Structural Variations and Polymorphism of Some ... - ACS Publications

CRYSTAL GROWTH & DESIGN

Structural Variations and Polymorphism of Some Derivatives of 6-Amino-2-phenylsulfonylimino1,2-dihydropyridine

2004 VOL. 4, NO. 4 701-709

Michael T. Kirchner,† L. Sreenivas Reddy,† Gautam R. Desiraju,*,† Ram K. R. Jetti,‡ and Roland Boese*,‡ School of Chemistry, University of Hyderabad, Hyderabad 500 046, India, and Institut fu¨ r Anorganische Chemie, Universita¨ t Duisburg-Essen, Standort Essen, Universita¨ tsstrasse 5-7, D-45177 Essen, Germany Received March 9, 2004

ABSTRACT: Following up from a study of polymorphism in the title compound, 6-amino-2-phenylsulfonylimino1,2-dihydropyridine, 12 phenyl-substituted derivatives were examined for polymorphism by recrystallization from a large number of solvents. The title compound contains hydrogen bond donor (D) and acceptor sites (A) located in a AADD juxtapositioning. Two structural families that differ in their use of the AADD hydrogen bond functionality may be identified. Methyl and chloro substitution in the ortho position or fluoro substitution in the para position leads to a catemer motif, which is related to the kinetic form of the title molecule. Meta Substitution by methyl and larger substituents in the para position block this structural option. While meta-methyl and para-chloro substitution lead to unique structures, two polymorphs of the para-methyl-substituted derivative could be crystallized and these adopt the dimer arrangement. With the larger bromo, iodo, methoxy, and trifluoromethyl substituents in the para position, dimers arranged into interconnected layers are obtained. These dimer structures are reminiscent of the thermodynamic form of the title compound. It is noteworthy that the majority of derivatives of the title compound fail to show polymorphic behavior, and this shows that our understanding of polymorphism is still far from complete. Introduction

Scheme 1

6-Amino-2-phenylsulfonylimino-1,2-dihydropyridine, 1a (Scheme 1), is an interesting molecule that exhibits polymorphism.1 This molecule was supplied as one of the “blind” test molecules for crystal structure prediction (CSP) in 2001,2 and our previous publication delineated several interesting features of the structural chemistry of this compound.1 Notably, it had been felt during the blind test that CSP of this molecule would be very difficult because of its seeming propensity for polymorphism.3 We showed, however, that CSP was problematic not because polymorphism was endemic (we could isolate a second form only with much difficulty) but because the commonly occurring form (answer in blind test) is kinetically favored.1 We recollect here that methods of CSP based on structure minimization with atom potentials tend to generate the thermodynamic form.2,4 Sulfonimide 1a is noteworthy not only in its two hydrogen bond donor and two acceptor sites but also in the arrangement of these four sites in an acceptoracceptor-donor-donor (AADD) fashion on one of the edges of this L-shaped molecule (Scheme 2). The pattern of N-H‚‚‚O and N-H‚‚‚N hydrogen bonds formed at these sites constitutes the basis for polymorphism of this molecule.1 The kinetic form (I) has a catemer two-point supramolecular synthon (one-dimensional hydrogen bond pattern) while the thermodynamic form (II) has a four-point dimer synthon (zero-dimensional hydrogen bond pattern). The operation of factors that alter the

balance between kinetic and thermodynamic crystals is a key question in the control of polymorphism.5 This also raises the question as to how polymorphism is influenced by small changes, e.g., in the molecular structure. Accordingly, we present here the structural chemistry of several derivatives of molecule 1a (Scheme 1), systematically substituting the H-atoms on the ortho, meta, and para positions of the phenyl ring. In addition, it is interesting to note that 1a and its derivatives contain a rare example of a quadruple hydrogen bonding unit and this might be of value for supramolecular synthesis.6

* To whom correspondence should be addressed. G.R.D. E-mail: [email protected]. R.B E-mail: [email protected]. † University of Hyderabad. ‡ Universita ¨ t Duisburg-Essen.

Experimental Section The synthesis and crystallography of the title compound were reported earlier.1

10.1021/cg049912t CCC: $27.50 © 2004 American Chemical Society Published on Web 06/18/2004

702

Crystal Growth & Design, Vol. 4, No. 4, 2004

Kirchner et al.

Scheme 2. Kinetic (Left) and Thermodynamic (Right) Polymorphs of the Title Moleculea

a

A ) acceptor; D ) donor of hydrogen bonds.

