4 kcal·mol<sup>−</sup>1). In view of these properties, the introduction of fluorine into molecular structures enhances latent functions or promotes the emergence of new ones and is, therefore, a powerful approach for the development of novel organic functional materials.

Our group has developed efficient and selective synthetic routes to various fluorinated organic molecules [12,13] including those exhibiting photoluminescence (PL) and liquidcrystalline (LC) properties [14]. The results obtained so far indicate that the introduction of fluorine atoms substantially increases PL intensity in the solid state and induces the emergence of mesophases between crystalline (Cry) and isotropic (Iso) phases.

Previously, we prepared a tricyclic molecule with a central tetrafluorocyclohexa-1,3 diene ring (**1a**) as a guest molecule with negative dielectric anisotropy to develop vertical alignment-type LC materials [15–18] and showed that **1a** exhibits blue PL in the crystalline

**Citation:** Ohsato, H.; Yamada, S.; Yasui, M.; Konno, T. Effects of Tetrafluorocyclohexa-1,3-Diene Ring Position on Photoluminescence and Liquid-Crystalline Properties of Tricyclic π-Conjugated Molecules. *Crystals* **2023**, *13*, 1208. https:// doi.org/10.3390/cryst13081208

Academic Editor: Ingo Dierking

Received: 10 July 2023 Revised: 28 July 2023 Accepted: 29 July 2023 Published: 3 August 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

and solution states. On this basis, we synthesized analogous tricyclic molecules with controlled electron density along the long molecular axis (**1b** and **1c**) and revealed that their PL behavior is greatly affected by the electron density distribution, which, in turn, is influenced by the electronic properties of terminal substituents (Figure 1) [19].

**Figure 1.** Previously synthesized tricyclic molecules **1a**–**c** with central tetrafluorocyclohexa-1,3-diene rings.

Building on the abovementioned results, we herein synthesized and characterized molecules **2a**–**d** to examine how PL and LC properties are affected by the position of the tetrafluorocyclohexa-1,3-diene ring in the tricyclic structure and the electronic properties of terminal substituents (Figure 2) [15,18].

**Figure 2.** Structures of tricyclic molecules **2a**–**d** with terminal tetrafluorocyclohexa-1,3-diene rings.

### **2. Materials and Methods**

### *2.1. General Characterization*

Melting points (*T*m) were measured on a Shimadzu DSC-60 Plus instrument using at least three heating/cooling cycles at a scan rate of 5.0 ◦C·min−1. 1H and 13C nuclear magnetic resonance (NMR) spectra were recorded on a Bruker AVANCE III 400 spectrometer (1H: 400.13 MHz, 13C: 100.61 MHz) in chloroform-*d* (CDCl3). Chemical shifts were reported on the basis of the residual proton or carbon signal of CHCl3 (*δ*<sup>H</sup> = 7.26 ppm, *δ*<sup>C</sup> = 77 ppm) in parts per million (ppm). 19F-NMR (376.46 MHz) spectra were recorded on the Bruker AVANCE III 400 spectrometer in CDCl3 using trichlorofluoromethane (CFCl3, *δ*<sup>F</sup> = 0.00 ppm) as an internal standard. Infrared (IR) spectra were acquired using the KBr method on a JASCO FT/IR-4100 type A spectrometer. High-resolution mass spectra (HRMS) were recorded on a JEOL JMS-700MS spectrometer using fast atom bombardment (FAB+) methods. Column chromatography was performed using Wakogel® 60N (38–100 μm), and thin-layer chromatography was performed using the corresponding silica gel plates (silica gel 60F254, Merck, Darmstadt, Germany).

### *2.2. Materials*

The target molecules were synthesized according to a previously reported method from readily available precursors, namely, dimethyl 2,2,3,3-tetrafluorosuccinate (**2a** and **2d**; Scheme 1a) [18] and 4-bromo-3,3,4,4-tetrafluorobut-1-ene (**2b** and **2c**, Scheme 1b) [15].

Detailed synthetic procedures are provided in Schemes S1 and S2, and Figures S1–S42 shown in the Supplementary Materials. Characterization data are presented below (for **2a**–**d**) and in the Supplementary Materials (for other molecules).

### 2.2.1. 4-Ethyl-5,5,6,6-tetrafluoro-1-[4-(4-n-propylphenyl)phenyl]cyclohexa-1,3-diene (**2a**)

Yield: 90% (0.25 g, 0.67 mmol); yellow solid; *T*m: 92 ◦C; 1H-NMR (CDCl3): *δ* 0.98 (t, *J* = 7.2 Hz, 3H), 1.20 (t, *J* = 7.6 Hz, 3H), 1.69 (sext, *J* = 7.6 Hz, 2H), 2.40 (q, *J* = 7.2 Hz, 2H), 2.64 (t, *J* = 7.2 Hz, 2H), 6.09 (d, *J* = 6.0 Hz, 1H), 6.39 (d, *J* = 6.0 Hz, 1H), 7.27 (d, *J* = 7.8 Hz, 2H), 7.53 (d, *J* = 8.4 Hz, 4H), 7.62 (d, *J* = 7.6 Hz, 2H); 13C-NMR (CDCl3): *δ* 11.4, 13.8, 21.6, 24.5, 37.7, 114.0 (tt, *J* = 251.87, 26.82 Hz), 114.1 (tt, *J* = 251.9, 26.8 Hz), 123.1 (t, *J* = 9.2 Hz), 125.8 (t, *J* = 8.79 Hz), 126.8, 126.9, 127.4, 129.0, 131.7, 133.6 (t, *J* = 22.0 Hz), 137.6, 137.8 (t, *J* = 21.9 Hz), 141.5, 142.3; 19F-NMR (CDCl3): *<sup>δ</sup>* −126.57 (d, *<sup>J</sup>* = 4.76 Hz, 2F), −122.23 (d, *<sup>J</sup>* = 4.86 Hz, 2F). The above characterization data were consistent with those reported previously [15,18].

### 2.2.2. 5,5,6,6-Tetrafluoro-1-(4-methoxyphenyl)phenylcyclohexa-1,3-diene (**2b**)

Yield: 80% (0.53 g, 1.6 mmol); white solid; *T*m: 130 ◦C; 1H-NMR (CDCl3): *δ* 3.86 (s, 3H), 6.03–6.12 (m, 1H), 6.37–6.46 (m, 2H), 7.00 (d, *J* = 8.8 Hz, 2H), 7.51–7.63 (m, 6H); 13C-NMR (CDCl3): *δ* 55.3, 110.0–116.0 (m, 2C of *C*F2*C*F2), 114.3, 122.9 (t, *J* = 25.8 Hz), 124.7 (t, *J* = 8.0 Hz), 126.7, 127.7, 128.1, 129.9 (t, *J* = 11.8 Hz), 130.9, 132.6, 136.4 (t, *J* = 22.0 Hz), 141.6, 159.5; 19F-NMR (CDCl3): *<sup>δ</sup>* −121.29 (s, 2F), −121.67 (s, 2F); IR (KBr): *<sup>ν</sup>* 3026, 2968, 2844, 1649, 1604, 1576, 1530, 1445, 1399, 1312, 1289, 1202, 1183, 1021, 1011, 879, 787 cm−1; HRMS (FAB) calculated for C19H14F4O [M]+: 334.0980, found: 334.0980. Crystal data for C19H14F4O (*M* = 334.30 g/mol): orthorhombic, space group *P* 21 21 21, *a* = 5.5586(7) Å, *b* = 9.2038(15) Å, *c* = 29.282(4) Å, *α* = 90◦, *β* = 90◦, *γ* = 90◦, *V* = 1498.1(4) Å3, *Z* = 4, *T* = 173 K, *μ*(MoKα) = 0.710 mm<sup>−</sup>1, *D*calc = 1.482 g/cm3, 98,894 reflections measured (3.042◦ ≤ 2*θ* ≤ 27.480◦), 7212 unique (*R*int = 0.0476, *R*sigma = 0.0950), which were used in all calculations. The final *R*<sup>1</sup> was 0.0631 (*I* > 2*σ*(*I*)) and *wR*<sup>2</sup> was 0.1269 (all data).

### 2.2.3. 5,5,6,6-Tetrafluoro-1-{4-(trifluoromethyl)phenyl}phenylcyclohexa-1,3-diene (**2c**)

Yield: 70% (0.15 g, 1.4 mmol); white solid; *T*m: 138 ◦C; 1H-NMR (CDCl3): *δ* 6.06–6.18 (m, 1H), 6.40–6.54 (m, 2H), 7.60 (d, *J* = 8.8 Hz, 2H), 7.64 (d, *J* = 8.8 Hz, 2H), 7.72 (s, 4H); 13C-NMR (CDCl3): *δ* 112.7 (tt, *J* = 249.4, 27.2 Hz), 113.4 (tt, *J* = 253.0, 26.5 Hz), 124.2 (q, *J* = 272.1 Hz), 123.4 (t, *J* = 25.6 Hz), 125.5 (t, *J* = 8.0 Hz), 125.8 (q, *J* = 3.6 Hz), 127.3, 127.4, 127.9, 129.7 (q, *J* = 32.3 Hz), 129.8 (t, *J* = 11.8 Hz), 132.6, 136.1 (t, *J* = 22.7 Hz), 140.4, 143.6; 19F-NMR (CDCl3): *<sup>δ</sup>* −62.43 (s, 3F), −121.31 (s, 2F), −121.72 (s, 2F); IR (KBr): *<sup>ν</sup>* 3088, 2362, 1919, 1690, 1616, 1502, 1425, 1274, 1210, 968, 875, 794, 739, 729 cm<sup>−</sup>1; HRMS (FAB) calculated for C19H11F7 [M]+: 372.0749, found: 372.0759.

