Crystal engineering rescues a solution organic synthesis in a cocrystallization that confirms the configuration of a molecular ladder

Abstract

Treatment of an achiral molecular ladder of C_2h symmetry composed of five edge-sharing cyclobutane rings, or a [5]-ladderane, with acid results in cis- to trans-isomerization of end pyridyl groups. Solution NMR spectroscopy and quantum chemical calculations support the isomerization to generate two diastereomers. The NMR data, however, could not lead to unambiguous configurational assignments of the two isomers. Single-crystal X-ray diffraction was employed to determine each configuration. One isomer readily crystallized as a pure form and X-ray diffraction revealed the molecule as being achiral based on C_i symmetry. The second isomer resisted crystallization under a variety of conditions. Consequently, a strategy based on a cocrystallization was developed to generate single crystals of the second isomer. Cocrystallization of the isomer with a carboxylic acid readily afforded single crystals that confirmed a chiral ladderane based on C₂ symmetry. The chiral ladderane and acid self-assembled to generate a five-component hydrogen-bonded complex that packs to form large solvent-filled homochiral channels of nanometer-scale dimensions. Whereas cocrystallizations are frequently applied to structure determinations of proteins, our study represents the first application of a cocrystallization to confirm the relative configuration of a small-molecule diastereomer generated in a solution-phase organic synthesis.

Crystal engineering involves the rational design of solids with predictable and/or tunable properties (1, 2). During the past two decades, crystal engineering has undergone remarkable growth with applications in areas such as catalysis (3), energy storage (4), electronics (5, 6), and pharmaceutics (7, 8). A burgeoning area in the field of crystal engineering involves the design and properties of cocrystals (7–11). Cocrystals are multicomponent solids with organic compounds assembled in combination to form a crystalline solid with properties different than the individual components (9–11). A cocrystal typically consists of a target molecule crystallized with a second molecule, or cocrystal former (CCF), employed to influence properties of the target (e.g., solubility, conductivity). The CCF interacts with the target via intermolecular forces (e.g., hydrogen bonding) that serve to unite and hold the components together.

In this context, we have described a cocrystal approach to synthesize [n]-ladderanes (where: n = 3 or 5) in the solid state (12, 13). The [n]-ladderanes are rod-shaped molecules composed of n edge-fused cyclobutane rings that define a molecular equivalent of a macroscopic ladder (14, 15) (Scheme 1). Ladderanes are promising building blocks in optoelectronics and have been recently discovered as building blocks of natural products, being present as lipids in anammox (anaerobic ammonium oxidizing) marine bacteria (16–18). The ladderane lipids serve a structural role of providing extraordinary rigidity to internal membrane components. More specifically, we have shown that a CCF based on 1,3-dihydroxybenzene, or resorcinol (res), acts as a template to assemble polyenes via hydrogen bonds into supramolecular assemblies for [2 + 2] photocycloadditions that generate [3]- and [5]-ladderanes (12). Cocrystallization of 5-methoxyresorcinol (5-OMe-res) with either an all-trans-4-pyridyl-substituted 1,4-butadiene or 1,6-hexatriene yielded assemblies of composition 2(template)·2(polyene). UV-irradiation produced the corresponding end-functionalized C_2h-symmetric ladderane 4-pyr[n]-lad stereospecifically, in quantitative yield, and in gram amounts. The ladderanes we generate in the solid state are cumbersome to make using conventional solution methods of organic synthesis. Difficulties are evidenced by the fact that ladderanes lacking internal functional groups, such as those in the natural products, are rare, with exceptions being ladderanes derived from intramolecular reactions of cyclophanes (14, 15). Recent reports by Corey and coworkers on the first total and enantioselective syntheses of ladderane lipids in the form of (±)-pentacycloanammoxic acid, as well as attempted syntheses from polyenes in solution and the solid state, highlight these difficulties (19–21).

Scheme 1.

