Crystal Engineering of 1D Exciton Systems Composed of Single‐ and Double‐Stranded Perylene Bisimide J‐Aggregates

Single crystals of three at bay area tetraphenoxy‐substituted perylene bisimide dyes are grown by vacuum sublimation. X‐ray analysis reveals the self‐assembly of these highly twisted perylene bisimides (PBIs) in the solid state via imide–imide hydrogen bonding into hydrogen‐bonded PBI chains. The crystallographic insights disclose that the conformation and sterical congestion imparted by the phenoxy substituents can be controlled by ortho‐substituents. Accordingly, whilst sterically less demanding methyl and isopropyl substituents afford double‐stranded PBI chains of complementary P and M atropo‐enantiomers, single hydrogen‐bonded chains of homochiral PBIs are observed for the sterically more demanding ortho‐phenyl substituents. Investigation of the absorption and fluorescence properties of microcrystals and thin films of these PBIs allow for an unambiguous interpretation of these exciton systems. Thus, the J‐aggregates of the double‐stranded crystals exhibit a much larger (negative) exciton coupling than the single‐stranded one, which in contrast has the higher solid‐state fluorescence quantum yield.


DOI: 10.1002/adom.202000926
control of coloristic properties to afford high grade color pigments with hues from red to maroon by tuning the lateral and rotational displacement of the cofacially stacked chromophores. [10] Based on the knowledge acquired by this research, more recently PBI packing arrangements could be tailored to provide outstanding n-channel organic semiconductor materials for organic transistors and solar cells. [11][12][13] However, with their outstanding optical properties including fluorescence quantum yields up to unity, [14] PBIs are interesting candidates for a broad variety of photonic applications as well. [15][16][17] In this regard it is important that most recent research recognized for a variety of covalent and supramolecular PBI dimers as well as larger self-assembled PBI dye aggregates that, depending on the mutual arrangement of the PBIs, entirely different photophysical properties can be tailored. [18][19][20] These insights from small model systems now motivate the engineering of unprecedented PBI packing arrangements in solid-state materials with the goal to tailor functional properties for applications of these dyes beyond their conventional ones as color pigments. [21][22][23][24][25] Here of particular interest is the longitudinal slip-stack arrangement of the dyes that affords dye aggregates with significant bathochromic shifts of the absorption and emission bands and high exciton mobilities originating from J-type exciton coupling. [26] Accordingly, the aim of this work was to engineer a slip-stacked packing arrangement of PBI dyes to afford a new class of PBI pigments with solid-state properties that are distinguished from the common cofacially stacked color pigment materials.
In our previous work, we applied the concept of hydrogen bond directed self-assembly to the class of at bay area fourfold phenoxy-substituted PBIs (Figure 1). By this, well-defined J-aggregates could be achieved with stacking motifs from double-up to sextuple-stranded helices in solution, [27,28] in gels, [29,30] and in the liquid-crystalline state. [31][32][33] In-depth photophysical studies for these J-aggregates revealed highly desired properties including outstanding fluorescence quantum yields up to 90%, strong exciton coupling, and exciton migration for distances up to 100 nm. [34,35] However, because all of these dye aggregates had to be equipped with long aliphatic or oligoethylene glycol chains for solubilization purpose, they are rather model systems than useful materials for desirable applications.
Single crystals of three at bay area tetraphenoxy-substituted perylene bisimide dyes are grown by vacuum sublimation. X-ray analysis reveals the self-assembly of these highly twisted perylene bisimides (PBIs) in the solid state via imide-imide hydrogen bonding into hydrogen-bonded PBI chains. The crystallographic insights disclose that the conformation and sterical congestion imparted by the phenoxy substituents can be controlled by ortho-substituents. Accordingly, whilst sterically less demanding methyl and isopropyl substituents afford double-stranded PBI chains of complementary P and M atropo-enantiomers, single hydrogen-bonded chains of homochiral PBIs are observed for the sterically more demanding ortho-phenyl substituents. Investigation of the absorption and fluorescence properties of microcrystals and thin films of these PBIs allow for an unambiguous interpretation of these exciton systems. Thus, the J-aggregates of the double-stranded crystals exhibit a much larger (negative) exciton coupling than the single-stranded one, which in contrast has the higher solid-state fluorescence quantum yield.