### 2.2.4. 4-Ethyl-5,5,6,6-tetrafluoro-1-[4-{4-(n-octyloxy)phenyl}phenyl]cyclohexa-1,3-diene(**2d**)

Yield: 83% (0.32 g, 0.69 mmol); pale-yellow solid; *T*m: 71 ◦C; 1H-NMR (CDCl3): *δ* 0.90 (t, *J* = 6.8 Hz, 3H), 1.19 (t, *J* = 7.2 Hz, 3H), 1.26–1.42 (m, 8H), 1.48 (quin, *J* = 8.0 Hz, 2H), 1.81 (quin, *J* = 7.2 Hz, 2H), 2.40 (q, *J* = 7.2 Hz, 2H), 4.00 (t, *J* = 6.8 Hz, 2H), 6.09 (d, *J* = 6.0 Hz, 1H), 6.38 (d, *J* = 6.0 Hz, 1H), 6.98 (d, *J* = 8.8 Hz, 2H), 7.48–7.62 (m, 6H); 13C-NMR (CDCl3):*δ* 11.5, 14.1, 21.7, 22.7, 26.1, 29.26, 29.31, 29.4, 31.8, 68.1, 110–125 (m, 2C of *C*F2*C*F2), 114.9, 123.1 (t, *J* = 8.8 Hz), 125.6 (t, *J* = 8.8 Hz), 126.7, 127.5, 128.0, 131.3, 132.6, 133.7 (t, *J* = 23.1 Hz), 137.8 (t, *J* = 21.2 Hz), 141.3, 159.1; 19F-NMR (CDCl3): *δ* –123.55 (s, 2F), –127.90 (s, 2F); IR (KBr): *ν* 3038, 2926, 2852, 1885, 1654, 1606, 1579, 1529, 1500, 1253, 1132, 907, 864 cm<sup>−</sup>1; HRMS (FAB) calculated for C28H32F4O [M]+: 460.2389, found: 460.2382.

### *2.3. Single-Crystal X-ray Diffraction (XRD)*

Single-crystal XRD patterns were recorded on an XtaLAB AFC11 diffractometer (Rigaku, Tokyo, Japan). The reflection data were integrated, scaled, and averaged using CrysAlisPro software (v. 1.171.39.43a; Rigaku Corporation, Akishima, Japan), and empirical absorption corrections were applied using the SCALE 3 ABSPACK scaling algorithm (CrysAlisPro). Structures were identified using a direct method (SHELXT-2018/2 [20]), refined using a full-matrix least-squares method (SHELXL-2018/3 [21]), and visualized using OLEX2 [22]. The crystallographic data were deposited in the Cambridge Crystallographic Data Center (CCDC) database (CCDC 2269760 for **2b**) and can be obtained free of charge

from the CCDC, 12 Union Road, Cambridge CB2 1EZ, UK; Fax: +44-1223-336033; e-mail: deposit@ccdc.cam.ac.uk.

### *2.4. Photophysical Properties*

JASCO V-750 absorption (JASCO, Tokyo, Japan) and FP-6600 fluorescence (JASCO, Tokyo, Japan) spectrometers were used to acquire solution-phase ultraviolet/visible (UV/vis) absorption and PL spectra. A Quantaurus-QY C11347-01 instrument (Hamamatsu Photonics, Hamamatsu, Japan) was used for PL quantum yield measurements, and a Quantaurus-Tau fluorescence lifetime spectrometer (C11367-34; Hamamatsu Photonics, Japan) was employed for PL lifetime determination.

### *2.5. LC Properties*

Polarizing optical microscopy (POM) measurements were carried out using an Olympus BX53 microscope (Tokyo, Japan) equipped with cooling and heating stages (10,002 L, Linkam Scientific Instruments, Surrey, UK) to assess LC properties. Thermodynamic properties were assessed using differential scanning calorimetry (DSC; DSC-60 Plus, Shimadzu, Kyoto, Japan) at heating and cooling rates of 5.0 ◦C·min−<sup>1</sup> under N2. Variable-temperature powder X-ray diffraction (VT-PXRD) analyses were carried out using an X-ray diffractometer (Rigaku, MiniFlex600, Tokyo, Japan) equipped with an X-ray tube (Cu *K*α, *λ* = 1.54 Å) and semiconductor detector (D/teX Ultra2). The sample powder was mounted on a nonreflecting silicon plate set on a benchtop stage (Anton Paar, BTS-500). The temperature, heating/cooling rate, and X-ray exposure time were controlled.

### *2.6. Theoretical Calculations*

All computations were performed using the Gaussian 16 program set [23] with density functional theory (DFT) at the level of the M06-2X hybrid functional [24] and the 6-31+G(d) (for all atoms) basis set with a conductor-like polarizable continuum model (CPCM) [25] for CHCl3. Theoretical vertical transitions were calculated using the time-dependent DFT (TD-DFT) method at the same theoretical level using the same solvation model.

### **3. Results and Discussion**

### *3.1. Synthesis*

Compounds **2a** and **2d**, featuring an ethyl group attached to the longitudinal molecular terminal, were synthesized from the readily available dimethyl 2,2,3,3-tetrafluorosuccinate according to a reported procedure (Scheme 1a) [18]. The reaction of dimethyl tetrafluorosuccinate with 4-(4-*n*-propylphenyl)phenylmagnesium bromide in THF at −78 ◦C overnight followed by hydrolysis under acidic conditions afforded ketoester **3a** in 64% yield. Compound **3a** was treated with 3.6 equivalents of vinylmagnesium chloride in Et2O, and the reaction mixture was stirred overnight at reflux to afford 4,4,5,5-tetrafluoroocta-1,7 diene (**4a**) in 37% yield. In the presence of a second-generation Grubbs catalyst, the ring-closing metathesis of **4a** in CH2Cl2 (40 ◦C, 24 h) furnished 1-aryl-4-ethyl-5,5,6,6 tetrafluorocyclohex-2-ene-1,4-diol (**5a**) in 49% yield. The 24 h exposure of **5a** in methanol to H2 at room temperature resulted in catalytic hydrogenation and furnished 1-aryl-4-ethyl 2,2,3,3-tetrafluorocyclohexan-1,4-diol (**6a**) in 70% yield. Subsequent dehydration with phosphoryl chloride in pyridine at 90 ◦C for 24 h produced **2a** in 90% yield. The octyloxy chain-bearing structural analog **2d** was prepared by a similar procedure starting with the addition of 4-(4-octyloxyphenyl)phenylmagnesium bromide.

Compound **2b**, featuring an electron-donating methoxy group, and **2c**, featuring an electron-withdrawing trifluoromethyl (CF3) group at the longitudinal molecular end, were synthesized according to a previously reported procedure (Scheme 1b) [15]. The Barbier-type nucleophilic addition of 1,1,2,2-tetrafluorobut-3-enyllithium (prepared in situ from 4-bromo-3,3,4-4-tetrafluorobut-1-ene and LiBr-free MeLi) to *p*-anisaldehyde in tetrahydrofuran (THF) at −78 ◦C for 2 h gave tetrafluorohomoallyl alcohol **7b** in 70% yield. The oxidation of **7b** with Oxone® in the presence of sodium 2-iodobenzenesulfonate

(*pre*-IBS; 5 mol%) in acetonitrile at 90 ◦C for 16 h afforded 1-aryl-2,2,3,3-tetrafluoropent-4 en-1-one (**8b**) in 77% yield. Compound **8b** was treated with allylmagnesium bromide in THF at −78 ◦C for 2 h to produce 4-aryl-5,5,6,6-tetrafluoroocta-1,7-diene-4-ol (**9b**) in 58% yield. Compound **9b** underwent ring-closing metathesis upon treatment with a secondgeneration Grubbs catalyst (3 mol.%) to furnish 4-aryl-5,5,6,6-cyclohex-1-en-4-ol (**10b**) in 73% yield. The dehydration of **10b** with phosphoryl chloride in pyridine at 90 ◦C for 24 h produced the target methoxy-substituted species (**2b**) in 80% yield. The CF3-substituted **2c** was synthesized using the same reaction sequence.