Synthesis. All compounds discussed here were synthesized according to the literature procedure given for 2-(aminobenzenesulfonylamino)-6-aminopyridine by taking the corresponding sulfonyl chloride with 2,6-diaminopyridine.7 All compounds were characterized by IR and NMR. 1H NMR spectra were recorded at 200 MHz on a Bruker ACF instrument, and IR spectra were recorded on a Jasco 5300 spectrophotometer. Melting points were recorded on a Fisher-John apparatus. 6-Amino-2-phenylsulfonylimino-1,2-dihydropyridine (1a). Yield 72%. IR (cm-1): 3373, 3327, 3200, 1371, 1130. 1H NMR (200 MHz, dimethyl sulfoxide (DMSO)-d6, δ ppm): 5.91 (d, 1H), 6.15 (d, 1H), 6.15 (b, NH2), 7.27 (t, 1H), 7.49 (m, 3H), 7.82 (d, 2H), 11.24 (b, NH). mp 235 °C. 6-Amino-2-(2-chlorophenylsulfonylimino)-1,2-dihydropyridine (1b). Yield 60%. IR (cm-1): 3365, 3329, 3204, 1365, 1134. 1H NMR (200 MHz, DMSO-d6, δ ppm): 5.85 (d, 1H), 6.10 (d, 1H), 6.42 (b, NH2), 7.27 (t, 1H), 7.49 (m, 3H), 7.82 (d, 1H), 11.86 (b, NH). mp 222-225 °C. 6-Amino-2-(2-methylphenylsulfonylimino)-1,2-dihydropyridine (1c). Yield 65%. IR (cm-1): 3357, 3337, 1373, 1114. 1H NMR (200 MHz, DMSO-d6, δ ppm): 2.59 (s, 3H), 5.86 (d, 1H), 6.10 (d, 1H), 6.23 (b, NH2), 7.22 (t, 1H), 7.40 (m, 3H), 7.94 (d, 1H), 11.20 (b, NH). mp 210-211 °C. 6-Amino-2-(3-chlorophenylsulfonylimino)-1,2-dihydropyridine (2a). Yield 56%. IR (cm-1): 3355, 3187, 1352, 1123. 1 H NMR (200 MHz, DMSO-d6, δ ppm): 5.87 (d, 1H), 6.10 (d, 1H), 6.22 (b, NH2), 7.23 (t, 1H), 7.35 (t, 1H), 7.42 (d, 1H), 7.92 (d, 1H), 8.32 (s, 1H) 11.72 (b, NH). mp 225-227 °C. 6-Amino-2-(3-methylphenylsulfonylimino)-1,2-dihydropyridine (2b). Yield 73%. IR (cm-1): 3375, 3335, 3200, 1362, 1134. 1H NMR (200 MHz, DMSO-d6, δ ppm): 2.35 (s, 3H), 5.89 (d, 1H), 6.16 (d, 1H), 6.16 (b, NH2), 7.23 (t, 1H), 7.30 (t, 1H), 7.42 (d, 1H), 7.65 (s, 1H), 7.67 (s, 1H), 11.14 (b, NH). mp 235-236 °C. 6-Amino-2-(4-fluorophenylsulfonylimino)-1,2-dihydropyridine (3a). Yield 79%. IR (cm-1): 3305, 3292, 3223, 1340, 1128. 1H NMR (200 MHz, DMSO-d6, δ ppm): 5.89 (d, 1H), 6.18 (d, 1H), 6.28 (b, NH2), 7.29 (t, 1H), 7.35 (d, 2H), 7.94 (d, 2H), 11.40 (b, NH). mp 182-185 °C. 6-Amino-2-(4-chlorophenylsulfonylimino)-1,2-dihydropyridine (3b). Yield 58%. IR (cm-1): 3393, 3203, 1375, 1134. 1 H NMR (200 MHz, DMSO-d6, δ ppm): 5.89 (d, 1H), 6.16 (d, 1H), 6.34 (b, NH2), 7.27 (t, 1H), 7.35 (d, 2H), 7.84 (d, 2H), 11.52 (b, NH). mp 192 °C. 6-Amino-2-(4-bromophenylsulfonylimino)-1,2-dihydropyridine (3c). Yield 62%. IR (cm-1): 3423, 3327, 3275, 1369, 1126. 1H NMR (200 MHz, DMSO-d6, δ ppm): 5.89 (d, 1H), 6.13 (d, 1H), 6.31 (b, NH2), 7.27 (t, 1H), 7.67 (d, 2H), 7.78 (d, 2H), 11.54 (b, NH). mp 215 °C. 6-Amino-2-(4-iodophenylsulfonylimino)-1,2-dihydropyridine (3d). Yield 60%. IR (cm-1): 3435, 3362, 1350, 1125.