Introduction
Crystal engineering of π-scaffolds in general [1][2][3][4][5][6] and of perylene bisimides (PBIs) in particular [7][8][9] has a long tradition. For several decades PBI functional material research was focused on the Furthermore, it is noteworthy that the assignment of the number of self-assembled strands in these aggregates remained only tentative because neither for the aggregates in solution nor for those in the liquid-crystalline state an unambiguous structure determination by single crystal X-ray diffraction was possible. In addition, we were wondering how the helical packing structures considered for these self-assembled materials, i.e., each hydrogen-bonded aggregate fiber consisted of exclusively either P or M PBI atropo-enantiomers to give P and M helices (which are present in equal amounts in the columnar LC phases) could lead to crystalline unit cells.

Results and Discussion
For our current study we synthesized three at bay area tetraphenoxy-substituted PBIs 1(Me), [29] 1(Pr), and 1(Ph) (Figure 1) from methylbenzyl-protected 1,6,7,12-tetrachloro PBI with three distinct phenol derivatives and subsequent deprotection of the imide positions with boron tribromide (for details on synthesis and characterization see Supporting Information). By using either a methyl, iso-propyl or phenyl substituent in orthoposition of the phenoxy group, the sterical demand around the π-scaffold of the chromophores is varied. It is noteworthy that the tetraphenoxy-substituted PBIs exhibit high conformational flexibility which improves the solubility of the otherwise insoluble pigment. [36] However, this feature renders the crystallization of these dyes challenging. Accordingly, different from earlier (unsuccessful) attempts [29,32] here crystallization was carried out on silicon wafer pieces in vacuum by thermal gradient sublimation at 305-330 °C and a pressure of <2 × 10 −6 mbar, thereby exploiting the rapid interconversion of the various conformational states at elevated temperatures ( Figure S1, Supporting Information). As another advantage of this method, the otherwise common inclusion of solvent molecules in the crystal lattice could be avoided and high-quality solvent-free single crystals suitable for single crystal X-ray analysis as well as spectroscopic investigations could be obtained (Figure 2).
The molecular structures of 1(Me), 1(Pr), and 1(Ph) in the single crystals are depicted in Figure 2. What becomes immediately apparent from these molecular structures is a broken symmetry, i.e., in all cases the dihedral angles of two bay areas are unequal (for more details, see Supporting Information) and also the orientation of the ortho-substituents is not the same for the four phenoxy substituents. Due to the sterical congestion of the four phenoxy substituents, the twist angle between the two naphthalene units rises from about 24.5° for 1(Me), over 26.2° for 1(Pr) up to 29.5° for 1(Ph) with increasing size of the orthosubstituents ( Table 1). This trend of increasing the inherent twist of the chromophore's π-scaffold is also measureable by the increasing dihedral angles for the four bay ∠(C1-C12b-C12a-C12) and four inner carbon atoms ∠(C13a-C12a-C12b-C13b) up to 32.8° and 26.9° for 1(Ph), respectively (for numbering see Figure 1a).
The increasing sterical demand of these substituents has another interesting effect on the molecular arrangement in the solid state: While the broken symmetry of 1(Me) and 1(Pr) affords one shielded π-surface and another one accessible for π-π interactions with neighboring molecules, such intermolecular π-π interactions are completely prohibited in case of 1(Ph). Here, the phenyl-residues fully surround the PBI core with one of them even attaching to the PBI surface by π-π interactions (3.62 Å; Figure 2c).