Compounds **2a**–**d** were purified by column chromatography (eluent: hexane/EtOAc = 3/1 for **2a** or 10/1 for **2b**–**d**) and recrystallization from a 1:1 (*v*/*v*) mixture of CH2Cl2 and hexane. The molecular structures of the target molecules were confirmed by NMR spectroscopy, IR spectroscopy, and HRMS, and the related purities were sufficient for photophysical and LC property analyses.

Among **2a**–**d**, only the methoxy-substituted **2b** furnished single crystals appropriate for X-ray crystallographic analysis upon recrystallization, whereas **2a**, **2c**, and **2d** did not furnish single crystals even after multiple recrystallizations. Figure 3 shows the crystal structure of **2b** obtained by X-ray structure analysis.

**Figure 3.** (**a**) Molecular structure and (**b**,**c**) packing of **2b** in the crystalline lattice. Display notation: space-filling model for rearmost molecules, ball-and-stick model for middle molecules, and wireframe model for frontmost molecules.

Compound **2b** crystallized in an orthorhombic system (*P* 21 21 21 space group) and featured a unit cell with four molecules. The dihedral angle between the two aromatic rings of the biphenyl moiety was approximately 4◦, and that between the tetrafluorocyclohexa-1,3-diene ring and the biphenyl moiety was approximately 19◦ (Figure 3a). In **1b**, which has a central tetrafluorocyclohexa-1,3-diene ring, the dihedral angle between the cyclohexa-1,3-diene ring and the adjacent aromatic ring was at least 31◦ [19]. On the basis of the molecular structures of **1b** and **2b**, we concluded that the change in the position of the tetrafluorocyclohexa-1,3-diene ring from central to terminal favored a more planar structure. The space-filling model representation in Figure 3b suggests that the π/F interactions [26,27] between the π-electrons and F atoms of tetrafluorocyclohexa-1,3-diene resulted in the formation of a stacked structure along the *<sup>a</sup>*-axis. The C(sp2)···F interatomic distance corresponding to the π/F interaction (304.4 pm) was shorter than the sum of van der Waals radii (317 pm) of carbon (170 pm) and fluorine (147 pm) atoms [10]. The molecule

represented by the space-filling model formed molecular packings featuring two pairs of CH/π interactions [28] with the molecule represented by the ball-and-stick model along the *<sup>b</sup>*-axis direction. The C(sp2)···H interatomic distance corresponding to the CH/<sup>π</sup> interaction worked (284.5 pm) was also shorter than the sum of the van der Waals radii (290 pm) of carbon (170 pm) and hydrogen (120 pm). In addition to the short distance between the C(sp2) and H atoms, the carbon atom of the methoxy group was in close contact with the fluorine atom at a distance (302.3 pm) shorter than the sum of carbon (170 pm) and fluorine (147 pm) van der Waals radii. The molecule represented by the ball-and-stick model also formed a stacked structure with the molecule represented by the wire-frame model along the *a*-axis via CH/π interactions (short contact: 286.7 pm) and O/H hydrogen bonds (short contact: 270.5 pm) (Figure 3c). Accordingly, herringbone-type packing structures were formed through multiple intermolecular interactions. However, unlike the packing structure of **1b** [19], which features a central cyclohexa-1,3-diene ring, the packing structure of **2b** did not feature intermolecular π/π stacking.

### *3.2. Photophysical Properties*

Figure 4 shows the UV/vis absorption spectra, PL spectra, and PL color chromaticity diagrams (as defined by the Commission Internationale de l'Eclailage (CIE)) of **2a**–**d**, and Table 1 lists the related photophysical data.

**Figure 4.** (**a**) Ultraviolet/visible absorption spectra (concentration: 1.0 <sup>×</sup> 10–5 mol·L<sup>−</sup>1) and (**b**) photoluminescence (PL) spectra (concentration: 1.0 <sup>×</sup> <sup>10</sup>–6 mol·L−1) of **2a**–**<sup>d</sup>** measured in chloroform (CHCl3). Inset: photographs of PL in CHCl3 solution under UV irradiation (*λ*ex = 365 nm). (**c**) Commission Internationale de l'Eclailage (CIE) chromaticity diagram for PL colors of **2a**–**d**.


**Table 1.** Photophysical data of **2a**–**d** in CHCl3 solution.

<sup>1</sup> Concentration: 1.0 × <sup>10</sup>−<sup>5</sup> mol·L−1. <sup>2</sup> Concentration: 1.0 × <sup>10</sup>−<sup>6</sup> mol·L−1. <sup>3</sup> Measured using an integrating sphere. <sup>4</sup> Radiative deactivation rate constant (*k*r) = *Φ*PL/*τ*. <sup>5</sup> Nonradiative deactivation rate constant (*k*nr) = (1 – *Φ*PL)/*τ*.

Compound **2a**, possessing ethyl and *n*-propyl substituents at longitudinal molecular terminals, exhibited a single absorption band with a maximum absorption wavelength (*λ*abs) of ~330 nm in CHCl3. Compound **2b**, featuring a strongly electron-donating methoxy group, exhibited a red-shifted *λ*abs of 337 nm, whereas **2c**, featuring a strongly electronwithdrawing CF3 group, exhibited a blue-shifted *λ*abs of 320 nm. In CHCl3, the *λ*abs of **2d** with ethyl and *n*-octyloxy groups as longitudinal terminal substituents was 337 nm, i.e., equal to that of **2b**.

The theoretical vertical transition was modeled using Gaussian software [23] with time-dependent density functional theory (TD-DFT). Figure 5 shows the distributions of the highest occupied molecular orbitals (HOMOs) and lowest unoccupied molecular orbitals (LUMOs) for **2a**–**d**, with the related theoretical data summarized in Table 2. The detailed orbital distributions are shown in Figures S44–47 in Supplementary Materials.


**Table 2.** Theoretical data of **2a**–**d** obtained using Gaussian software with time-dependent density functional theory 1.

<sup>1</sup> Calculated at the M06-2X/6-31+G(d) level of theory using a conductor-like polarizable continuum model for CHCl3.

According to Figure 5, the HOMOs of **2a**–**d** were spread throughout the π-conjugated structure, whereas the LUMOs were localized on the tetrafluorocyclohexa-1,3-diene ring of the tricyclic π-conjugated framework. Substituents at longitudinal molecular ends affected the HOMO and LUMO energies, e.g., electron-donating substituents such as alkoxy groups increased the HOMO energy, whereas electron-withdrawing substituents had the opposite effect. The alkoxy group at the opposite end did not affect the energy of the LUMO, as this orbital was localized on the tetrafluorocyclohexa-1,3-diene ring, whereas the electron-donating ethyl group introduced into the tetrafluorocyclohexa-1,3-diene skeleton increased the LUMO energies of **2a** and **2d**. The absorption wavelengths (*λ*calcd) of **2a**–**d** determined by TD-DFT calculations (331 nm for **2a**, 338 nm for **2b**, 325 nm for **2c**, and 335 nm for **2d**) were close to the measured *λ*abs values listed in Table 1. The transitions from the ground to the first excited states were calculated to be of π–π\* HOMO→LUMO and HOMO–1/HOMO–2→LUMO types.

When a solution of **2a** in CHCl3 was excited by irradiation with UV light at *λ*abs (330 nm), a single PL band with a maximum PL wavelength (*λ*PL) of approximately 437 nm was observed (Figure 4b). Compared to that of **2a**, the PL band of **2b** with an electrondonating methoxy group (*λ*PL = 463 nm in CHCl3) was red-shifted by 26 nm, whereas the PL band of **2c** with an electron-withdrawing CF3 group (*λ*PL = 416 nm) was substantially blue-shifted. Similar to the methoxy-substituted **2b**, the *n*-octyloxy-substituted **2d** exhibited PL (*λ*PL = 463 nm). In the case of **2c** with a large HOMO–LUMO overlap, radiative deactivation probably occurred from the locally excited state, whereas **2b** or **2d** with a locally existing LUMO luminesced through the radiative deactivation of the intramolecular charge transfer (ICT) excited state, which can be reasonably explained by the Lippert– Mataga plot [29,30] shown in Figure S49.

The high-energy PL of **2c** corresponded to dark-blue color represented by CIE chromaticity coordinates of (*x, y*) = (0.157, 0.036) (Figure 4c). In contrast, the low-energy PL of **2b** and **2d** emitted from ICT states corresponded to light-blue color with CIE coordinates of (*x, y*) = (0.153, 0.189). The quantum yields (*Φ*PL) and PL lifetimes (*τ*) of **2a**–**d** were determined as 0.24–0.94 and ~2.11 ns, respectively. This value of *τ* indicates that the light emitted by **2a**–**d** was fluorescent. Among the four compounds, **2c** exhibited the lowest *Φ*PL (0.24) and a very short *τ* (<1.0 ns). The radiative (*k*r) and nonradiative (*k*nr) deactivation rate constants of **2c** were calculated from *<sup>Φ</sup>*PL and *<sup>τ</sup>* as 3.30 × 108 <sup>s</sup>−<sup>1</sup> and 10.44 × 108 <sup>s</sup><sup>−</sup>1, respectively. Notably, *k*r was not significantly different between **2a** and **d**, whereas the *k*nr of **2c** was 5–37 times higher than those of other derivatives. These results suggested the occurrence of fluorescence reabsorption (self-absorption) in **2c**, which resulted in decreased *Φ*PL and increased *k*nr.