In this paper, we report an application of a cocrystallization to determine the relative configuration of a chiral ladderane generated in a solution-phase isomerization. Our interests lie in using ladderanes synthesized in the solid state as precursors to the natural lipids. During experiments to treat 4-pyr-[5]-lad with acid, the terminal pyridyl groups were determined to undergo a cis- to trans-isomerization that generates two isomers. Multidimensional solution NMR spectroscopy—a tool commonly used to elucidate the structures of organic molecules—supported the isomerization, along with quantum chemical calculations, to produce two diastereomers; namely, achiral 1a and chiral 1b (Scheme 2). The NMR data, however, could not be used to unambiguously determine the relative configuration of each isomer owing to close structural similarities of the two molecules.

Scheme 2.

Thus, to confirm the structures of the two ladderanes, we turned to single-crystal X-ray diffraction. Moreover, whereas one isomer readily crystallized and was confirmed as achiral 1a, the second isomer resisted crystallization under a variety of conditions. To circumvent difficulties to obtain single crystals, we revisited principles of crystal engineering by employing a cocrystallization (Scheme 3). Cocrystallizations are often used to determine structures of biomolecules (e.g., proteins) (22), yet applications to parallel problems involving small molecules have been less explored, with absolute structure determinations of steroids (23, 24) and, more recently, separations and structure determinations of natural products (25, 26) and a bicyclobutyl being the only examples (27). Here, a cocrystallization of the second isomer with 3,5-dinitrobenzoic acid (3,5-DNBA) readily afforded single crystals that enabled the relative configuration of chiral 1b to be confirmed. The isomer 1b details a unique case wherein chirality of a ladderane is generated from the dispositions of identical end substituents. The components of the cocrystal form a hydrogen-bonded complex that packs to give a homochiral solid with solvent-filled channels of nanometer-scale dimensions. We expect the cocrystal approach described here to be applicable to similar problems of structure determination in organic chemistry where single crystals are difficult to obtain and structure cannot be confirmed using spectroscopic methods alone.

Scheme 3.

Results and Discussion

Our study begins with achiral C_2h-symmetric 4-pyr-[5]-lad, which is obtained using 5-OMe-res as a template. Treatment of a related rctt-monocyclobutane was reported (28) to result in a cis- to trans-isomerization of pyridyl groups to give the rtct-stereoisomer in quantitative yield. Our goal was to determine if treatment of 4-pyr-[5]-lad with acid could generate more complex ladderanes as products.

When 4-pyr-[5]-lad, upon removal from the template, was reacted with glacial acetic acid (AcOH) for 12 h at 50 °C, a ¹H NMR spectrum suggested that the ladderane underwent an isomerization (Fig. 1). An isomerization was evidenced by the disappearance of peaks of 4-pyr-[5]-lad and multiplication of the ¹H resonances in the pyridyl (7.00–8.50 ppm) and polycyclobutyl (2.40–4.50 ppm) regions (Fig. 1A). A ¹H-¹H COSY spectrum of the reaction mixture revealed 12 groups of resonances in the polycyclobutyl region with two sets of six groups being of the same relative intensity and correlated as two separate J-coupling networks (Fig. 1B). The two coupling networks suggested the reaction generated two isomers in a 1∶1 ratio. When the reaction was conducted at 25 °C, the ¹H NMR spectrum was more complex with additional resonances in both the pyridyl and polycyclobutyl regions (SI Text). Collectively, our observations are consistent with the terminal 4-pyridyl groups of 4-pyr-[5]-lad undergoing an acid-catalyzed cis- to trans-isomerization that initially generated partially isomerized unsymmetrical ladderanes that converted to two symmetrical isomers in equal amounts.

Fig. 1.

NMR spectra of reaction of 4-pyr-[5]-lad with AcOH. (A) The ¹H spectrum showing the pyridyl and polycyclobutyl regions of reactant and products and (B) ¹H-¹H COSY spectrum of products showing two separate coupling networks (dashed and nondashed arrows correspond to a single group).