However, nothing influences the solid-state packing as strongly as the propensity of the PBI dyes to interact with each other by imide-imide hydrogen bonding, ranging from 2.83 to 2.94 Å between the oxygen and nitrogen atoms of the CO···HN units (Table 1). Thus, all imide units connect by twofold hydrogen bonding to neighbor molecules, leading to the line-up of the PBI molecules in 1D hydrogen-bonded supramolecular polymer chains (Figure 3). These strands always consist exclusively either of P (blue colored molecules) or M (red colored molecules) enantiomers, respectively, with neighboring strands always containing the opposite stereoisomer. In the case of 1(Me) and 1(Pr) the two neighboring single-strands are attached closely at van der Waals distances of about 3.48-3.51 Å (Table 1) between slip-stacked, longitudinally displaced PBI molecules, thereby affording a textbook example for J-aggregates with slip angle (Θ center-center ) of about 25.4°-29.8° ( Figure 3 and Table 1). It is important to emphasize the complementary π-π stacking motif of each PBI dye to both longitudinally displaced dyes in the next strand guided by the alternating M-and P-helical twist encoded in the PBIs within the double-strands. Interestingly, while the double-strands of 1(Me) adopt a herringbone-like packing in the solid state, the chromophores of double-and single-strands of PBIs 1(Pr) and 1(Ph) are almost coplanar oriented in the crystal lattice ( Figure 3).
The center-to-center distances of the parallel double-strands of PBIs 1(Me) and 1(Pr) are with 12.2 and 10.7 Å, respectively, even more separated than the single-strands of 1(Ph) (9.89 Å), thereby affording negligible impact on the optical properties in the solid state (vide infra). In contrast to PBI J-aggregates observed in solution, the double-stranded arrangements do not show any helicity along the H-bonding axis. This originates from illustrating the accessible and shielded sides of the PBI π-surfaces as well as the torsion angles of the two naphthalene subunits (gray). Phenoxy substituents are displayed in wireframe (left, middle) or space filling (right) while the ellipsoids of the PBI chromophore are set to 50% probability. Hydrogen atoms were partially omitted for clarity (C: gray, O: red, N: blue, H: white). The π-stacked nearest neighboring molecule is depicted in (a) and (b) in light gray. Table 1. Structural parameters found in the solvent-free single crystal structures of 1(Me), 1(Pr), and 1(Ph). a) Angle between the PBI's two naphthalene subunits; b) Numbering of carbon atoms are according to Figure 1; c) Distances determined for the diastereomer with three phenoxy substituents facing in the same direction; d) Distances determined for the diastereomer with two residues facing upward and downward, respectively; e) For the definition of the center-to-center distance and the slip angle, see Figure 3.
the heterochiral assembly of P and M atropo-enantiomers that differs from the homochiral assembly found in solution. [23,27] This information from single crystal X-ray analysis is now most useful to elucidate the impact of the molecular packing of the PBIs dyes in single-and double-stranded arrangements on optical properties in the solid state ( Figure S2, Supporting Information). The absorption and fluorescence properties of molecularly dissolved dyes in dichloromethane solution are similar to those reported for other tetraphenoxy PBI with longer solubilizing side chains. [32] Thus, PBIs 1(Me), 1(Pr), and 1(Ph) display almost identical absorption and fluorescence profiles with perfect mirror image band shapes ( Figure S3, Supporting Information). Compared to tetraphenoxy PBIs without ortho-substituents we note a more defined vibronic progression that is attributed to the more restricted conformational space for the phenoxy groups. [32] The absorption (λ abs ) and emission (λ em ) maxima are at 580-582 and 608-609 nm, respectively, giving rise to small Stokes shifts ( Stokes  ν ∆ ) of only 760-820 cm −1 ( Table 2). As expected, high fluorescence quantum yields (Φ F ) between 83% and 93% with corresponding fluorescence lifetimes (τ F ) between 6.62 ns for 1(Me) and 7.41 ns for 1(Ph) were determined (Table 2; Figure S4 In contrast to molecularly dissolved dyes, absorption and emission studies on solid-state samples are challenging for chromophores with high tinctorial strength (ε max , µ eg ). This holds in particular true for the thicker crystals (with thicknesses of several hundred nanometers) where their high optical densities (OD) leads to strong reabsorption which has a pronounced impact on both absorption and emission spectra as illustrated in Figure 4 for 1(Me). [37] In Figure 4 we compare optical spectra obtained from spin-coated (10 −4 m, CH 2 CH 2 , 10 000 rpm) and solvent-treated vacuum-deposited (15 nm) thin films on quartz with spectra of an individual microcrystal grown from solution as well as an ensemble of single crystals obtained by gradient sublimation (for details see Supporting Information).