Most molecules exhibiting luminescence in solution generally experience luminescence quenching through intermolecular energy transfer at high concentrations or in the solid state. However, **2a**–**d** exhibited strong luminescence even in the crystalline state. Figure 6 shows the PL spectra of crystalline **2a**–**d**, the related CIE chromaticity diagram, and photographs of crystals under 365 nm UV light. The corresponding photophysical data are summarized in Table 3.

**Figure 6.** (**a**) PL spectra of crystalline **2a**–**d**. (**b**) CIE chromaticity diagram and photographs of **2a**–**d** crystals under 365 nm ultraviolet light.

**Table 3.** Photophysical data of crystalline **2a**–**d**.


<sup>1</sup> Measured using an integrating sphere. <sup>2</sup> Radiative deactivation rate constant (*k*r) = *Φ*PL/*τ*. <sup>3</sup> Nonradiative deactivation rate constant (*k*nr) = (1 − *Φ*PL)/*τ*.

Crystalline **2a** with two alkyl groups at longitudinal molecular ends exhibited green PL with a single PL band at *λ*PL around 509 nm, which was red-shifted relative to the value in CHCl3 solution by 72 nm. However, the PL behavior of **2b**–**d** did not substantially change upon the transition from the CHCl3 solution to the crystalline state. In the crystalline state, **2b** crystallized mainly via CH/π, π/F, and hydrogen bonds; π/π stacking between the intermolecular aromatic rings was not observed. The similarity between the *λ*PL and *Φ*PL values observed in the crystalline state and CHCl3 solution was ascribed to the absence of π/π stacking interactions in crystalline **2b**, which suppressed the nonradiative deactivation induced by the formation of molecular aggregates. Given that crystalline **2c** and **2d** also exhibited PL behavior similar to that in the CHCl3 solution state, we concluded that their conjugated structures were also not involved in intermolecular interactions, although their crystal structures have not yet been elucidated. On the other hand, we inferred that crystalline **2a**, which featured a PL wavelength and PL color different from those observed in the solution state, interacted with the π-conjugated site through the formation of molecular aggregates, unlike in the dilute solution, although the crystal structure of **2a** also remained veiled. PL lifetime measurements showed that the *τ* of crystalline **2a**–**d** was 1.84–3.19 ns and, therefore, also indicative of fluorescence. The PL decays of **2a**, **2c**, and **2d** were well modeled by a mono-exponential function, and the related PL originated from a single excited state. In contrast, the PL decay of **2b** was fitted by a biexponential function assuming a radiative deactivation pathway from any two excited states, although the related excited-state details remain unknown.

Compared with the previously reported **1b** and **1c** with a central tetrafluorocyclohexa-1,3-diene ring [19], **2b** and **2c** featured a shorter (by 20–25 nm) *λ*abs in CHCl3, which was

ascribed to the significantly increased LUMO level of the latter molecules. However, *λ*PL was found to be almost the same, except for **2c**, which had a CF3 group at the molecular terminal. In the crystalline state, the *λ*PL of **2c** was blue-shifted relative to that of **1c**, although almost identical *λ*PL values were observed for **2b** and **1b**. The CHCl3 solution-phase *Φ*PL values of **2b** and **2c** exceeded those of **1b** and **1c**. In contrast, the opposite trend was observed in the crystalline state, i.e., the *Φ*PL values of **2b** and **2c** were lower than those of **1b** and **1c**. In **2b** and **2c**, which greatly differ from **1b** and **1c** [19], the biphenyl moiety was planar and formed a herringbone structure because of CH/π interactions. We concluded that weak intermolecular interactions did not lead to molecular motion suppression, resulting in decreased *Φ*PL. Accordingly, the positional change of the tetrafluorocyclohexa-1,3-diene ring in the tricyclic scaffold had a relatively large effect on the crystalline-state behavior, and the position of this ring affected intermolecular interactions and, hence, the extent of molecular motion inhibition and *Φ*PL.

### *3.3. LC Properties*

The previously reported **1a**–**c** exhibited transitions only between their Cry and Iso phases upon heating and cooling, i.e., no LC phases were observed [15]. To understand how the position of the tetrafluorocyclohexa-1,3-diene ring in tricyclic molecules affects their LC properties, we used POM and DSC to examine the LC behavior of **2a**–**d**, which showed PL in both dilute solution and crystalline states (Figures S53–S56 in Supplementary Materials). Compounds **2b** and **2c** exhibited only a Cry→Iso phase transition but did not form any mesophase upon heating and cooling. In contrast, for **2a** and **2d**, a fluid bright-field POM image was observed between the Cry and Iso phases, indicating the formation of an LC phase upon cooling (**2a**) or heating/cooling (**2d**). Figure 7 shows the DSC curves of **2a** and **2d** and the POM images of the corresponding mesophases. Table 4 lists the phase transition behaviors of **2a**–**d**, namely, their phase sequences, as well as phase transition temperatures and enthalpies in the second heating and cooling processes.

**Figure 7.** Differential scanning calorimetry (DSC) curves of (**a**) **2a** and (**b**) **2d** recorded during the second heating and cooling processes at a scan rate of 5 ◦C·min−<sup>1</sup> under N2. Polarizing optical microscopy textures in the mesophases of (**c**) **2a** and (**d**) **2d**.


**Table 4.** Phase transition data of **2a**–**d** during the second heating and cooling processes.

<sup>1</sup> Determined by DSC (scan rate: 5 ◦C·min−1, atmosphere: N2). Abbreviations: Cry, crystal; Iso, isotropic; N, nematic; SmA, smectic A; SmC, smectic C phase.

In the case of the **2a** mesophase, POM revealed that a fluid four-brush Schlieren texture formed at 90 ◦C after the slow cooling from the dark-field-image Iso phase. Given that POM indicated the formation of a nematic (N) phase with only orientational order, the mesophase appearing during the cooling of **2a** was classified as the N phase. Further cooling from the N-phase state of **2a** resulted in fluidity loss at 81 ◦C and a phase transition to the hard Cry phase. In the case of the **2d** mesophase, the nonfluidic bright-field POM image corresponding to the Cry phase changed to a fluidic fan-shaped POM image at 71 ◦C upon heating. Further heating induced an optical texture change to a Schlieren-patterned N phase at 110 ◦C followed by a phase transition to the Iso phase in the dark-field POM image at 133 ◦C. Upon cooling, the N-phase Schlieren texture appeared at 136 ◦C, and a transition to a phase with a fan-shaped texture occurred at 114 ◦C. Upon further cooling, a broken fan-shaped texture was observed at 68 ◦C, followed by a phase transition to the nonfluidic Cry phase at 40 ◦C. The fan-shaped optical texture observed in the mesophase of **2d** is characteristic of the smectic (Sm) phase, which has an orientational and positional order. Notably, in the case of **2d**, the Sm phase appeared at a lower temperature than the N phase.

Further insights into the LC phases exhibited by **2a** and **2d** were provided by VT-PXRD measurements. The pattern of **2a** recorded after cooling from the Iso phase and holding at 70 ◦C featured no Cry phase peaks but contained a halo peak centered around 2*θ* = 18◦ (Figure S57). This result strongly suggests that the mesophase appearing in **2a** is the N phase without positional order. PXRD measurements were also performed for **2d** at 124, 89, and 49 ◦C after cooling from the Iso phase. A halo peak centered around 18◦ was also observed in the pattern recorded at 124 ◦C, and the mesophase appearing at this temperature was determined to be the N phase (Figure S57). The PXRD pattern recorded at 89 ◦C featured a sharp peak at 3.75◦ and a weak peak at 7.45◦ (Figure 8a).

**Figure 8.** Powder X-ray diffraction patterns of **2d** recorded at (**a**) 89 and (**b**) 49 ◦C.

These diffraction peaks corresponded to the plane indices of (*hkl*) = (001) and (002). The peak at 3.75◦ in the low-angle region corresponded to a *d*-spacing of 2.35 nm, according to Bragg's equation, which was consistent with the longitudinal molecular length of **2d** (Figure 8a) This consistency of the interlayer distance with the molecular length agreed with the formation of a smectic A (SmA) phase with a layered periodic structure wherein the long molecular axis was oriented in the direction of the layer normal. In the pattern recorded at 49 ◦C, the peak of the (001) plane appeared at 3.95◦ and corresponded to a *d*-spacing of 2.23 nm, which was shorter than the molecular length along the long molecular axis (2.35 nm) (Figure 8b). This result indicated the presence of a smectic C (SmC) phase featuring a tilt angle with respect to the layer normal (Figure 8b).