In principle, a cis- to trans-conversion of the end groups of 4-pyr-[5]-lad can generate two stereoisomers. One isomer 1a contains pyridyl groups that point in opposite directions at the top and bottom halves of the molecule. The isomer 1a is achiral, being based on inversion symmetry (C_i symmetry). The second isomer 1b possesses pyridyls that point to the corners of a tetrahedron. The elongated shape of the ladderane in combination with the orientation of the pyridyl groups prohibits an improper rotation axis in the molecule. As a result, 1b is chiral (C₂ symmetry). From quantum chemical calculations, 1a and 1b are determined to be nearly equal in energy (ΔE difference approximately 0.27 kcal/mol) and lower than the parent 4-pyr-[5]-lad (approximately 5 kcal/mol). The relatively low energies can be attributed to the more sterically favorable trans-configuration of the end pyridyl groups. Isomers 1a and 1b are, thus, closely related diastereomers based on constitutionally identical centers of chirality (C), with the senses of the chirality being opposite (C_RC_RC_SC_S) and identical (C_RC_RC_RC_R or C_SC_SC_SC_S) at the corners of the ladderanes, respectively. Given that isomer 1a is achiral, 1a is a meso form of the ladderane. Notwithstanding the ladderanes derived from the marine bacteria (14–21), 1b details a unique case of how terminal substituents generate a chiral ladderane.

Whereas ¹H NMR data and quantum chemical calculations were consistent with an isomerization of 4-pyr-[5]-lad that generates two products, attempts to separate the isomers were initially unsuccessful. Thin-layer and column chromatographies using solvents of different polarities (e.g., ethyl acetate, methanol, hexanes) produced yellow streaks that were unresolvable using solvent combinations. To further our efforts, we turned to solubility experiments. Moreover, by subjecting the reaction mixture to different crystallization conditions, we anticipated that one of the ladderane isomers could be isolated via selective crystallization. A selective crystallization would be particularly favored given that only two isomers formed in the reaction.

When the crude solid from the reaction was dissolved in acetonitrile, a white precipitate formed. A ¹H NMR spectrum of the solid showed, in contrast to the reaction mixture, only two doublets in the pyridyl region and six resonances in the polycyclobutyl region (Fig. 2). The peaks in the pyridyl and polycyclobutyl regions were of equal intensities (Fig. 2A) whereas a ¹H-¹H COSY experiment revealed the signals to correlate as a single J-coupling network. A ¹H NMR spectrum of the solid from the remainder of the reaction mixture, obtained by allowing the acetonitrile solution to evaporate, showed very similar sets of peaks that corresponded to a single product (Fig. 2B). Overall, the splitting patterns, relative intensities, and number of peaks in each spectrum were consistent with an isomerization of 4-pyr-[5]-lad that generated 1a and 1b.

Fig. 2.

The ¹H NMR spectra of isolated ladderanes of reaction of 4-pyr-[5]-lad with AcOH: (A) initial precipitate and (B) remaining solid of mixture (solvent: DMSO-d₆).

Although we managed to separate the two ladderanes via a selective crystallization, the NMR data could not be used to unambiguously confirm the relative configuration of each isomer. NOESY data were inadequate to identify stereochemical differences whereas Ha and Ha′, which reflect relative orientations of the cyclobutyl protons adjacent to the pyridyls, exhibited indistinguishable J-coupling and splitting patterns (SI Text). As a result, we next attempted to grow single crystals of each isomer to employ X-ray diffraction to determine each configuration.

From extensive crystallization studies, we obtained single crystals of the initial white precipitate by slow evaporation of a benzene solution over a period of ca. 2 d. Moreover, a single-crystal X-ray diffraction study confirmed the stereochemistry as achiral 1a in the form of the solvate 1a·2(benzene). As expected, the pyridyl groups of the ladderane, which sit around a crystallographic center of inversion, point in opposite directions at the top and bottom of the molecule (Fig. 3). The C–C lengths and C–C–C angles of the cyclobutanes compare favorably to 4-pyr-[5]-lad (12). The ladderanes assemble in the solid via C-H⋯N and C-H⋯π forces to form layers separated by disordered benzene molecules.

