OD- and non-OD-polytypism of 9-(3-chloropyridin-4-yl)-9H-carbazole

Synthesis and crystal growth

NaO^tBu (2.40 g, 25.00 mmol) was added to a solution of 4-amino-3-chloropyridine (643 mg, 5.00 mmol), 2,2′-dibromo-1,1′-biphenyl (1.56 g, 5.00 mmol), Pd₂(dba)₃ (183 mg, 0.20 mmol) and 1,1′-bis(diphenylphosphino)ferrocene (222 mg, 0.40 mmol) in degassed toluene (20 mL) under argon atmosphere in a three-necked flask. The reaction mixture was refluxed for 16 h. After cooling, the reaction mixture was filtered through a celite pad and washed with CH₂Cl₂. The filtrate was concentrated under reduced pressure and purified by column chromatography (light petrol/CH₂Cl₂ 1:1) yielding 1 as white solid (1.03 g, 3.70 mmol, 74%). Single crystals of 1 were grown by recrystallization from EtOH.

¹H NMR (CDCl₃, 400 MHz): δ=8.93 (s, 1H), 8.72 (d, J=5.1 Hz, 1H), 8.16 (d, J=7.8 Hz, 2H), 7.51 (d, J=5.1 Hz, 1H), 7.44 (ddd, J=8.2 Hz, 7.0 Hz, 1.2 Hz, 2H), 7.35 (ddd, J=7.8 Hz, 7.0 Hz, 0.8 Hz, 2H), 7.16 (d, J=8.2 Hz, 2H) ppm. ¹³C NMR (CDCl₃, 100 MHz): δ=152.0, 149.3, 142.9, 139.7, 130.2, 126.2, 124.4, 123.9, 120.9, 120.5, 110.1 ppm. HRMS: calculated for C₁₇H₁₁ClN₂ [M+H]⁺ 279.0684; found 279.0687.

NMR spectra were recorded on a Bruker Avance DRX-400 spectrometer. An Agilent 6230 LC TOFMS mass spectrometer equipped with an Agilent Dual AJS ESI-Source was used for HRMS.

Single crystal diffraction

Crystals of 1 were selected under a polarizing microscope, embedded in perfluorinated oil and attached to Kapton^® micromounts. Data were collected at 100 K in a dry stream of nitrogen on a Bruker KAPPA APEX II CCD diffractometer system [4] using graphite monochromatized MoKα¯ radiation. Data reduction was performed with SAINT-Plus [4] and corrections for absorption effects applied with SADABS or TWINABS [4], as applicable. The structures were solved using the dual-space method implemented in SHELXT [5] and refined with JANA2006 [6].

In preliminary experiments, multiple platy crystals were subjected to short ω-scans. The crystals featured arcing of reflections along 2θ, a sharp intensity drop-off at higher diffraction angles and characteristic signs of twinning (splitting of reflections). In addition to these plates, the sample contained a single crystal with a distinct columnar habit. This crystal diffracted to distinctly higher angles than the platy crystals and did not feature splitting of reflections. Therefore, a routine data collection and refinement was performed. The structure with the orthorhombic Ic2a symmetry will henceforth be called polytype I. The non-standard setting of the Iba2 space group was chosen for comparability with the other polytype.

Because the reason of systematic twinning could not be derived from the structure of polytype I, a structure determination using a platy crystal was attempted. To obtain reasonable intensities, a rather large plate was measured first, which featured distinct arcing (Figure 2a). The crystal was a twin of two domains with monoclinic P2₁ symmetry, called polytype II. Even though a structure model could be derived from this data set, the data quality was unsatisfactory. Only 94% of intensity data were determined, owing to difficulties during data reduction caused by smearing of reflections. Moreover, the estimated standard uncertainty (ESU) on the cell parameters and bond lengths were large. Therefore, a tiny plate without arcing of reflections (Figure 2b) was measured next with long exposition times. Despite distinctly weaker intensities, ESUs improved significantly and a complete data set could be obtained. The structure model of polytype II derived from this second plate will constitute the basis of the discussion below.

Fig. 2:

Reciprocal h2l plane of (a) a large and (b) a tiny plate of polytype II of 1 reconstructed from CCD data.

More details on data collection and refinement are compiled in Table 1.