### *3.4. PL Properties of 2d in Various Molecular Aggregation States*

Compound **2d**, which forms various mesophases, was selected to investigate PL behavior changes associated with the phase transition-induced alterations in molecular aggregate structure. PL behavior was examined using a fluorescence spectrometer equipped with a self-made temperature control unit. The samples were cooled from the Iso phase and held for 5 min at each temperature during cooling. Figure 9 shows the thus obtained PL spectra and CIE chromaticity diagrams, and Table 5 summarizes the related photophysical data.

**Figure 9.** (**a**) PL spectra of **2d** recorded at different temperatures upon cooling. (**b**) CIE chromaticity diagram for PL color of **2d** at different temperatures.



<sup>1</sup> PL intensity *I* of each phase with respect to the PL intensity (*I*N) of the N phase.

In the case of **2d**, a PL band with *λ*PL ≈ 466 nm appeared in the N phase; however, the related PL intensity (*I*N) decreased because of the accelerated nonradiative deactivation by micro-Brownian motion upon heating. The N→SmA phase transition observed upon cooling induced a 2.0-fold PL intensity increase (*I*/*I*<sup>N</sup> = 2.0) along with a slight red shift in *λ*PL. No significant change was observed in *λ*PL or PL intensity upon the transition to the SmC phase, whereas the transition to the Cry phase induced a blue shift of *λ*PL by 12 nm and a 7.8-fold increase in PL intensity relative to the N phase (*I*/*I*<sup>N</sup> = 7.8). The CIE chromaticity diagram shown in Figure 9b demonstrates that the PL color of **2d** changed from dark blue to light blue owing to the phase transition-induced alteration of the molecular aggregate structure.

### **4. Conclusions**

Tricyclic π-conjugated molecules with terminal tetrafluorocyclohexa-1,3-diene rings and different substituents introduced at the longitudinal molecular ends (**2a**–**d**) were synthesized in five steps from dimethyl 2,2,3,3-tetrafluorosuccinate or 4-bromo-3,3,4,4 tetrafluorobut-1-ene and evaluated in terms of their photophysical and LC behaviors. All four molecules exhibited PL in both dilute solutions and crystalline states. In dilute solutions, the PL wavelength varied in the range of 416–463 nm, which reflected the effect of substituent electron-donating/withdrawing nature on molecular orbital energy. *Φ*PL was maximal (0.94) for **2a** and minimal (0.24) for **2c**, which had the shortest *λ*PL. The low *Φ*PL observed in the latter case was ascribed to self-absorption caused by the overlap of absorption and PL spectra. In the crystalline state, the PL behaviors of **2b**–**d** were similar to those in dilute solution, whereas **2a**, which had two alkyl groups at both ends, exhibited green PL with substantially red-shifted *λ*PL. Regarding phase transition behavior, a mesophase was observed for **2a** and **2d** with an ethyl group at one molecular end. Only the N phase with an orientational order appeared in the case of **2a**, whereas the Sm phase with both orientational and positional orders, as well as the N phase, appeared in the case of **2d**. The N phase observed for **2d** exhibited weak blue PL during cooling. The PL intensity increased upon the N→SmA phase transition during cooling, did not substantially change upon the SmA→SmC phase transition, and strongly increased upon the SmC→Cry phase transition on further cooling. Concomitantly, the PL color changed from dark blue to light blue, i.e., temperature-responsive PL behavior was observed. The results described herein expand the applicability of CF2CF2-containing tricyclic molecules as next-generation PL, LC, and PL-LC materials.

**Supplementary Materials:** The following supporting information can be downloaded from https://www.mdpi.com/article/10.3390/cryst13081208/s1: Scheme S1. Synthetic procedure of **2a** and **2d** starting from commercially available dimethyl 2,2,3,3-tetrafluorosuccinate. Scheme S2. Synthetic procedure of **2b** and **2c** starting from commercially available 4-bromo-3,3,4,4-tetrafluorobut-1-ene. Figures S1–S42. 1H-, 13C-, and 19F-NMR spectra; Figure S43. ORTEP-type crystal structure of **2b**; Figures S44–S47. HOMO-1/HOMO-2, HOMO, and LUMO distributions and differential density between HOMO and LUMO; Figure S48. UV/vis absorption and PL spectra of **2a**–**d** in CHCl3 solution; Figure S49. PL spectra of **2a**–**c** in different solvents and related Lippert–Mataga plots; Figure S50. PL decay profiles of **2a**–**d** in CHCl3 solution; Figure S51. Excitation and PL spectra of **2a**–**d** in crystalline states; Figure S52. PL decay profiles of **2a**–**d** in crystalline states; Figures S53–S56. DSC thermograms and POM images for **2a**–**d**; Figure S57. VT-PXRD patterns of **2a** and **2d** recorded at different temperatures; Table S1. Crystallographic data for **2b**; Tables S2–S5. Cartesian coordinates for **2a**–**d**; Tables S6–S9. Phase transition behaviors of **2a**–**d** observed by DSC.

**Author Contributions:** Conceptualization, H.O., S.Y., and T.K.; methodology, H.O., S.Y., and T.K.; validation, H.O., S.Y., and T.K.; formal analysis, H.O., S.Y., and T.K.; investigation, H.O., S.Y., and T.K.; resources, S.Y. and T.K.; data curation, H.O., S.Y., and T.K.; writing—original draft preparation, H.O., S.Y., and T.K.; writing—review and editing, H.O., S.Y., M.Y., and T.K.; Visualization, H.O., S.Y., and T.K.; supervision, T.K.; project administration, T.K.; funding acquisition, S.Y. and T.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** Data supporting the presented findings are contained within the article and Supplementary Materials.

**Acknowledgments:** The authors express their sincere gratitude to Tosoh Finechem Corporation for providing 4-bromo-3,3,4,4-tetrafluorobut-1-ene and to Profs. Sakurai and Shimizu (Kyoto Institute of Technology) for help with VT-PXRD measurements. The authors acknowledge the use of equipment shared in the MEXT project to promote public utilization of advanced research infrastructure (program for supporting the introduction of the new sharing system), grant number JPMXS0421800222.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

## *Article* **Side-Chain Labeling Strategy for Forming Self-Sorted Columnar Liquid Crystals from Binary Discotic Systems**

**Tsuneaki Sakurai 1,\*, Kenichi Kato <sup>2</sup> and Masaki Shimizu <sup>1</sup>**


**Abstract:** The spontaneous formation of self-sorted columnar structures of electron-donating and accepting π-conjugated molecules is attractive for photoconducting and photovoltaic properties. However, the simple mixing of donor–acceptor discotic molecules usually results in the formation of mixed-stacked or alternating-stacked columns. As a new strategy for overcoming this problem, here, we report the "side-chain labeling" approach using binary discotic systems and realize the preferential formation of such self-sorted columnar structures in a thermodynamically stable phase. The demonstrated key strategy involves the use of hydrophobic and hydrophilic side chains. The prepared blend is composed of liquid crystalline phthalocyanine with branched alkyl chains (**H2Pc**) and perylenediimide (PDI) carrying alkyl chains at one side and triethyleneglycol (TEG) chains at the other side (**PDIC12/TEG**). To avoid the thermodynamically unfavorable contact among hydrophobic and hydrophilic chains, **PDIC12/TEG** self-assembles to stack up on top of each other and **H2Pc** as well, forming a homo-stacked pair of columns (self-sort). Importantly, **H2Pc** and **PDIC12/TEG** in the blend are macroscopically miscible and uniform, and mesoscopically segregated. The columnar liquid crystalline microdomains of **H2Pc** and **PDIC12/TEG** are homeotropically aligned in a glass sandwiched cell. The "labeling" strategy demonstrated here is potentially applicable to any binary discotic system and enables the preferential formation of self-sorted columnar structures.

**Keywords:** self-sort; segregated columns; binary mixture; amphiphilicity; homeotropic alignment

### **1. Introduction**

The control of nanostructures and miscibility of binary blends is important for tuning the physical properties of the blended organic materials. Historically, polymer blend has been a famous notion, where the miscibility and compatibility of blended polymers have been well discussed, especially in view of their effect on thermal and mechanical properties [1–4]. The bulk heterojunction of conjugated polymers and fullerene derivatives is another famous concept utilized for blend films in organic photovoltaic cells [5–7]. More recently, blends of electron donors and acceptors based on conjugated polymers or small molecules have been used for active layers in organic electronic devices, including photovoltaic cells [8–10], electrochemical transistors [11], ambipolar transistors [12], and so on. In these blends, not only large interfaces of donor and acceptor molecules (or macromolecules) but also hole/electron-transporting bicontinuous interpenetrating networks are essential for the device operation. The optimization of such nanostructures in the blends is usually performed by a try-and-error approach using spin-coating methods. Meanwhile, hydrogen-bond-assisted organogelator systems have been demonstrated as a more elaborated molecular design [13–16]. In these systems, self-sorted fibrous one-dimensional assemblies were developed by the simple mixing of electron donor and acceptor molecules. The different distances of two hydrogen bonding sites between the donor and acceptor molecules are critical for the formation of self-sorting fibers. Although methodologies of

**Citation:** Sakurai, T.; Kato, K.; Shimizu, M. Side-Chain Labeling Strategy for Forming Self-Sorted Columnar Liquid Crystals from Binary Discotic Systems. *Crystals* **2023**, *13*, 1473. https://doi.org/ 10.3390/cryst13101473

Academic Editor: Benoit Heinrich

Received: 27 August 2023 Revised: 6 October 2023 Accepted: 7 October 2023 Published: 10 October 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

bulk heterojunctions as well as binary organogelator systems have been established, they have a critical drawback: the obtained nanostructures are kinetically controlled but not thermodynamically stable in the long term. Thus, the construction of thermodynamic bicontinuous structures of electron donor and acceptor materials has been awaited.