Fig. 3.

X-ray structure of achiral ladderane 1a: (A) ball-and-stick, (B) space-filling, and (C) bilayer packing with benzene. Selected C–C distances (Å): C6–C7 1.563(2), C8–C17 1.569(2), C9–C16 1.586(2), C10–C15 1.587(2), C11–C14 1.577(2), C12–C13 1.553(2).

Whereas the configuration of 1a was confirmed from a benzene solvate, the second isomer repeatedly resisted crystallization. Attempts to grow crystals from different solvents consistently afforded yellow oils and amorphous solids. That the compound resisted crystallization, however, provided a measure of support for the isomer being chiral because molecules of relatively low symmetry tend to pack less favorably in a lattice (29).

To generate single crystals of the ladderane, we decided to revisit principles of crystal engineering by employing a cocrystallization. In particular, we hypothesized that cocrystallization of the ladderane with a CCF would provide a means to modify the surface of the ladderane so as to extend the available crystallization energy landscape of the molecule (22). A CCF would enable us to effectively embed the ladderane into a lattice different compared to that provided by the pure molecule and, thereby, improve the probability of obtaining an ordered solid.

For a CCF, we selected a carboxylic acid, which was expected to interact with the ladderane via the robust pyridyl⋯carboxylic acid supramolecular synthon (Scheme 4) (25). Cocrystallization of the ladderane with a monoacid (1∶4 ratio) was expected to generate a two-component solid wherein the ladderane and acid self-assemble via O-H(acid)⋯N(pyridyl) hydrogen bonds. Each ladderane would interact with up to four acid molecules in a discrete five-component complex. The interaction between a CCF and ladderane would enable the components to pack more efficiently relative to the pure ladderane. Whereas cocrystallizations are often successfully used to aide the elucidation of the structures biological molecules (22), the application of a cocrystal strategy to determine the relative configuration of a diastereomer had not been reported (23–27).

Scheme 4.

Whereas crystallization of the ladderane alone afforded an amorphous solid, cocrystallization with 3,5-dinitrobenzoic acid (3,5-DNBA) (ratio: 1∶4) in 1∶3 (v/v) methanol∶nitromethane readily afforded colorless single crystals upon slow evaporation over 1 d (30, 31). A ¹H NMR spectrum demonstrated both the ladderane and 3,5-DNBA to be present in a 1∶4 ratio in the solid whereas an infrared spectrum ruled out possible salt formation, with the acid groups (C = O solid: 1,703 cm^-1) being present in the neutral form (C = O neutral: 1,708 cm^-1) (30, 31). Importantly, a single-crystal X-ray diffraction study of the cocrystal confirmed the stereochemistry as chiral 1b in crystalline (1b)·4(3,5-DNBA)·5(CH₃NO₂).

The components of the cocrystal crystallize in the chiral trigonal space group P3₂21 (Fig. 4). The asymmetric unit consists of one half of a molecule of 1b and two molecules of 3,5-DNBA, with one acid being disordered over two sites (occupancies: 0.62∶0.38). As expected, the ladderane and acid interact via four O-H(acid)⋯N(pyridyl) hydrogen bonds [O-H⋯N distances (Å): O(1)⋯N(1) 2.566(8), O(7a)⋯N(2) 2.58(1), O(7b)⋯N(2) 2.57(2)] (Fig. 4A). The pyridyl groups of the chiral [5]-ladderane point to the corners of a distorted tetrahedron [edges (Å): 6.57 and 7.81] with the C–C bond lengths and C–C–C angles being comparable to 1a (Fig. 4B). The geometry of the carbonyl group of the ordered 3,5-DNBA also supports the acid being in the neutral form (d(C = O) = 1.185(9) Å; d(C-O) = 1.315(7) Å). The spontaneous resolution of the components to form a chiral solid presumably afforded a racemic conglomerate (32).