Figure 1. Top: a catemer chain of the para-fluoro derivative 3a along [010]. Bottom: view down [001h ] showing two catemer chains with a herringbone pattern between phenyl rings. The F-atom points toward an S-atom of an adjacent chain (F‚‚‚S 3.222 Å; C-F‚‚‚S, 173°). 1

H NMR (200 MHz, DMSO-d6, δ ppm): 5.89 (d, 1H), 6.12 (d, 1H), 6.32 (b, NH2), 7.26 (t, 1H), 7.60 (d, 2H), 7.89 (d, 2H), 11.40 (b, NH). mp 240-242 °C. 6-Amino-2-(4-methylphenylsulfonylimino)-1,2-dihydropyridine (3e). Yield 79%. IR (cm-1): 3377, 3198, 1371, 1132. 1H NMR (200 MHz, DMSO-d6, δ ppm): 2.34 (s, 3H), 5.92 (d, 1H), 6.12 (d, 1H), 6.17 (b, NH2), 7.25 (t, 1H), 7.33 (d, 2H), 7.77 (d, 2H), 11.40 (b, NH). mp 205-207 °C. 6-Amino-2-[4-(trifluoromethyl)phenylsulfonylimino]1,2-dihydropyridine (3f). Yield 79%. IR (cm-1): 3387, 3318, 3214, 1365, 1138. 1H NMR (200 MHz, DMSO-d6, δ ppm): 5.89 (d, 1H), 6.16 (d, 1H), 6.40 (b, NH2), 7.30 (t, 1H), 7.87 (d, 2H), 8.02 (d, 2H), 11.72 (b, NH). mp 220-223 °C. 6-Amino-2-(4-methoxyphenylsulfonylimino)-1,2-dihydropyridine (3g). Yield 68%. IR (cm-1): 3467, 3372, 1356, 1130. 1H NMR (200 MHz, DMSO-d6, δ ppm): 3.70 (s, 3H), 5.73 (d, 1H), 6.15 (d, 1H), 5.95 (b, NH2), 6.90 (d, 2H), 7.24 (t, 1H), 7.79 (d, 2H), 10.84 (b, NH). mp 160 °C. 6-Amino-2-(4-nitrophenylsulfonylimino)-1,2-dihydropyridine (3h). Yield 80%. IR (cm-1): 3355, 3282, 3243, 1480, 1345, 1132. 1H NMR (200 MHz, DMSO-d6, δ ppm): 5.90 (d, 1H), 6.16 (d, 1H), 6.50 (b, NH2), 7.36 (t, 1H), 8.02 (d, 2H), 8.30 (d, 2H), 10.00 (b, NH). mp 213-215 °C. Recrystallization Experiments. Having possible polymorphism in mind, crystals were grown from acetone, acetonitrile, CCl4, chloroform, dichloromethane, dimethylform-

6-Amino-2-phenylsulfonylimino-1,2-dihydropyridine

Crystal Growth & Design, Vol. 4, No. 4, 2004 703

Table 1. Crystallographic Data and Structure Refinement Parameters empirical formula formula wt crystal system space group T (K)a a (Å) b (Å) c (Å) R (°) β (°) γ (°) Z V (Å3) Dcalcd (g cm-3) reflns collected unique reflns observed reflns R1[I > 2σ(I)] wR2 (all) goodness-of-fit structure type

1a (form I)

1a (form II)

1b

1c

2b

3a

3b

C11H11N3O2S 249.30 monoclinic


C11H10ClN3O2S 283.73 monoclinic



C11H10FN3O2S 267.28 monoclinic

C11H10ClN3O2S 283.73 monoclinic

P21/c 293 8.5450 (13) 9.0638 (17) 15.151 (2) 90 91.492 (12) 90 4 1173.0 (3) 1.412

P21/c 203 12.1099 (15) 10.7924 (12) 17.464 (2) 90 97.318 (2) 90 8 2263.9 (5) 1.463

P21/c 298 9.232 (6) 8.900 (6) 15.102 (9) 90 96.081 (11) 90 4 1233.8 (13) 1.527

P21/c 298 9.169 (2) 8.867 (2) 15.222 (4) 90 96.511 (5) 90 4 1229.5 (5) 1.422

P21/c 223 12.434 (4) 10.396 (4) 9.113 (3) 90 92.247 (6) 90 4 1177.2 (7) 1.486

P21/c 203 9.000 (6) 8.618 (6) 14.643 (10) 90 91.698 (17) 90 4 1135.2 (13) 1.564

C2/c 203 13.725 (6) 7.312 (3) 25.419 (11) 90 92.886 (8) 90 8 2547.8 (19) 1.479

3526

14219

10435

10199

14514

5392

15199

3403

5443

3061

2909

2916

1385

3167

2687

2871

1881

2135

1971

804

2065

0.0406 0.1117 1.02 catemer

0.0579 0.1525 0.96 dimer

0.0574 0.1553 0.97 catemer

0.0558 0.1451 1.04 catemer

0.0596 0.1563 1.04 intermediate

0.1082 0.3162 1.02 catemer

0.0700 0.1802 1.06 intermediate

3c empirical formula formula wt crystal system space group T (K)a a (Å) b (Å) c (Å) R (°) β (°) γ (°) Z V (Å3) Dcalcd (g cm-3) reflns collected unique reflns observed reflns R1[I > 2σ(I)] wR2 (all) goodness-of-fit structure type a

3d

3e (form I)

3e (form II)