As could be expected from the single crystal structures of all PBIs, their microcrystals show strong anisotropy when studied with (fluorescence) polarization optical microscopy due to the unidirectional alignment of all chromophores in their respective crystal lattice (Figure 5; Figure S5, Supporting Information). However, while the gradient-sublimed crystalline material of PBIs 1(Me) and 1(Pr) show a dark bluish color, the powder of 1(Ph) appears rather brownish/red ( Figure S6, Supporting Information). This first indication for a smaller bathochromic shift of the absorption band for single-stranded PBI 1(Ph) is supported by the absorption profiles obtained from the reflectance spectra of gradient sublimed material in BaSO 4 trituration by the Kubelka-Munk theory [38] (Table 2, Figure 6). We like to note, that due to grinding partial deaggregation occurs and therefore these spectra are contaminated by less perfectly ordered PBIs. While all PBIs show bathochromically shifted absorption maxima compared to the monomers in solution, this shift is with 320 cm −1 (11 nm) for 1(Ph) less than half than for 1(Me) with 1640 cm −1 (61 nm). The same holds true for the microcrystal's emission spectra on quartz substrates ( Figure 6). This can be rationalized by the missing additional exciton coupling (J i ) to the π-stacked nearest neighbor in double-strand PBIs in the H-bonded chain arrangement of 1(Ph) (Figure 3). A quantitative assessment by Kasha's exciton theory [39] based on the point-dipole approximation applied on the data of the single crystal structures is provided in Figure S8 and Table S2 in the Supporting Information.
The absorption and excitation spectra of spin-coated and vacuum-deposited thin films of 1(Me) exhibit the same onset and additionally allow us to study the spectral band shape. [40] The thin-film spectra of 1(Me) (Figure 4, blue line) are almost identical to those previously reported for 1(Me) J-aggregate nanofibers in low-polar methylcyclohexane (MCH) [29] with a narrowed absorption band and an about 70 nm red-shifted absorption maximum at 651 nm ( Figure 4, gray dotted line). Fluorescence spectra of these thin films reveal a minor Stokes shift of only 160 cm −1 (λ em = 658 nm) for the thinnest spincoated layer. With increasing sample thickness, the emission maximum is strongly falsified and red-shifted up to 749 nm for an ensemble of the crystalline material due to reabsorption (Figure 4), as OD increases in case of 1(Me) by about 0.01 per nanometer PBI layer. Accordingly, the fluorescence maxima shift from the almost reabsorption-free spin-coated thin film (658 nm, OD = 0.04), the crystal grown from solution on quartz substrates (685 nm) over the vacuum-deposited thin film (690 nm, OD = 0.16) up to the bulk material (749 nm, OD > 10) by almost 1850 cm −1 (Figure 4). The emission maxima obtained from spectra of microcrystals grown from solution on quartz substrates of 1(Me) (685 nm), 1(Pr) (674 nm), and 1(Ph) (634 nm) further support the observation, that double-stranded PBIs exhibit larger bathochromic shifts than their singlestranded counterpart (Table 2, Figure 6).