Powder diffraction

High temperature X-ray powder diffraction (XRPD) experiments were performed on a Panalytical X’Pert Pro diffractometer equipped with an Anton Paar HTK-1200N high temperature chamber in Bragg-Brentano geometry using CuKα_1,2 radiation (λ=1.540598, 1.544426 Å) with Ni filter and an X’celerator multi-channel detector. The ground bulk sample was placed on a Si single crystal cut along the (711) plane. Scans were recorded under helium atmosphere in the 2θ=10−70° range in 5 K steps from 30 to 140°C and back to 30°C with heating and cooling rates of 1 K/min between scans.

Calorimetry

Differential scanning calorimetry (DSC) measurements were recorded on a NETZSCH DSC 200 F³ with a heating rate of 10 K under argon atmosphere. Two heating and cooling cycles in the 30–135°C range were performed.

Polytypism

The polytypes I and II of 1 are made up of non-polar (with respect to the stacking direction) layers of the same kind (Figure 3), which will be designated as A_n, where n is a sequential number. The A_n layers possess (in polytype I actual; in polytype II approximate) P(n)2b symmetry. In the tradition of the OD literature [7], parentheses in the layer group symbol indicate the direction lacking translational symmetry.

Fig. 3:

The structure of (a) polytype I and (b) polytype II of 1 viewed down [010]. C, N and Cl atoms are represented by gray, blue and green ellipsoids drawn at the 50% probability levels. H atoms were omitted for clarity. Layer names are indicated to the right.

In both polytypes every pair of adjacent layers is geometrically equivalent to every other pair of adjacent layers in the same polytype. Thus, both polytypes fulfill the vicinity condition (VC) of OD theory [2]. There are two kinds of structures fulfilling the VC. In proper OD structures there is an infinity of possibilities of arranging the layers without violating the VC. These polytypes form an OD family of structures. In fully ordered structures, on the other hand, there is only one way of achieving the VC. One can say that the OD family of a fully ordered structure is made up of only one member.

Even though both polytypes fulfill the VC, pairs of adjacent layers in polytype I are not equivalent to pairs of adjacent layers in polytype II. Therefore, the polytypism relating polytypes I and II is of the non-OD kind, i.e. both polytypes belong to different OD families.

To describe the symmetry of polytypes, the OD theory introduced the concept of partial operations (PO). A PO is a restriction of a motion of Euclidean space E³ to the space occupied by a distinct layer. A PO is characterized by a motion of E³, a source and a target layer. POs can be composed only if the source layer of the second is the target layer of the first. Thus, the composition of all POs of a polytype (or any modular structure) forms a groupoid [8]. In OD theory, POs that map a layer onto itself are called λ-POs and POs that map different layers σ-POs.

The core of the OD theory is a formalism to classify the symmetry of families of OD polytypes, viz. OD groupoid families [2]. OD groupoid families are the analogs of space group types. Whereas space group types abstract from metrics, OD groupoid families additionally abstract from the different stacking possibilities in (proper) OD families. In both polytypes, adjacent A_n layers are related, among other operations, by a glide reflection with a plane parallel to (100). Therefore, the groupoids of polytype I and II belong to the same OD groupoid family with the symbol

P(c)21b{(nr,s)2r + 1n2,r + 1}

according to the notation of [2], even though the polytypes belong to different OD families. An analogous situation would be different polymorphs that possess the same space group type.

The first line of the OD groupoid family symbol indicates the layer group (λ-POs), the second line the operations relating adjacent layers (σ-POs). Here, n_r,s designates a glide reflection with the intrinsic translation vector (rb+sc)/2, 2_r+1 a screw rotation with the translation vector (r+1)b/2 and n_2,r+1 a glide reflection with the translation vector a₀+(r+1)b/2. a₀ is the vector perpendicular to (100) with the length of one layer width. Because a₀ differs in polytypes I and II it will be written as a0I or a0II, if necessary.

Polytype I

In polytype I, the metric parameters of the OD groupoid family are (r, s)=(1, 0). This can be expressed by the symbol

P(c)21b{(b)2a2}.

The NFZ relationship [9], a further central concept of OD theory, is used to determine in how many ways layers can be arranged in such a way that pairs of adjacent layers are equivalent. It is based on the groups Γ_n of those operations of A_n that do not reverse the orientation with respect to the stacking direction. POs that maintain the orientation are called τ-POs and therefore Γ_n can also be considered as the group of λ-τ-POs of the A_n layer. The subgroup Γ_n∩Γ_n+1<Γ_n are those λ-τ-POs of A_n that also apply to A_n+1. Here, because the b-glide planes of adjacent layers overlap, Γ_n=Γ_n∩Γ_n+1=P(1)1b. The NFZ relationship reads as Z=N/F=[G_n:G_n∩Γ_n+1]=[P(1)1b:P(1)1b]=1 and therefore given an A_n layer, there is only Z=1 way of placing the A_n+1 layer. Polytype I is fully ordered with the symmetry Ic2b and a=2a0I, as schematized in Figure 4a. Any other arrangement of A_n layers would result in a locally different polytype and thus violate the VC.