Fig. 4.

X-ray structure of chiral ladderane 1b determined in the cocrystal (1b)·4(3,5-DNBA)·5(CH₃NO₂): (A) ORTEP of ladderane, (B) five-component hydrogen-bonded assembly, and (C) 1D channels along c-axis. Selected C–C distances (Å): C6–C11 1.549(8), C7–C10 1.618(9), C8–C9 1.567(9) (dark gray = [5]-ladderane; light gray = 3,5-DNBA).

Whereas the ladderane and acid form a discrete assembly, the complexes pack to produce large homochiral channels (diameter: approximately 1.7 nm) (Fig. 4C) with walls composed of alternating ladderane and 3,5-DNBA molecules. Peaks associated with solvent molecules located in the channels were determined to be highly diffuse so as to not allow the solvent to be modeled. The X-ray data were, thus, treated with the SQUEEZE routine (33) and determined to correspond to 15 CH₃NO₂ molecules per unit-cell, with the solvent-accessible voids accounting for ca. 23% of the unit-cell volume (i.e., 1,251 Å³). The ability of a ladderane to function as a building block of a host solid has not been reported (34). Having established the relative configuration of both 1a and 1b from our X-ray experiments, we were able to fully assign the ¹H NMR data (SI Text).

In this report, we have described an application of a cocrystallization to determine the relative configuration of a diastereomer of a chiral [5]-ladderane that resists crystallization as a pure form. The ladderane is one of two isomers with relative configurations that could not be assigned using spectroscopy alone. Given that a wide variety of supramolecular synthons are now recognized to sustain the structures of multicomponents solids (1, 2), we expect cocrystallizations to experience more widespread applicability to problems of structure determination in organic synthetic chemistry. We also expect the isomerization reported here to be useful for studies that focus upon the properties of ladderanes, where applications of ladderanes as building blocks in materials science, synthetic chemistry, and biology are being actively pursued (14–21).

Materials and Methods

Reaction of 4-pyr-[5]-lad with AcOH.

In a closed reaction vessel 4-pyr-[5]-lad (100 mg, 0.21 mmol) was dissolved in 1 mL of AcOH. The solution was stirred for 12 h at 50 °C. The solution was quenched by addition of excess 1 M KOH (30 mL) to afford an off-white precipitate. The precipitate was filtered, washed with 30 mL of distilled water, and dried to yield 97 mg of solid.

Isolation of 1a.

The precipitate from the reaction of 4-pyr-[5]-lad with AcOH was dissolved in 10 mL of acetonitrile and allowed to slowly evaporate overnight to afford 1a as a white precipitate in 35% yield (35 mg, 0.07 mmol). ¹H NMR (DMSO-d₆, 600 MHz) δ 8.52 (d, J = 5.6 Hz, 1H), 8.48 (d, J = 5.4 Hz, 1H), 7.34 (d, J = 5.4 Hz, 1H), 7.24 (d, J = 5.6 Hz, 1H), 4.20 (dd, J = 8.7 Hz, 8.7 Hz 1H) two peaks are overlapped, 3.97 (dd, J = 8.9 Hz, 3.1 Hz 1H), 3.20 (dd, J = 8.4 Hz, 3.8 Hz 1H), 2.91 (s, 1H), 2.77 (s, 1H), 2.61 (s, 1H); ¹³C NMR (DMSO-d₆, 600 MHz) δ 153.2, 149.5, 149.4, 149.2, 122.0, 121.6, 48.3, 46.9, 46.1, 43.1, 41.8, 41.4. Single crystals of 1a·2(benzene) were obtained in benzene by hot filtration and slow evaporation of the solution (3 mL/50 mg) for a period of 1 d.

Isolation of 1b.