Even more challenging is the determination of the absolute quantum yield (Φ F ) of the crystalline pigments, as the measurement setup including an integrating sphere is fundamentally flawed due to multiple reflections within the sphere and the large amount of strongly absorbing pigment needed. Values for Φ F of 2.0%, 1.7%, and 15.4% for PBIs 1(Me), 1(Pr), and 1(Ph), respectively, where determined upon excitation at 520 nm, which proves indisputably the higher solid-state  [41] using the respective microcrystal spectra; b) Fluorescence lifetimes (τ fl ) and their relative contributions (Rel.) to the fluorescence decay curves after excitation with an EPL laser diode at 505.8 nm in front face setup with 22.5° geometry; c) Data taken von ref. [29].; d) Ensemble property of crystalline material obtained from gradient sublimation. Absorption spectra and maxima (λ abs ) where determined by applying the Kubelka-Munk theory [38] on reflectance spectra of samples in BaSO 4 trituration; e) Microcrystal grown from solution on quartz substrate. fluorescence for the single-stranded, more heavily shielded PBI 1(Ph) ( Table 2; Figure S6, Supporting Information). Corrections for reabsorption [41] were conducted yielding lower limits of 6%, 3% and 59% for 1(Me), 1(Pr), and 1(Ph), respectively, under the assumption that the emission profile of the individual microcrystals match the intrinsic solid-state fluorescence spectra (Figures 4 and 6). The value of 6% for 1(Me), albeit significantly reduced in comparison to the value of its J-aggregates in MCH solution (77%), [29] might still be considered high because fast exciton diffusion in such J-type coupled chromophore architectures [34,35] may lead to nonradiative relaxation on unavoidable trap sites in addition to other known relaxation pathways such as strong intermolecular electronic coupling [42] or singlet fission into triplet states. [22] Further insights are obtained from fluorescence lifetime (τ F ) measurements for the crystalline materials, which can be recorded without any contamination by reabsorption. Fluorescence decay curves were recorded by time correlated single photon counting (TCSPC) with an EPL laser diode (505.8 nm) in front-face geometry (22.5°) through detection at higher wavelengths than 711 nm ( Figure S7, Supporting Information). While, for all PBIs the τ F are significantly reduced, only singlestranded PBI 1(Ph) shows still monoexponential decay with the longest τ F = 2.65 ns (Table 2), which is about one third of the τ F of monomers in solution (7.41 ns). By contrast, PBIs 1(Me) and 1(Pr) show multiexponential decay, whose major components are 0.25 and 0.26 ns, respectively. This tenfold decrease must be related to stronger electronic interactions within doublestranded packing arrangements of the less-shielded PBIs 1(Me) and 1(Pr) which beyond exciton coupling may also include charge transfer interactions. [26] Figure 6. Normalized absorption spectra (dotted lines) determined by applying the Kubelka-Munk theory [38] on the reflectance spectra of BaSO 4 trituration of PBIs a) 1(Me) (red), b) 1(Pr) (orange), and c) 1(Ph) (violet). The monomer absorption spectrum of 1(Me) in CH 2 Cl 2 (gray dotted line) as well as the normalized fluorescence spectra (solid lines) of microcrystals grown from solution on quartz substrates are depicted as well. Inset: respective fluorescence polarization microscope images of the microcrystals grown on quartz substrates. Scale bar: 10 µm.

Conclusion
In conclusion, we were able to grow solvent-free single crystals of three H-bonded tetraphenoxy-substituted perylene bisimide dyes by gradient sublimation in vacuum which adopt unique packing arrangements, i.e., double-(1(Me), 1(Pr)) and single-(1(Ph)) strands in the solid state of strongly J-coupled chromophores. These H-bonded chains are composed of only one atropo-enantiomer (P or M), with neighboring strands always containing the opposite stereoisomer. This heterochiral ("racemic") π-stacking motif is in contrast to the narcissistic self-sorting [43] observed for related dyes with solubilizing chains in solution and LC phases, where consistently homochiral helical aggregates composed of the same enantiomer are observed. Double-stranded systems show slip-stacked packing arrangement of π-stacked chromophores, while the individual strands of PBI 1(Ph) are isolated by their phenyl-residues in orthoposition. Comparison of the spectroscopic properties in solution, crystalline powders as well as of individual single crystals revealed a decrease in effective fluorescence quantum yield in solid state compared to solution, however less severe for singlestranded PBI 1(Ph) with 59% in comparison to double-stranded PBI 1(Me) with 6%. Double-stranded J-aggregates of 1(Me) and 1(Pr) exhibit larger bathochromic shifts as well as strongly reduced fluorescence lifetimes (<0.26 ns) in contrast to singlestranded PBI 1(Ph) (2.65 ns). These materials are interesting candidates for future studies on the exciton dynamics in welldefined 1D chains by ultrafast spectroscopy [35] and for applications in light amplification [44] and single crystal lasing. [6]

Experimental Section
Synthesis and Characterization: The synthesis of precursors and of compound 1(Me) has been described previously. [29,46] The synthesis and characterization of 1(Pr) and 1(Ph) is given in the Supporting Information.