Fig. 4:

Schematic representation of the symmetry of (a) the polytype I and (b,c) the MDO polytypes of the OD family of polytype II. 1 molecules are represented by triangles. A gray color indicates translation by b/2. λ-POs of layers and σ-POs relating adjacent layers are indicated using the usual graphical symbols [10]. Operations not valid for the whole polytype are drawn in gray. The unit cells of the polytypes are indicated by red polygons.

Polytype II

In polytype II, r=0, but s∉ℤ, which can be expressed by the symbol

P(c)21b{(cs)21n2,1}.

Here, the b-glide reflections of adjacent layers do not overlap and therefore Z=N/F=[P(1)1b:P(1)11]=2. Thus, given an A_n layer, the adjacent A_n+1 layer can be placed in Z=2 ways. These two positions of A_n+1 are related by the b-glide reflection of A_n. Polytype II is therefore a proper OD structure.

The A_n layers can be arranged to an infinity of polytypes, in which pairs of adjacent layers are geometrically equivalent and which belong to the same OD family as polytype II. If interactions over more than one layer width are neglected, these polytypes are also energetically equivalent.

Two of the OD polytypes are of a maximum degree of order (MDO) [11], because they cannot be decomposed into fragments of simpler polytypes of the same family. MDO₁ [a=a0II+(s−1)c/2,P2₁, Figure 4b] is generated by repeated application of c_s-glide reflections. MDO₂ [a=2a0II,Pc2₁n, Figure 4c] is generated by alternating application of c_s and c_−s -glide reflections. All other polytypes of the OD family can be decomposed into fragments of MDO₁ and MDO₂. Polytype II is the MDO₁ polytype of the OD family. The metric parameter s can be deduced from the cell parameters of MDO₁ as s=a cos β/c+1=0.771.

Polytype family

As proven by the existence of polytypes I and II, given a layer A_n, there are two kinds of placing the adjacent A_n+1 layer, resulting in non-equivalent A_nA_n+1 pairs. These layer contacts will henceforth be designated as of the I (A_n mapped onto A_n+1 by a b-glide reflection) and II (c_s- or c_−s -glide reflection) type. Thus, polytype I and II are member of a large polytype family which can be be classified into three subfamilies (Figure 5):

Fig. 5:

The polytype family of 1 and its subfamilies. Arrows indicate a set/subset relationship.

1. Only I contacts are realized. This subfamily contains only the polytype I, which fulfills the VC.

2. Only II contacts are realized. This subfamily is a family of proper OD structures, which fulfill the VC.

3. I and II-types of contacts are realized. These polytypes do not fulfill the VC.

Even though the concept of MDO was originally only defined for OD families [11], it is just as compelling in the case of non-OD polytype families. With respect to the whole polytype family, polytype I, MDO₁ and MDO₂ (see above) are the MDO polytypes. A polytype containing layer contacts of both kinds violates the MDO condition [11]. The MDO polytypes are made up of only a subset of kinds of pairs of adjacent layers. Thus, it is again demonstrated that, by a large margin, most observed ordered polytypes are of the MDO type.

Symmetry reduction and desymmetrization

A characteristic phenomenon of ordered polytypes of proper OD families is a reduction of the symmetry of the actual layers compared to the idealized (prototype) layers [12].

Indeed, in the polytype II, the symmetry of the A_n layers is related to the symmetry of the prototype layers by symmetry reduction of index 2 from Pc2₁(b) to P12₁(1). The position of the 1 molecule is split in two (Z′=2). The two different molecules will be called II and II′.

In the fully ordered polytype I, on the other hand, the A_n layers retain the Pc2₁(b) symmetry and there is only one independent molecule (Z′=1), which will be called molecule I.

The desymmetrization can therefore be quantified in two aspects: firstly, the geometric deviation of the layers in polytype I compared to those of polytype II and secondly the deviation of the layers in polytype II from the prototype P(c)2₁a symmetry.

The first point is reflected in a variation of the cell parameters of the layer lattice. The b parameter in polytype II is slightly enlarged in comparison to polytype I. The effect on the surface of the fundamental parallelogram of the layer lattice is compensated by a smaller c [bc=7.6200·14.7552=112.43 Ų vs. bc=7.7720· 14.4752=112.50 Ų].