Liquid crystal (LC) phases are usually thermodynamically stable and appropriate as a platform for constructing the arrays of π-conjugated systems through self-assembly, resulting in the functional soft materials [17–23]. However, there has been no example of binary LC blends with bicontinuous structures. In columnar LC phases of discotic π-systems, columnarly stacked π-conjugated molecules enable one-dimensional charge transport pathways. If electron-donating and accepting π-conjugated molecules form a selfsorted columnar structure, we can realize intracolumnar hole/electron transport pathways as well as intercolumnar *p*/*n* heterojunctions with large interfaces, which equips a longterm structural stability. However, in relevant previous studies, mixed-stacking columnar structures were reported from LC phthalocyanine (*p*-type: electron donor) and perylene diimide (PDI) (*n*-type: electron acceptor) molecules [24–26]. This is quite reasonable because they are entropically favored (Figure 1a). Special molecular designs are required to accomplish thermodynamically stable self-sorted columnar structures by the self-assembly of LC electron donor and acceptor molecules. Here, we report a "side-chain labeling" strategy to realize self-sorted columnar structures from LC mixtures composed of freebase phthalocyanine (**H2Pc**) and PDI derivatives **PDIC12/C12**, **PDIC12/TEG**, and **PDITEG/TEG** (Figure 2). The dissymmetric introduction of both hydrophobic and hydrophilic side chains in **PDIC12/TEG** gives large enthalpic gain to the homo-stacked PDI columns, and thus the self-sorted structure is stabilized when the **PDIC12/TEG** is mixed with **H2Pc** carrying hydrophobic chains (Figure 1b). Furthermore, the resulting self-sorted LC columns of the **PDIC12/TEG** and **H2Pc** molecules align homeotropically in a sandwiched glass cell, which is desirable for efficient charge transport in photovoltaic applications.

**Figure 1.** Schematic illustrations of side-chain-directed molecular assembly of electron donors (orange) and acceptor (red) π-systems (**a**) substituted with hydrophobic side chains (green) alone and (**b**) site-specifically substituted with hydrophobic (green) and hydrophilic (blue) side chains.

Entropically favored mixed-stacking structures can be formed because both electron donor and acceptor molecules are decorated with alkyl chains and they are molecularly miscible. We focused on strong enthalpic interactions of immiscible hydrophobic and hydrophilic chains. If donor (acceptor) molecules are substituted with alkyl chains and acceptor (donor) molecules with hydrophilic oxyethylene chains, they are not miscible but macroscopically segregated [27]. How do we access the thermodynamic self-sorted nanostructure? The clue is hidden in the amphiphilic molecular design used in previous works, including ours [28–32]. **PDIC12/TEG** (Figure 2) is a Janus-type amphiphilic compound forming a columnar LC phase at room temperature. **PDIC12/TEG** molecules pack into the rectangular columnar phase with *p2mg* symmetry to minimize the unfavorable contact among immiscible hydrophobic and hydrophilic side chains incorporated in a single PDI core. When **H2Pc**, a compound carrying hydrophobic chains, is blended with **PDIC12/TEG**, **H2Pc** molecules may not intercalate into a column of **PDIC12/TEG** to avoid the

enthalpic penalty of increasing contacts between hydrophobic and hydrophilic segments. Nevertheless, they are macroscopically miscible with one another due to their hydrophobic chains. We expected that **H2Pc** molecules would form a homo-stacked columnar assembly and laterally contact with the hydrophobic chains of **PDIC12/TEG**, resulting in the self-sorted columnar assembly (Figure 1b). The preferential formation of a self-sorted nanostructure will be discussed in detail in the section of Results and Discussion.

**Figure 2.** Chemical structures of liquid crystalline phthalocyanine **H2Pc** and perylenediimides **PDIC12/C12**, **PDIC12/TEG**, and **PDITEG/TEG**.

### **2. Materials and Methods**

*2.1. Synthesis and Characterization of H2Pc and PDIs*

**H2Pc**, **PDIC12/C12**, **PDIC12/TEG**, and **PDITEG/TEG** were synthesized according to the previous reports [31,33], and characterized by 1H NMR spectroscopy in CDCl3 on a Varian model Mercury 400 spectrometer, operating at 400 MHz, where chemical shifts were determined with respect to tetramethylsilane as an internal reference. MALDI-TOF mass spectrometry was performed on an Autoflex III spectrometer from Bruker, Japan, using dithranol as a matrix. In addition, 1:1 molar ratio mixtures of **H2Pc** and PDI derivatives were prepared from their CH2Cl2 solutions in a glass vial. The solvent was evaporated in each solution, allowing for as-prepared waxy LC mixture.

### *2.2. Characterization of LC Mesophases*

The optical textures were recorded by a BX53-P polarizing optical microscope (POM) from Olympus, Japan, equipped with an EOS kiss X7i digital camera from Canon, Japan. The sample was loaded, by use of a capillary action, into an LC cell without any surface treatment. The LC cell was prepared as follows. The glasses with a size of 16 × 22 × 0.5 mm were purchased from Matsunami Glass Ind., Ltd. (Haemacytomer Cover Glasses), Osaka, Japan. Silica beads with 5 μm diameter were dispersed in a drop of fast curing optical adhesive (NOA81) purchased from THORLABS, and the beads-dispersed adhesive was spotted at four places in a rectangle on one glass. Another glass was placed onto the adhesivespotted glass with a few millimeter offset along with the long axis. The sandwiched glass cell was irradiated with 365 nm light from a SLUV-4 handy UV lamp purchased from AS ONE, Japan, to complete the curing of the adhesive.

The temperature of the sample was controlled by a HS82 hot-stage from Mettler Toledo, Japan. Differential scanning calorimetry (DSC) measurements were performed on a DSC 822e differential scanning calorimeter from Mettler Toledo, Japan. Cooling and heating profiles were recorded and analyzed with the STARe system. Samples were put into an aluminum pan and allowed to be measured under N2 gas flow.

X-ray diffraction measurements were carried out using a synchrotron radiation X-ray beam with a wavelength of 0.108 nm on BL44B2 at the Super Photon Ring (SPring-8, Hyogo, Japan) [34]. A large Debye–Scherrer camera was used in conjunction with an imaging plate as a detector, and all diffraction patterns were recorded with a 0.01◦ step in 2*θ*. The samples were loaded by capillary action at the isotropic liquid melts into a 0.5 mm thick soda glass

capillary purchased from WJM-Glas/Muller GmbH. During the measurements, samples were continuously rotated along the capillary axis to obtain a homogeneous diffraction pattern. The exposure time to the X-ray beam was 1.5 min each.

### *2.3. Evaluation of Intracolumnar Molecular Order*

Electronic absorption spectra were recorded on a V-730 UV/VIS/NIR spectrophotometer from JASCO, Japan, where the scan rate, response, and band width were set at 1000 nm min<sup>−</sup>1, 0.06 s, and 1.0 nm. The CHCl3 solution samples were prepared at 2.0 × <sup>10</sup>−<sup>5</sup> <sup>M</sup> and measured in a quartz cell equipped with a screw cap. The optical path length of the cell is 1.0 cm. Spin-coated films were prepared from CHCl3 solutions of the single compound or 1:1 molar ratio **H2Pc**/PDI mixtures onto a quartz substrate with a size of 9 × 40 × 1 mm. The spin-coating was performed at 1500 rpm for 30 s using a Mikasa model MS B-100 spin coater.

### **3. Results and Discussion**

### *3.1. Homeotropic Alignment Capability of H2Pc and PDIs*

The phase transition behaviors of **H2Pc**, **PDIC12/C12**, **PDIC12/TEG**, and **PDITEG/TEG** were characterized by DSC (Figure S1). They all showed LC mesophases and their clearing points are 180, 223, 189, and 165 ◦C on cooling, respectively, which is almost identical with the previous reports [31,33]. A spontaneous homeotropic alignment of discotic columnar LCs was often reported for hexagonal columnar mesophases [35–38]. The homeotropic alignment capability of **H2Pc** discotic columns was already reported in a previous study [39]. The capability of spontaneous homeotropic alignment for the PDI derivatives was monitored by means of POM using samples.