3f

3g

3h

C11H10BrN3O2S C11H10IN3O2S C12H13N3O2S C12H13N3O2S C12H10F3N3O2S C12H13N3O3S (C11H10N4O4S)‚(H2O) 328.19 375.18 263.31 263.31 317.29 279.31 312.31 monoclinic orthorhombic monoclinic monoclinic orthorhombic orthorhombic monoclinic P21/c 203 10.7941 (9) 18.2559 (15) 26.335 (2) 90 94.330 (2) 90 16 5175.2 (7) 1.685

Pbca 203 10.755 (2) 18.300 (3) 26.828 (5) 90 90 90 16 5280.1 (17) 1.888

P21/n 253 15.015 (6) 10.880 (5) 17.066 (7) 90 113.457 (8) 90 8 2557.6 (19) 1.368

P21/c 203 10.384 (3) 17.749 (4) 26.473 (7) 90 94.531 (7) 90 16 5074 (2) 1.379

Pbca 203 10.8897 (13) 18.343 (2) 26.640 (3) 90 90 90 16 5321.3 (11) 1.584

Pbca 298 10.614 (3) 18.276 (5) 27.402 (7) 90 90 90 16 5315 (2) 1.396

P21/c 293 14.663 (16) 12.276 (16) 7.589 (11) 90 96.16 (9) 90 4 1358 (3) 1.527

65572

39788

32477

19916

51767

48710

1925

12882

3444

6436

6524

6621

4660

1763

7255

2482

3422

2974

4550

3997

1241

0.0656 0.1848 1.04 dimer

0.0433 0.1245 1.08 dimer

0.0565 0.1688 1.00 dimer

0.0737 0.2601 0.90 dimer

0.0589 0.1481 1.05 dimer

0.1024 0.2560 1.12 dimer

0.0781 0.2323 1.05 intermediate

Temperature of data collection.

amide, DMSO, dioxane, ethanol, ethyl acetate, methanol, nitromethane, and tetrahydrofuran for all derivatives. For 3ae, additionally, benzene, n-butanol, ether, and toluene were tried. Up to five crystals from each batch, and around 250 crystals in all, were checked for their cell parameters on the diffractometer. Only for compound 3e were two polymorphs found. Crystallization from nitromethane, which successfully yielded the thermodynamic form for 1a, was not successful in generating polymorphs of derivatives 1c, 2b, and 3a-e,g,h.8 Compound 2a did not yield crystals, and 3h crystallized as a hydrate. X-ray Data Collection and Crystal Structure Determinations. X-ray data for 3h were collected on a Bruker P4 diffractometer, while those for 1b,c, 2b, and 3a-g were collected on a SMART diffractometer using Mo KR radiation. The structure solution and refinement were carried out using SHELXL programs built in with the SHELXTL (Version 6.12) package.9 The positions of the H-atoms bound to phenyl groups in 3a-h and amino groups in 3a,c,d, 3e form II, and 3f were generated by a riding model on idealized geometries with Uiso-

(H) ) 1.2Ueq(C) or 1.2Ueq(N), while the H-atoms of the amino groups in 1b,c, 2b, 3b, 3e form I, 3g,h, and methyl groups in 3e form I were located in difference Fourier maps and these H-atoms were also refined as riding with Uiso(H) ) 1.2Ueq(N) or 1.5Ueq(C). The hydrogen atoms of the water molecule in 3h were also taken from Fourier maps and also refined as riding with Uiso(H) ) 1.5Ueq(O). The F-atoms of the CF3 group in 3f were disordered over two sites with occupancies of 0.5 each. In some cases, the U values seem to be a bit too large or too small and this is because of poor crystal quality, especially in 3a. The details of the X-ray data collection, structure solution, and refinement are given in the Supporting Information.

Results and Discussion The packing of the parent molecule 1a and its derivatives is dominated by an AADD hydrogen bond functionality with the sulfonylimino O-atom and N-atom acting as acceptors, respectively, to the hydropyridine