A more detailed quantification was performed by moving the origins of both structure models to (0, 0, 14)T (polytype I) and (14, 0, 12)T (polytype II), transforming the coordinates into an orthonormal coordinate system and applying the b_[001]-glide reflection of the A_n layer to the molecule II′. Thus, an overlap of three molecules was obtained (Figure 6). The distances between the corresponding atoms in the three molecules are compiled in Table 2.

Fig. 6:

Overlap of the molecule in polytype I (red) with both molecules in polytype II (blue: molecule II and green: molecule II′, glide reflected by b_[001]).

Intuitively, because the A_n layers in polytype I are in a geometrically different environment than the layers in polytype II, one might expect that the deviation between polytypes I and II is larger than between the two molecules in polytype II. Nevertheless, the opposite behavior is observed. The molecule in polytype I adopts an intermediate position of the molecules in polytype II. Overall, desymmetrization is small with a maximum of 0.283 Å for the C16 atom of the pyridine ring.

Twinning

The point group of the OD groupoid family (the group generated by the linear parts of all POs of a member [13]) of both polytypes is m2m. The fully ordered polytype I retains this point symmetry. In polytype II the point symmetry is reduced to 2. Since [m2m:2]=2, one can expect this polytype to form OD twins [2] with two orientation states. The twin law of such a twin is obtained by coset decomposition of 2 in m2m. It is made up of m₍₁₀₀₎ and m_[001] operations, which correspond to the linear parts of the c₍₁₀₀₎ and b_[001] of the A_n layers that are lost in MDO₂. In the MDO₂ polytype, the m2m point symmetry of the OD groupoid family is retained. A stacking fault in such a polytype does not lead to twinning, but antiphase domains [14].

The observed twin individuals of the polytype II crystal are indeed the two orientations of the MDO₁ polytype of the OD-family. The twins can therefore be considered as classical OD twins [2]. It has to be noted that, even though unlikely, at the twin interface could also be located a fragment of the orthorhombic polytype I. Such a twin would not classify as an OD twin, because the fragment at the composition plane is not a fragment of any polytype of the same OD family.

The structures of both polytypes are merohedral [point groups m2m (polytype I) and 2 (polytype II)]. Nevertheless, owing to the resonant scatterer Cl, additional twinning by inversion could be ruled out. The crystal of the fully ordered polytype I [Flack parameter [15] −0.03(5)] was not twinned. The crystal of polytype II [Flack parameter −0.05(5)] was made up of only two domains related by reflection at (100), but not of domains related to the first by inversion or twofold rotation about a^*. In summary, the observed twinning behavior is in perfect agreement with the OD interpretation of the polytypes.

Crystal chemistry

A comparison of the non-equivalent layer contacts of both polytypes is difficult owing to a lack of structure directing hydrogen or halogen bonds. Using intermolecular distances can be treacherous, because they depend on arbitrary thresholds. One method to objectify such a description is the analysis of Hirshfeld surfaces [16]. Notably, in d_i/d_e fingerprint plots [17] short intermolecular contacts can be identified and compared visually. In Figure 7 such plots are given for the molecules in both polytypes.

Fig. 7:

d_i/d_e fingerprint plots of the Hirshfeld surfaces of the molecules in (a) polytype I and (b,c) polytype II, calculated with CrystalExplorer [18]. Regions of the histograms where H···H, C···H and N···H contacts dominate are drawn in green, blue and yellow, respectively. Other kinds of contacts are drawn in gray. A brighter color indicates a larger fraction of the Hirshfeld surface. For comparison, the plots of the polytype II molecules are overlaid with the outline of the fingerprint plot of the other molecule in the same polytype; the plot of polytype I with the outline of both polytype II molecules.

A characteristic feature in the plots of all three molecules are the spikes of short C–H···C interactions marked with an A letter. They are very similar for all three molecules, because they are intralayer contacts, which are, as demanded by polytypism, geometrically equivalent. These are a short C5–H···C7 carbazole to carbazole contact [polytype I: 2.696 Å; polytype II: 2.661 Å and 2.646 Å] and a slightly longer C16–H···C2 pyridine to carbazole contact [polytype I: 2.770 Å; polytype II: 2.774 Å and 2.927 Å]. The latter is distinctly longer for the II′ molecules, owing to desymmetrization. Both contacts are of the H···π kind and connect molecules to rods extending along [010] (Figure 8).