The supernatant from the acetonitrile solution that afforded 1a was evaporated to dryness to give an off-white yellow solid (35 mg, 0.07 mmol) of 1b in 35% yield (35 mg, 0.07 mmol). ¹H NMR (DMSO-d₆, 600 MHz) δ 8.51 (d, J = 6.0 Hz, 1H), 8.48 (d, J = 5.4 Hz, 1H), 7.37 (d, J = 5.4 Hz, 1H), 7.22 (d, J = 6.0 Hz, 1H), 4.18 (dd, J = 8.7 Hz, 8.6 Hz 1H) two peaks are overlapped, 4.00 (dd, J = 8.7 Hz, 3.2 Hz 1H), 3.17 (dd, J = 8.6 Hz, 3.6 Hz 1H), 2.90 (s, 1H), 2.87 (s, 1H), 2.48 (s, 1H); ¹³C NMR (DMSO-d₆, 600 MHz) δ 153.3, 149.8, 149.7, 149.6, 122.4, 122.1, 48.1, 47.2, 46.1, 43.8, 42.4, 42.1. Cocrystals of composition (1b)·4(3,5-DNBA)·5(CH₃NO₂) were obtained by a cocrystallization with 3,5-DNBA (63.3 mg, 0.29 mmol) in 1∶3 mL (v∶v) (1 mL/25 mg) methanol∶nitromethane. The solution was allowed to evaporate slowly for a period of 3 d to afford transparent colorless needles. The crystals were allowed to sit in 20 mL 1 M KOH solution for a period of 2 d. The white solid was filtered and allowed to dry to give clean 1b in 9% yield (9 mg, 0.02 mmol).

NMR Characterization

All products were characterized using Avance-400 and Avance-600 Bruker NMR spectrometers operating at 400 and 600 MHz, respectively. ¹H and ¹³C chemical shifts were referenced (x) with the residual proton and carbon chemical shifts of the solvents (DMSO-d₆, ¹H, 2.50 ppm; ¹³C, 39.5 pm). Fractions with milligram quantities of the product were characterized by a battery of 1D and 2D homonuclear and ¹H-¹³C heteronuclear experiments [¹H, 1D correlated spectroscopy (COSY), nuclear Overhauser effect spectroscopy (NOESY), ¹³C, ¹³C-distortionless enhancement by polarization transfer (DEPT), heteronuclear multiple quantum coherence (HMQC), and heteronuclear multiple bond correlation (HMBC)] (35). Gradient-assisted versions of the pulse sequences and inverse detection were used for these 2D experiments. Typical parameters for the NMR experiments were as follows: ¹H [time domain data points (TD), 64k; number of scans (NS), 4k], ¹³C (TD, 128k; NS, 10,000), ¹³C-DEPT (TD, 128k; NS, 5,000), 1D COSY (TD, 64k; NS, 512), NOESY [TD, 2k; TD1, 256; NS, 32; dummy scans (DS), 32; mixing times, 1.0, 1.5, and 2.0s], ¹³C-1H HMQC (TD, 2k; TD1, 128; NS, 32; DS, 128) and ¹³C-¹H HMBC (TD, 2k; TD1, 128; NS, 32; DS, 128). TD, NS, and DS refer to time domain data points, number of scans, and dummy scans, respectively. All NMR data were processed with TOPSPIN 1.3 suite of software programs. The 1D ¹H data were processed with zero-filling to 64k data points and 0.2 Hz exponential line broadening, whereas ¹³C spectra were processed with zero-filling to 128k data points and 1.0 Hz of exponential line broadening. The 2D NMR data were processed with the zero-filling to 2,048 points and 1,024 points in acquisition and second dimension, respectively. Relative numbers of proton signals multiplied by the integral areas were used for the quantification.

IR Spectroscopy.

Fourier transform infrared spectra were recorded from NaCl plates on a Nicolet 380 FT-IR spectrometer. OMNIC software was employed for data analysis using 4 cm^-1 resolution and the averaging of 8 scans. All spectra were normalized and baseline-corrected.

Computational Chemistry.