Spectroscopy in Solution: UV/vis absorption spectra were measured on a Lambda 950 spectrophotometer (Perkin-Elmer). Fluorescence spectra were recorded on an Edinburgh Instruments FLS980-D2D2-ST spectrometer and were corrected against the photomultiplier sensitivity and the lamp intensity. Fluorescence lifetimes were determined with an EPL picosecond pulsed diode laser (λ ex = 505.8 nm) for TCSPC with an Edinburgh Instruments FLS980-D2D2-ST spectrometer. Measurements were performed using 1 mm or 1 cm quartz cuvettes (SUPRASIL) and spectroscopic solvents (Merck). UV/vis absorption and emission spectra of 1(Me), 1(Pr), and 1(Ph) were measured in CH 2 Cl 2 and MCH at RT.
Spectroscopy in Solid State: Optical investigations were performed on a Zeiss Axio Imager optical polarization microscope. Emission spectra of individual single crystals were measured on the same instrument with an integrated Zeiss CCD spectrometer customized by A.S. & Co with a spot resolution of up to 4 × 4 µm 2 . Thin films for optical investigations of 1(Me) were prepared as follows: 1) A saturated solution of 1(Me) in CHCl 3 was filtered (polytetrafluoroethylene (PTFE) filters, 0.22 µm). Then spin-coating was performed at 10 000 rpm, 5000 rpm s −1 for 60 s. Suprasil quartz plates (Type: 202-QS) were used as substrates. Microcrystal Growth: Microcrystals of 1(Me) and 1(Pr) were grown based on the slow evaporation method. [46] A saturated solution of PBI was prepared and filtered (PTFE, 0.22 µm) and then mixed with MeOH in a ratio of 1:1 (v/v). Screw caps were filled with 250 µL of CHCl 3 :MeOH in a ratio of 1:1 (v/v). Suprasil quartz plates (Type: 202-QS) were used as substrates. It took three days for the solvent to evaporate completely from the closed vessels. Single crystals of 1(Ph) were grown by drop-casting the compound on quartz plates from a CHCl 3 solution (2 mg mL −1 ) and subsequent solvent vapor annealing with CHCl 3 for 30 min.
Single Crystal Growth: Solvent-free single crystals of 1(Me), 1(Pr), and 1(Ph) were obtained by vacuum sublimation, using a thermal gradient sublimation system of HTM Reetz GmbH. About 5 mg of raw material were placed into a standard NMR tube as a fine powder. Small pieces of the (100) wafers with a width of 2-3 mm and a length off around 10 mm were placed into the NMR tube (polished side facing up). The NMR tube in turn was positioned centrally between the 1st and the 2nd heating zone of the thermal gradient sublimation system. Prior to heating, the system was evacuated for about 2 h until a pressure of at least 2 × 10 −6 mbar was reached. Then the 1st and the 2nd heating zone were heated up with a rate of ≈10 K min −1 . The experiment was stopped when single crystals of sufficient size had grown, mainly on the edges of the wafer pieces facing toward the organic material.
X-Ray Diffraction: X-ray diffraction measurements of 1(Me), 1(Pr), and 1(Ph) were performed on a D8 Quest Kappa diffractometer of Bruker working with a Photon100 complementary metal-oxide-semiconductor (CMOS) or with a Photon II CMOS detector, Cu Kα radiation (0.15418 nm) and at a temperature of 100 K. Crystal structures were solved using direct methods, expanded by Fourier techniques and were refined with the Shelx software package. [47] Anisotropical refinement was performed on nonhydrogen atoms, while hydrogen atoms were included in the structure factor calculation on geometrically idealized positions. Powder XRD measurements were performed on a D8 Discover (Bruker) with a LynxEye-1D-Detector using Cu Kα radiation (0.15418 nm). Powder XRD patterns were measured of the crystalline bulk material of 1(Me), 1(Pr), and 1(Ph) obtained by thermal gradient sublimation. To ensure high crystallinity of the investigated solid-state materials powder XRD measurements were performed on the purified and solvent-free PBIs after thermal gradient sublimation and compared them to the simulated diffraction patterns of their respective single crystal structures. The complementarity of these patterns without any significant additional peaks led to the conclusion that the majority of bulk material inherits the desired molecular packing as in the single crystal structures.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.