After being loaded into the glass cell with a capillary action at the isotropic liquid phase (Iso), the sample was slowly cooled at 1.0 K/min. Then, the growth of dendritic textures was observed without a polarizer for all the PDI derivatives at around their clearing points (Figure 3a–c). At the same time, no optical texture appeared under crossed polarizers (Figure 3a–c). A similar behavior was seen for **H2Pc** with slow cooling at 1.0 K/min (Figure 3d), while defect areas with homogeneous alignment were confirmed upon rapid cooling at 10 K/min (Figure S2). These microscopic observations indicate the strong homeotropic tendency for the hexagonally arranged discotic columns from all four compounds (Figure 3e). Interestingly, after the phase transition from a hexagonal to rectangular columnar mesophase at around 110 ◦C upon cooling, the micrograph of **PDIC12/TEG** was almost unchanged, suggesting that the homeotropic orientation was kept upon the hexagonal–rectangular structural transformation.

**Figure 3.** Crossed polarized (**left**) and optical (**right**) microscopy images of (**a**) **PDIC12/C12**, (**b**) **PDIC12/TEG**, (**c**) **PDITEG/TEG**, and (**d**) **H2Pc** in glass sandwich cell without any treatment. (**a**–**d**) were taken at 222, 193, 168, and 181 ◦C, respectively, after cooling from their isotropic liquid phases at 1.0 K/min. Scale bars represent 200 μm. (**e**) Schematic illustration of LC samples in 5 μm thick sandwiched glass cell and homeotropic alignment of discotic columns formed in LC.

### *3.2. Orientation, Phase Transition Behavior, and Phase Structure of H2Pc/PDI Mixtures*

In order to confirm our hypothesis of the side-chain labeling strategy, 1:1 molar ratio mixtures of **H2Pc**/**PDIC12/C12**, **H2Pc**/**PDIC12/TEG**, and **H2Pc**/**PDITEG/TEG** were prepared and their phase behaviors were characterized. The three blend samples were loaded into a sandwich glass cell over 210 ◦C, and their optical textures were recorded upon cooling. Figure 4 shows optical micrographs with and without crossed polarizers and the dependence of the optical textures on the cooling rate. The mixture of **H2Pc**/**PDITEG/TEG** gave the most distinctive picture (Figure 4c,f). Independent of the cooling rate, the mixture clearly gave green and red color areas, which most likely correspond to the domains of **H2Pc** and **PDITEG/TEG**, respectively. The hydrophobic **H2Pc** and hydrophilic **PDITEG/TEG** are immiscible with each other and segregated macroscopically [27]. In contrast, **H2Pc**/**PDIC12/C12** and **H2Pc**/**PDIC12/TEG** appear to have a homogeneous phase in the field of microscope view. Upon rapid cooling from their isotropic phases, fan-shaped textures appeared in POM upon Iso-to-LC phase transitions for both **H2Pc**/**PDIC12/C12** and **H2Pc**/**PDIC12/TEG** (Figure 4a,b). The presence of textures indicates a non-homeotropic alignment of columnar structures. In contrast, the cooling rate was set at 1.0 K/min, and the growth of dendritic textures was seen in optical microscopy without polarizers, but almost dark field images were obtained under crossed polarizers (Figure 4d,e). Although the dark area ratio in these blends was a bit smaller than their constituent compounds, the homeotropic alignment capability was confirmed by POM observations.

**Figure 4.** Crossed polarized (**left**) and optical (**right**) microscopy images of 1:1 molar ratio mixtures of (**a**,**d**) **H2Pc**/**PDIC12/C12**, (**b**,**e**) **H2Pc**/**PDIC12/TEG**, and (**c**,**f**) **H2Pc**/**PDITEG/TEG** in glass sandwich cell without any treatment. Images (**a**–**c**) were taken at 25 ◦C after rapid cooling from their isotropic melt. (**d**–**f**) were taken at 198, 200, and 161 ◦C, respectively, after cooling from their isotropic liquid phases at 1.0 K/min. Scale bars represent 200 μm.

The phase transition behaviors of the 1:1 molar ratio mixture of **H2Pc**/**PDIC12/C12**, **H2Pc**/**PDIC12/TEG**, and **H2Pc**/**PDITEG/TEG** were characterized by DSC. The DSC traces of **H2Pc**/**PDIC12/C12** and **H2Pc**/**PDIC12/TEG** implied phase transitions from a mesoscopically uniform material (Figure 5). In the blend of **H2Pc**/**PDIC12/C12**, the clearing point (202 ◦C on cooling) is between **H2Pc** (180 ◦C) and **PDIC12/C12** (223 ◦C) (Figures 5a and S1), which is reasonable for molecularly miscible binary mixtures. In contrast, the clearing point of **H2Pc**/**PDIC12/TEG** (205 ◦C on heating) is higher than those of **H2Pc** (181 ◦C) and **PDIC12/TEG** (191 ◦C) (Figures 5b and S1). This pattern is quite rare and interesting to note—the LC phase of the blend is thermodynamically more stable than the parent columnar phases. We will discuss this phenomenon in more depth with the powder X-ray diffraction (PXRD) patterns (vide infra). In the blend of **H2Pc**/**PDITEG/TEG**, the melting and clearing points of both the compounds are detected, though the clearing point at 192 ◦C is higher than that of **H2Pc** (181 ◦C) (Figures 5 and S1). In other words, **H2Pc**/**PDITEG/TEG** affords the superimposed DSC chart of those of the constituent compounds. This is solely a sign of

the macroscopic phase separation of **H2Pc** and **PDITEG/TEG**, which is consistent with the POM images.

**Figure 5.** DSC traces of 1:1 molar ratio mixtures of (**a**) **H2Pc**/**PDIC12/C12**, (**b**) **H2Pc**/**PDIC12/TEG**, and (**c**) **H2Pc**/**PDITEG/TEG** on 2nd heating/cooling cycle at 10 K/min.

Although the clearing points for **H2Pc**/**PDIC12/C12** and **H2Pc**/**PDIC12/TEG** are almost identical, the values of phase transition enthalpy inform that the LC phase structure and degree of miscibility are completely different between these mixtures. The LC-to-Iso phase transition enthalpy changes (Δ*H*) were evaluated from the second heating trace in DSC (Figure S1) and are 4.8, 17.3, and 8.1 kJ mol−<sup>1</sup> for **H2Pc**, **PDIC12/C12**, and **PDIC12/TEG**, respectively. The entropy changes upon these phase transitions (Δ*S*) can be estimated from the principle that Gibbs free energy is constant upon phase transition, i.e., Δ*G* = Δ*H* – *T*Δ*S* = 0, where Δ*G* and *T* are Gibbs free energy change and absolute temperature. By substituting the evaluated Δ*H* and observed *T* into the above equation, the values of Δ*S* were estimated as 10.6, 34.8, and 17.4 J mol−<sup>1</sup> K−<sup>1</sup> for **H2Pc**, **PDIC12/C12**, and **PDIC12/TEG**, respectively. These values well explain the relatively larger entropic gain of linear dodecyloxy chains upon phase transition from columnar mesophase to isotropic liquid. The values of Δ*H* and Δ*S* are calculated for the 1:1 molar mixtures of **H2Pc**/**PDIC12/C12** and **H2Pc**/**PDIC12/TEG** in a similar way, except that the average molecular weight of the two components is used for transforming the observed heat change into enthalpy values. The values of Δ*S* were estimated as 18.9 and 15.9 J mol−<sup>1</sup> K−<sup>1</sup> for **H2Pc**/**PDIC12/C12** and **H2Pc/PDIC12/TEG**, respectively. The value of 18.9 J mol−<sup>1</sup> K−<sup>1</sup> for **H2Pc**/**PDIC12/C12** is smaller than the averaged Δ*S* values calculated from those of the parent compound (22.7 J mol−<sup>1</sup> K<sup>−</sup>1), implying that the molecules are disordered in the observed columnar mesophase. For example, one column is composed of **H2Pc** and **PDIC12/C12** molecules. In contrast, the value of 15.9 J mol−<sup>1</sup> K−<sup>1</sup> for **H2Pc/PDIC12/TEG** is a bit larger than and even close to the averaged Δ*S* values of the parent compound (14.0 J mol−<sup>1</sup> K−1). This similarity in the entropy values indicates the possibility that **H2Pc** and **PDIC12/TEG** form their respective microdomains. The molecular motion in the microdomains upon the phase transition would be consistent with that in the bulk of the corresponding compounds, while that at the interfaces of the microdomains is relatively limited. In this case, the Δ*S* value is expected to be smaller than the average values speculated from those for the parent compounds. Considering that the phase transition temperatures for these blends and parent compounds are close and in the range of 181–224 ◦C, the above speculations would have a certain level of significance. As below, we will directly discuss the molecular packing structures in the mesophase for LC blends based on the PXRD measurements.