704


and amino donors (Scheme 2).6,10 The second H-atom of the amino group and the second O-atom of the sulfonyl group provide a third strong donor-acceptor pair, which flanks the AADD fragment on both sides of the extended hydropyridine system. The three distinct hydrogen bonds so formed (N-H‚‚‚O, N-H‚‚‚N, and N-H‚‚‚O) are present in all 14 structures in this study. Variations in the arrangement of the molecules around these fixtures permit a classification of 11 crystal structures into two distinct families, termed here catemer and dimer, while the three others can possibly be visualized as intermediates between the families. Catemer Family. Form I of the parent molecule 1a heads the first family followed by 1b,c and 3a. All of these crystals are characterized by a catemer arrangement. The chain is of the type [‚‚‚AADD‚‚‚AADD‚‚‚ AADD] and is shown in Figure 1, top, for derivative 3a. Along the chain, additional strong hydrogen bonds from the second amino H-atom to the second O-atom of the sulfonyl group are present. As described previously, the catemer along a screw axis typifies the kinetic product of crystallization in this system being a one-dimensional rather than a zero-dimensional motif. Substitution of the ortho-H-atom in the parent molecule 1a with a Cl-atom or a CH3 or the para-H-atom with an F-atom leads to basically the same structure in 1b,c and 3a as demonstrated by the comparable cell parameters all within space group P21/c (see Table 1). The a-axis of 3a, along which the C-F bond is aligned, is ∼0.45 Å longer than in form I of 1a. This increase of the a-axis is because of the steric constraints of fitting the F-atom into the previously occupied position of the H-atom. This is a close fit, as indicated by a short intermolecular F‚‚‚S distance D ) 3.222 Å, which is a short, linear fluorine‚‚‚sulfonyl contact below the sum of the van der Waals radii (Figure 1, bottom).11 In compounds 1b,c, the conformations along the S-C bond are eclipsed with the substituted ortho-C-atom being gauche (θ ) -53.34° for 1c and θ ) -55.56° for 1b). This is distinct from the synclinal conformations in form I of 1a (θ ) -88.28°) and in 3a (θ ) -92.16°). Because of this conformational difference, the phenyl substituents in 1b,c are perpendicular to the direction of the chain and do not form a herringbone motif with the neighboring chain as in 1a and 3a; rather, they are π-stacked (Figure 2). The para-H-atom of the pyridine ring, which is important for the double layer structures (see later), is not active within this family. In summary, the packing of the catemer family can accommodate small substituents (like F) in the para position and larger substituents (like Cl and Me) in the ortho position, especially if the herringbone motif is avoided for the phenyl rings. Both of the meta C-H groups are used for C-H‚‚‚O interactions in form I of 1a and hence are not free for isostere substitution, i.e., with retention of this particular packing. This becomes relevant in compound 2b. Intermediate Structures 2b and 3b,h. In view of the steric crowding in 3a, it is not surprising that the corresponding chloro analogue, 3b, adopts a different and distinct packing. The space group is C2/c with Z′ ) 1, and the b-axis is the shortest observed in the group (7.3122 Å). Molecules related by a 2-fold axis form dimers of the type [AADD‚‚‚DDAA]. This is in contrast

Kirchner et al.

Figure 2. Compounds 1b (top) and 1c (bottom), view down [001 h ], showing two catemer chains. The conformation down the C-S bond is different from 3a (see Figure 1, bottom), changing the packing of phenyl rings from a herringbone to a π-stacking motif and allowing the ortho substituents to be directly incorporated into the phenyl ring packing.

to the dimer family described later in this paper in which symmetry-independent molecules form dimers. The third hydrogen bond, of the N-H‚‚‚O type (D ) 2.962 Å), is between stacked dimers threading them into chains running along [110] and [11h 0] (Figure 3). Eventually, a layer in (001) is formed containing all of the dimers and with the chlorophenyl substituents sticking out in the [001] and [001 h ] directions. Between layers these chlorophenyl substituents are π-stacked. Two C-H‚‚‚O interactions are formed by the metaH-atoms, a short one (d ) 2.241 Å, 149°) within (001) and a longer one (d ) 2.528 Å) between layers. An even longer bent interaction of the para-H-atom in the



Figure 3. Compound 3b showing the stacked dimers along [110] with the secondary N-H‚‚‚O hydrogen bond forming a chain motif.

pyridine ring (d ) 2.631 Å, 125°) is also formed. This last C-H‚‚‚O interaction is always found within the dimer family as is the four-point [AADD‚‚‚DDAA] dimer. On the other hand, the general arrangement of stacked phenyl substituents and their arrangement into a chain are more reminiscent of the catemer family (Figure 4). In this way, 3b may be regarded as an intermediate between the catemer and the dimer family. Incidentally, it should be noted that the density of 3b is only 1.479 g cm-3 while that of the isomeric 1b is 1.527 g cm-3. This is unusual, and we searched intensively for a polymorph of 3b. However, we did not find one. Still, we feel 3b is the most likely candidate for polymorphism within this series of compounds. As para substitution with substituents larger than fluorine leads to a distinct change in the observed packing (3a f 3b), so does meta substitution with a methyl group. In 2b, we find a two-point recognition pattern and a catemer arrangement. Again, additional N-H‚‚‚O hydrogen bonds (D ) 3.028 Å) are formed along the chain. However, this time, the para-H-atom of the pyridine ring is oriented toward the second sulfonyl O-atom, forming a C-H‚‚‚O hydrogen bond (d ) 2.492 Å) interconnecting the chains. This type of interaction is a feature of the dimer family. A twodimensional hydrogen bond network results (Figure 5). This network is very comparable to the two-dimensional hydrogen-bonded network in the dimer family, and it too forms a double layer with interdigitated phenyl residues, which are incorporated into a herringbone pattern with aromatic C-H‚‚‚π (d ) 2.730 Å) and aliphatic C-H‚‚‚π (d ) 2.898 Å) contacts. The paraH-atom of the phenyl ring forms a C-H‚‚‚O hydrogen bond (d ) 2.647 Å) to the second O-atom of the sulfonyl group. In this respect, too, 2b is an intermediate between the catemer and the dimer family. Only, the dimers are shifted relative to each other to form catemer chains. This is possible because the chain is formed not with screw axes as in the other structures of the catemer family but with a glide plane, putting the substituents on the same side, which is crucial for the interdigitation of the double layer. This is reflected in the cell parameters: a (2b) ) 12.4344 Å equals a (1a, form II) ) 12.1099 Å as the thickness of the double layers; b (2b) ) 10.3960 Å equals b (1a, form II) ) 10.7924 Å as the width of the dimer or chain; and c (2b) ) 9.1134 Å