Fig. 8:

Rod of 1 molecules connected by short C–H···π contacts in the polytype I viewed down [100]. Atoms are represented by spheres of arbitrary radius. Color codes as in Figure 3, H atoms are white. Short intermolecular contacts marked by A in Figure 7 are indicated by dotted lines. The corresponding rods in polytype II are virtually equivalent.

The short H···N and H···H contacts result in markedly different d_i/d_e fingerprints (yellow and green in Figure 7), because they are of the interlayer kind. Accordingly, all three molecules are in different environments concerning these interactions. The most notable short interlayer contact in the polytype I is C15–H···H–C15 [Region B, pyridine to pyridine, 2.468 Å]. This region is located on the d_i=d_e line, because the contact is between H atoms related by symmetry. The pronounced interactions in the polytype II are C3′–H···N2 [Region C, carbazole to pyridine, 2.635 Å], C3–H···H–C2 and C4–H···H–C2 [Region D, carbazole to carbazole, 2.476 and 2.475 Å]. Region C is a contact between the molecules II and II′ and therefore it appears in both plots, but reflected at the d_i=d_e line. Region D connects two II molecules and therefore is only present in one plot, where it is symmetric by reflection at d_i=d_e. All the discussed close interactions are shown in the actual structures in Figure 9.

Fig. 9:

Short C–H···N and C–H···H–C interlayer contacts in (a, b) polytype I and (c, d) polytype II of 1, indicated by dotted lines and marked with the letter of Figure 7. The structures are projected (a, c) along [010] and (b, d) on the layer plane (100). For clarity, in (b, d) only the contact plane is shown as indicated by a dashed outline in (a, c). Molecular fragments of the top and bottom layers are drawn in red and black, respectively.

Often it is assumed that tighter packed polymorphs are energetically favored, because the van-der-Waals interaction is maximized. Whereas the density of the layers are virtually identical (see above), a more pronounced difference is observed for the packing density along the stacking direction, which can be expressed by the length of the a₀ vector [polytype I: |a0I|=a/2=12.108 Å; polytype II: |a0II|=asinβ=12.151 Å]. Thus, the polytype I is slightly more tightly packed [1.360 vs. 1.354 g/cm³].

Bulk sample

The bulk sample was inspected under the microscope and only crystals with the characteristic platy habit of the polytype II were observed. The assignment was confirmed by short single-crystal diffraction experiments of a few selected plates.

To detect a potential II→I phase transition, differential scanning calorimetry (DSC) scans were performed on the bulk sample. The only significant effect on heating is the endothermic melting (Figure 10, onset 126.7°C). The corresponding exothermic crystallization peak is broad and features a strong hysteresis (onset 70.5°C). In a second cycle, the melting point is slightly lower (125.5°C), indicative of minor decomposition or imperfect crystallization. As in the first cycle, no other significant effects are observed, suggesting that the polytype II is obtained on crystallization. The sample did not crystallize during the second cooling cycle.

Fig. 10:

DSC trace of the bulk sample of 1 in the 30→135°C range. The heating curves are at the top. Endothermic effects are up.

As an additional confirmation of the identity of the bulk sample and the lack of a phase transition, high-temperature X-ray powder diffraction (XRPD) experiments were performed in the 30→140°C range (Figure 11). Nearly all peaks can be attributed to the polytype II with a very strong preferred orientation of the h00 reflections (corresponding to the stacking direction).

Fig. 11:

XRPD scans of 1 at T=30, 120, 125°C and at room temperature after crystallization of the melt. (a) Full view to emphasize the strong texture and (b) magnification of the ordinate to show the fine structure. Scans are translated along the ordinate for clarity. The peak form of the intense peaks is due to the employed Ni filter. Peaks in the T=30°C scan that could not be attributed to polytype II are marked with an asterisk.

Two small peaks at 2θ=11.84, 12.24° belong to an impurity and gradually vanish on heating (Figure 11). The peak at 12.24° might correspond to the 200 reflection of polytype I, but the peak at 11.84° cannot be attributed to a polytype I reflection. It is therefore unlikely that the impurity is polytype I.

At 125°C the sample was molten. As already observed in the DSC measurements, the sample did not readily crystallize on cooling. In an XRPD scan of the solidified sample at room temperature only reflections of the polytype II could be evidenced, with virtually only the h00 reflections visible.

In summary, the bulk sample, crystallized from EtOH or from melt, was composed almost exclusively of polytype II, and a transformation II→I could not be evidenced. A way to crystallize polytype I remains unknown.