The molecular packing structures in the mesophases were studied by means of PXRD measurements. In the mesophase at 80 ◦C, the 1:1 molar ratio mixture of **H2Pc**/**PDIC12/C12** gave a diffraction pattern that is assignable to a hexagonal columnar phase with the lattice parameter of *a* = 32.1 Å (Figure 6a). Variable-temperature PXRD measurements elucidated that the hexagonal columnar mesophase was present at 30–200 ◦C (Figure S4). The parent hexagonal columnar mesophases of **H2Pc** and **PDIC12/C12** have a lattice parameter of *a* = ~32 Å and ~31 Å, respectively (Figures S3 and S7). The size matching of these two

molecules may be one of the critical reasons for stabilizing a uniform hexagonal packing of mixed-stacked columns. With the clearing temperature information discussed in the DSC section, we conclude that the **H2Pc**/**PDIC12/C12** self-organized into molecularly miscible, entropically favored columns with hexagonal packing, as illustrated in Figure 7a. Interestingly, the mixture of **H2Pc**/**PDIC12/TEG** showed different behavior. Over 80 ◦C, the mixture formed a hexagonal columnar phase with *a* = ~32 Å (Figure S5). When being cooled down to 80 ◦C, the mixture changed its PXRD pattern to the superposition of those of **H2Pc** and **PDIC12/TEG** (Figure 6b), and similar superimposed patterns were also recorded at 50 and 30 ◦C (Figure S5). Namely, **H2Pc** and **PDIC12/TEG** are mesoscopically segregated but macroscopically miscible, as disclosed by PXRD and DSC measurements. The schematic illustration of **H2Pc**/**PDIC12/TEG** is shown in Figure 7b. Then, we tried to interpret the hexagonal columnar mesophase of **H2Pc**/**PDIC12/TEG** over 80 ◦C. Although a set of observed diffractions was assigned to a single hexagonal lattice, the (001) peak at *d* = ~3.4 Å, corresponding to the π-distance periodicity of **PDIC12/TEG**, obviously appeared as similar to those at 30–80 ◦C. In addition, as mentioned in the DSC analysis earlier, the clearing temperature of the mixture at 205 ◦C is higher than those of the parent compounds. Having these results in mind, we consider that **H2Pc** and **PDIC12/TEG** mainly form selfsorted columns even over 80 ◦C but the average domain size may be decreased. The blend **H2Pc**/**PDITEG/TEG** exhibited superimposed pattens of those of **H2Pc** and **PDITEG/TEG** below 240 ◦C (Figures 6 and S6). These results are consistent with the macroscopic phase separation derived from the POM and DSC results. The illustration of macroscopically phase-separated columnar phases is shown in Figure 7c.

**Figure 6.** XRD patterns of 1:1 molar ratio mixtures of (**a**) **H2Pc**/**PDIC12/C12** at 80 ◦C, (**b**) **H2Pc**/ **PDIC12/TEG** at 80 ◦C, and (**c**) **H2Pc**/**PDITEG/TEG** at 160 ◦C. For comparison, the XRD patterns of the components for the blends are represented in (**b**,**c**).

**Figure 7.** Schematic illustrations of proposed molecular assembly in columnar LC phases for (**a**) **H2Pc/ PDIC12/C12**, (**b**) **H2Pc/PDIC12/TEG**, and (**c**) **H2Pc/PDITEG/TEG**. Red and green disks represent corresponding H2Pc and PDI molecules.

### *3.3. Intracolumnar Molecular Order in H2Pc/PDI Mixtures*

The intracolumnar molecular order in the mesophases at room temperature was investigated by absorption spectroscopy of the thin film of the 1:1 molecular blends. In diluted CHCl3 solutions, both **H2Pc** and **PDIC12/C12** are molecularly dispersed and show characteristic absorption at 600–750 nm and 400–550 nm, respectively, with strong vibronic coupling features (Figure 8a). In spin-coated LC films, these absorption bands become broad and blue-shifted due to the columnar assembly of molecules with π–π interactions (H-like aggregation). The spectra of **PDIC12/TEG** and **PDITEG/TEG** in the films are essentially the same as that of **PDIC12/C12**. Then, the spectra of the blend films were analyzed similarly. As expected, in the macroscopically phase-separated **H2Pc**/**PDITEG/TEG** blend film, the shape of the absorption spectra is almost the superposition of those of the parent LC films (Figures 8b and S8). The heterotropic interactions hardly work due to the limited area of the interfaces between **H2Pc** and **PDITEG/TEG**. In the LC phase of **H2Pc**/**PDIC12/C12**, proposed as a molecularly miscible columnar phase, the absorption spectrum of the film is completely different from that of **H2Pc**/**PDITEG/TEG**. The characteristic two intense absorption bands from **H2Pc** and **PDIC12/C12** both show vibronic structures in the blend film, while these bands are broadened compared to their solution states (Figure 8b). This feature strongly indicates that homotropic molecular interactions in their columnar assembly are broken, supporting the proposed molecularly miscible columnar phase (Figure 7a). The film of the **H2Pc**/**PDIC12/TEG** mixture afforded basically the superimposed spectrum of those of **H2Pc** and **PDIC12/TEG**. However, shoulder vibronic peaks at around 670–730 nm suggest that a small part of **H2Pc** columnar assemblies is dissociated by the intercalation of **PDIC12/TEG**. Thus, the picture of mesoscopically segregated self-sorted assembly as illustrated in Figure 7b may almost be correct, but the structural purity is less than perfect.

**Figure 8.** (**a**) Absorption spectra of **H2Pc** (green) and **PDIC12/C12** (red) in spin-coated film (solid line) and in CHCl3 (dotted line). (**b**) Absorption spectra of spin-coated film of **H2Pc/PDIC12/C12** (red), **H2Pc/PDIC12/TEG** (green), and **H2Pc/PDITEG/TEG** (blue).

### **4. Conclusions**

Although nanosegregated, bicontinuous structures of electron-donating and accepting π-conjugated molecules have been recognized as important for photoconducting and photovoltaic properties, only the kinetic control of such nanostructures has been reported so far. We conceived the side-chain labeling strategy using hydrophobic/hydrophilic chains to induce the homotropic self-assembly of donor and acceptor molecules and demonstrated the preferential formation of donor/acceptor self-sorted columnar structures in thermodynamically stable LC binary mixtures. In this LC blend, the columnar mesophases of H2Pc and PDI molecules are macroscopically miscible and uniform but mesoscopically segregated as evidenced by DSC and PXRD results. In addition, the intercalation of PDI (H2Pc) to the H2Pc (PDI) columns is minimally inhibited as supported by absorption spectroscopy. In a more comprehensive view, self-sorted nanostructures of binary mixtures are entropically unfavored in general, but the present work clarified that they can be accessed thermodynamically by self-assembly processes with the help of enthalpic interactions of

immiscible side-chain pairs. Amphiphilic molecules—**PDIC12/TEG** in this work—induce mesoscopic phase separation and avoid macroscopic phase separation. This role is referred to as a compatibilizer in the research field of macromolecules [40]. While a small molecular compatibilizer has recently been reported [41], our work further extends the concept to the strategy of accessing self-sorted nanostructures. In future, the important subjects include the analysis and control of the size of donor and acceptor nano(micro)-domains, which will lead to the manipulation of photo and electronic functions originating from nanosegregated donor/acceptor blends.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cryst13101473/s1, Figure S1: DSC traces of **H2Pc**, **PDIC12/C12**, **PDIC12/TEG**, and **PDITEG/TEG**; Figure S2: Crossed polarized microscopy images of **H2Pc** in glass sandwich cell; Figure S3: Variable-temperature XRD patterns of **H2Pc**; Figure S4: Variable-temperature XRD patterns of 1:1 molar ratio mixture of **H2Pc/PDIC12/C12**; Figure S5: Variable-temperature XRD patterns of 1:1 molar ratio mixture of **H2Pc/PDIC12/TEG**; Figure S6: Variable-temperature XRD patterns of 1:1 molar ratio mixture of **H2Pc/PDITEG/TEG**; Figure S7: Schematic illustrations of columnar hexagonal and rectangular phases with corresponding lattice parameters and primary diffractions.

**Author Contributions:** T.S. conceived and designed the experiments; T.S. performed the experiments; T.S. and K.K. analyzed the data; T.S. and M.S. wrote the manuscript draft; K.K. revised the manuscript draft. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by JSPS KAKENHI numbers 20H02710 from the Japan Society for the Promotion of Science and a research grant from TEPCO Memorial Foundation.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The synchrotron radiation XRD experiments were performed at BL44B2 in SPring-8 with the approval of RIKEN. T.S. thanks the Leading Initiative for Excellent Young Researchers program by Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

MDPI St. Alban-Anlage 66 4052 Basel Switzerland www.mdpi.com

*Crystals* Editorial Office E-mail: crystals@mdpi.com www.mdpi.com/journal/crystals

Disclaimer/Publisher's Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Academic Open Access Publishing

mdpi.com ISBN 978-3-0365-9549-8