Figure 4. Top: compound 3b, view down [1 h 10] showing the stacking of chlorophenyl rings and chain motif of dimers. Bottom: compound 1b, view down [010] showing the stacking of chlorophenyl rings. The catemer chain is down the view axis. General similarities between the two structures may be seen.

equals c/2 (1a, form II) ) 8.732 Å as the length of the secondary N-H‚‚‚O linkage motif and b (1a, form I) ) 9.0636 Å as the length of the catemer repeat. The hydrate of the para-nitro derivative 3h is a clear exception to the rule that molecules bind to each other via the AADD fragment. Here, water accepts a N-H‚‚‚O hydrogen bond from the hydropyridine (D ) 2.890 Å); while the catemer chain is still formed, it incorporates only the flanking sulfonylimino O-atom and the amine N-atom. It might be added that while the activation of the phenyl H-atoms by the nitro and sulfonylimino groups is rather strong, neither the H-

706


Kirchner et al.

Figure 5. Compound 2b, view down [100] onto the plane of the two-dimensional N-H‚‚‚O and C-H‚‚‚O hydrogen bond network with a catemer motif cross-linked by C-H‚‚‚O interactions. Note that the phenyl substituents are all situated on the far side of the network, as opposed to the situation depicted in Figure 1, top. Compare this network with the one in Figure 6, top.

atoms nor the nitro groups are incorporated into the hydrogen bond network. Dimer Family. Form II, the thermodynamic polymorph of the parent molecule 1a, is the prototype of the dimer family. The two polymorphs of the methyl, 3e; bromo, 3c; iodo, 3d; methoxy, 3g; and trifluoromethyl, 3f derivatives fall into this family. Two symmetryindependent molecules form an [AADD‚‚‚DDAA] dimer synthon with N-H‚‚‚O and N-H‚‚‚N hydrogen bonds. Each dimer is linked to six others with N-H‚‚‚O and C-H‚‚‚O interactions (Figure 6, top) to form a layer, and all phenyl or tolyl substituents protrude from the same side of the layer. The double layer is constituted with interdigitated aryl rings forming a herringbone motif (Figure 6, bottom). The structures in the dimer family are characterized by this hydrogen bond network forming a double layer. Both polymorphs of the methyl derivative 3e are very similar to form II of the parent molecule 1a. Unlike in 1a where the dimorphs are structurally very distinct, the polymorphs of 3e are mere variants. Form I of 3e adopts the space group P21/n with Z′ ) 2. P21/n can be transformed to P21/c, and then, a is changed to 17.0663 Å, c is changed to 17.6836 Å, and β is changed to nonconventional 128.836°. With this transformation, b and c become directly comparable to the axial lengths of form II of 1a (see Table 1). The stacking of double layers is along transformed a, and this axis is obviously lengthened due to the para-methyl substituents. The hydrogen bond network is relatively flat, and the distance between the networks is 3.40 Å between double layers and 9.89 Å within a double layer. For the parent compound 1a (form II), the corresponding values are 3.37 and 8.64 Å. The increase of 1.25 Å represents the change from a C-H spatial requirement of 2.3 Å to a C-CH3 requirement of 3.3-3.8 Å. It should be noted that the density of crystalline 3e form I is only 1.367 g cm-3 (1.379 g cm-3 for form II) as compared to 1.463 g cm-3 in 1a form II. Form II of 3e adopts the space group P21/c with Z′ ) 4. Similarities with form II of 1a are again obvious from the axial similarities (Table 1). The two dimers in the asymmetric unit are related by pseudosymmetry and can be fitted to a dimer of form I plus its c-glide related equivalent with an rms deviation of less than

Figure 6. Top: compound 3e form I, view on the (101 h ) plane; the dimers are connected by four strong hydrogen bonds. Each dimer is surrounded by six others to which it is linked by N-H‚‚‚O and C-H‚‚‚O hydrogen bonds. All para-tolyl substituents are on the far side of the resulting two-dimensional hydrogen-bonded layer. Bottom: compound 3e form I, down [010]; hydrogen-bonded networks interdigitate with the phenyl substituents to form double layers.