Ground state geometry optimizations were performed in the gas phase using spin restricted resolution of identity (RI) B-P functional combined with TZVP basis set (36, 37). Symmetry constraints were used during the geometry optimization. Geometry optimization convergence was signaled when energy change between two consecutive optimization iterations was less than 10^-6 and change in gradient was less than 10^-3 atomic units. All geometry optimizations were performed using TURBOMOLE (38). Accurate single point energies were calculated in the gas phase using spin restricted RI approximation combined with MP2 level of theory and TZVP basis set using RI-BP/TZVP obtained geometries (37).

X-ray Crystallography.

All crystal data were collected on a Nonius KappaCCD single-crystal X-ray diffractometer using MoK_α radiation (λ = 0.7107 Å). Data collection, cell refinement, and data reduction were performed using Collect (39) and HKL Scalepack/Denzo, (40) respectively. Structure solution was accomplished with the aid of SHELXS-97, whereas refinement by full-matrix least-squares based on F² was conducted using SHELXL-97 (41). All non-hydrogen atoms were refined anisotropically. Hydrogen atoms attached to C- and O-atoms were fixed in geometrically constrained riding positions, and refined isotropically on the basis of corresponding C- and O-atoms [U(H) = 1.2 U_eq(C), U(H) = 1.5 U_eq(O)]. Structure solution and refinement for (1b)·4(3,5-DNBA)·5(CH₃NO₂) revealed large channels. Further refinement uncovered relatively weak electron density (< 0.85 eA^-3) in the Fourier map. The peaks were attributed to diffuse solvent molecules. Even after all electron-density peaks were assigned to solvent molecules, the structure exhibited solvent-accessible voids, which suggested that partial desolvation occurred during the X-ray experiment. The ill-defined solvent peaks are in agreement with the poor diffraction quality. Because a reasonable model of the solvent could not be achieved, the data were treated with the SQUEEZE routine in PLATON (33) to remove electron density related to the diffuse solvent molecules. Four solvent-accessible voids that collectively account for 23% of the unit-cell volume (i.e., 1,251 Å³) and 483 electrons, which correspond to approximately 15 molecules of CH₃NO₂ per unit cell, were identified. The sites of the disordered carboxylic acids were restrained to be the same as the ordered group. The occupancies of the disordered molecule refined to 0.62(1) and 0.38(1). Restraints were used on the thermal parameters of both the ordered and the disordered benzoic acids. Details of all structural analyses are summarized in Table 1.

Table 1.

Crystallographic parameters for 1a·2(benzene) and (1b)·4(3,5-DNBA)·5(CH₃NO₂)

Compound reference	1a·2(benzene)	(1b)·4(3,5-DNBA)·5(CH3NO2)
Chemical formula	(C32H28N4)·2(C6H6)	C32H28N4·4(C7H4N2O6)·5(C1H3N1O2)
Formula Mass	624.8	1,622.29
Crystal system	monoclinic	trigonal
a/Å	17.8888(19)	29.740(3)
b/Å	9.5512(11)	29.740(4)
c/Å	20.427(3)	7.0681(8)
α/°	90	90
β/°	100.356(5)	90
γ/°	90	120
Unit cell volume/Å3	3,433.3(7)	5,414.0(11)
Temperature/K	190(2)	150(2)
Space group	P21/c	P3221
Number of formula units per unit cell, Z	4	3
Radiation type	MoKα	MoKα
Absorption coefficient, μ/mm-1	0.071	0.124
Number of reflections measured	22,724	29,039
Number of independent reflections	6,040	3,379
Rint	0.0563	0.0550
Final R1 values (I > 2σ(I))	0.0497	0.0828
Final wR(F2) values (I > 2σ(I))	0.1190	0.2199
Final R1 values (all data)	0.0924	0.1095
Final wR(F2) values (all data)	0.1339	0.2377
Goodness of fit on F2	1.008	1.056
Largest diff. peak and hole/eA-3	0.152, −0.198	0.305, −0.280
CCDC number	768,180	768,181