0.1 Å. In effect, the geometries of the hydrogen-bonded networks and of the layers themselves are homomorphic. While the general features of the double layers of the three compounds in the series are the same, there are interesting differences in the stacking of the layers (Figure 7). If each dimer is envisioned as a sphere that is close-packed with six neighbors in the same layer, the change from form II of 1a to form I of 3e is equivalent to a change from ABABAB to quasi-ABCABC stacking of these layers. In form II of 3e, however, stacking of layers is achieved by a doubling of the asymmetric unit, effectively shifting layers by -c/4, in terms of form II of 1a. So, instead of having three neighbors in the next layer, a dimer has two nearest neighbors and two next-nearest neighbors. The next four derivatives 3c,d,f,g in the dimer family may be grouped into a subfamily since they show a common deviation from the methyl derivatives and the parent 1a form II packing. The general arrangement and hydrogen bond pattern stays the same (see Figure 8), but the dimer orientation is different in that the



Figure 8. Compound 3d, view down [001]. The central dimer is surrounded by six dimers and connected to four by N-H‚‚ ‚O hydrogen bonds. Note that the highlighted dimer on the right has its phenyl substituents pointing toward the observer, in contrast to the arrangement depicted in Figure 6, top.

Figure 9. Compound 3d, view down [010]; the slant of both pyridine ring systems can be seen, while the general features of the double layer formation are kept. Between the double layers, a C-H‚‚‚O hydrogen bond is formed.

Figure 7. Differences in the stacking of layers in the dimer family. The dimers are indicated by spheres. Top: compound 1a form II with ABABAB stacking. Middle: compound 3e form I with quasi-ABCABC stacking. Bottom: compound 3e form II stacking without closest packing.

pyridine rings are slanted with respect to the hydrogen bond network. In this way, the previously planar two-

dimensional hydrogen bond network becomes more corrugated, enabling weak C-H‚‚‚O hydrogen bridges between these networks (Figure 9). Because a corrugated surface will enhance the rate of crystal growth, this packing variation might be kinetically favored. While 3d,f,g share the same space group Pbca with Z′ ) 2 and have similar cell parameters, 3c (P21/c, b ) 94.33°, Z′ ) 4) is obviously comparable in its cell dimensions. The cell of bromo compound 3c is very similar to form II of 3e. The packing of layers is slightly different from that found in the parent 1a and the methyl derivatives 3e in that here every dimer exactly faces another dimer in the adjacent layer. In general, it might be said that the intermolecular geometry of the dimer changes in going from the planar to the slanted subfamilies. The mean N-H‚‚‚O bond length shortens from D ≈ 2.8 Å in the planar dimer to D ≈ 2.75 Å in the slanted dimer while the neighboring

708


Kirchner et al.

Figure 11. Overlay of AADD fragments in the dimers so that both C-atoms and the pyridine N-atom coincide. In the lower part, two different geometries within the dimers are discernible. Color scheme: 1a form II, red; 3b, orange; 3c, green; 3d, pink; 3e form I, brown; 3e form II, blue; 3f, yellow; and 3g, magenta. This figure is in effect a sampling of the packing landscape in this series.

Figure 10. Scatter plot of D (N‚‚‚O) distances vs neighboring D (N‚‚‚N) distances in the dimers. Lines connect D/D pairs within one dimer. The planar subfamily structures have shorter hydrogen bonds to the N-atom acceptor (lower right area, pink squares) while the slanted subfamily structures have shorter hydrogen bonds to the O-atom acceptor (upper left area, green squares). A blue circle is used for the chloro derivative 3b.

N-H‚‚‚N bond lengthens on average from D ≈ 3.0 Å to D ≈ 3.15 Å (Figure 10). As expected from the relative strengths of the typical N-H‚‚‚O and N-H‚‚‚N hydrogen bonds, the distance change is less for the stronger N-H‚‚‚O hydrogen bond. Apparently, this lengthening and shortening is mutually compensative; the overall stabilization of the four-point synthon seems to be unaffected. A calculation at the RHF/6-31G* level of both dimer geometries in form I 3e and 3g results in an energy difference of only 1.34 kcal mol-1. A possibly alternative DADA type of dimer geometry after formation of the tautomer is not found in any of the structures nor is the tautomer observed in DMSO solution by NMR. These results confirm the observations of Meijer that a prevalence of AADD over DADA is expected because of additional H-atom repulsion in alternating hydrogen bonds.6 However, these changes in mean hydrogen bond distances cannot be expressed in simple geometrical terms, and anyway, the AADD fragment was not optimized to give a perfectly flat dimer, as in the case of say Meijer’s systems.6 The pyridine ring and its immediate N-atom substituents are fixed in a planar geometry. However, the longer (covalent) bond distances of the S-atom put the acceptor O/N pair about 0.1 Å further apart from each other than the NH/NH donor pair, introducing a mismatch that cannot be aligned properly in an idealized flat [AADD‚‚‚DDAA] binding. Also, the tetrahedral angle around the S-atom does not fit into the line of the 120° angles along sp2 C- and

N-atoms, and the acceptor O-atom is situated in front of the N-atoms, which is unfavorable (Figure 6, top). The AADD system is not exactly flat either, because the exocyclic C-N bonds are subject to torsional reorientation. This is even true for the sulfonylimino C-N bond, which has formal double bond character but will adopt NCNS torsion angles down to 143°. With this torsional freedom, the monomer adapts to fine tune the nonplanar dimer. It should be noted that the three flexible bonds (C-N, N-S, and S-C) are linked and must cooperate to allow a molecular geometry for both hydrogen bonding and close packing. A superposition of the AADD dimers in this study is shown in Figure 11. A change of general orientation might be noted within the dimer family as is clearly seen by the gap between overlaid dimers in the lower part of Figure 11. The structures of 3e form II (blue) and 3c (green), which both have P21/c and Z′ ) 4 and surprisingly similar cell parameters, also share a common feature in Figure 11 in that only they have a dimer in each geometry. This might be seen as transitional packing between the two subfamilies. Nevertheless, chemically equivalent hydrogen bonds have lengths that differ up to 0.1 Å even within the same dimer, as can be seen by the length of the connecting lines in Figure 10. So, there is no real symmetrical, optimized arrangement of the dimer; rather, the close packing of the phenyl substituents within the double layer of the dimer family structures influences the dimer geometry. Conclusions This study began as an attempt to identify polymorphs of derivatives of a compound, 1a, which was known to be dimorphic and with the two structures being fundamentally different.1 A large number of derivatives, 12 in all, were prepared and crystallized, and each compound was crystallized around 6-10 times employing different conditions. None except 3e was found to be polymorphic, and the variations in the


crystal structure, had they been found in the same compound, would have constituted a realization of the originally stated aims of this work. However, polymorphism was not found in a general sense, and yet we have found that the crystal structures of the compounds in this study may be divided into two families. The catemer family has a one-dimensional hydrogen bond structure that closely resembles the kinetic form of the reference compound 1a. The dimer family has a zero-dimensional structure that resembles the thermodynamic polymorph of 1a. While 3e is technically dimorphic, both forms belong to the same (dimer) family. Derivatives 2b and 3c crystallize in crystal structures that might be seen as intermediates between the catemer and the dimer families. What is more notable is that none of the compounds studied here have structures in both the catemer and the dimer families and apparently the kinetic/thermodynamic contradiction that was such a major issue with compound 1a1 does not really matter for the substituted derivatives. Tentatively, we conclude that the kinetic and thermodynamic forms are one and the same for these derivatives, and therefore, polymorphism is not observed. It is truly ironic then that compound 1a was given for the CSP exercise as a blind test molecule.3 Had any of its many derivatives been given, the chances of a correct prediction seem (now, and in hindsight) much brighter. This having been said, it is possible that polymorphism might still be found in the chloro derivative 3b, since this compound has an intermediate packing pattern and an anomalously low density. To be sure, we tried many crystallization experiments with this compound, but no polymorphs emerged. Whether these will emerge in the future is entirely speculative at the present time. However, there is another way of considering the present results. If the possible structural variations for any given molecule might be said to constitute a “packing landscape”,5 the landscapes for all of the derivatives studied in this paper might be said to be broadly similar. Variation of experimental conditions of crystallization is one way of sampling a particular landscape, and when one arrives at a local minimum, a new polymorph is found. This study shows that substitutional variation is a good way of sampling different regions of different landscapes, but these different landscapes are generally similar within a chemical family. Therefore, structural variations that are found for differently substituted derivatives (although not yet formally polymorphic) represent putative polymorphic structures that might still be found for any one particular compound if the experimental conditions are made more exotic or very possibly be identified computationally with programs that generate large number of polymorph structures. We conclude by merely restating several facts that are well-known. (i) Polymorphism cannot be predicted. (ii) Mc Crone’s rule does not hold in a general sense if


all experiments are carried out at normal pressures. (iii) Crystal packing is full of subtleties, but the decisiveness with which a particular packing is adopted seems not to correlate with the subtleties of the individual interactions. The whole is certainly greater than the sum of the parts, but this then is a dictum in supramolecular thinking. Acknowledgment. M.T.K. thanks the Alexander von Humboldt Foundation for financial support. L.S.R. thanks the CSIR for a fellowship. R.K.R.J. and R.B. are grateful to DFG for financial assistance. Supporting Information Available: X-ray crystallographic information file (CIF) for all derivatives. This material is available free of charge via the Internet at http:// pubs.acs.org.

References (1) Jetti, R. K. R.; Boese R.; Sarma, J. A. R. P.; Reddy, L. S.; Vishweshwar, P.; Desiraju, G. R. Angew. Chem., Int. Ed. 2003, 42, 1963-1967. (2) Sarma, J. A. R. P.; Desiraju, G. R. Cryst. Growth Des. 2002, 2, 93-100. (3) Motherwell, W. D. S.; Ammon, H. L.; Dunitz, J. D.; Dzyabchenko, A.; Erk, P.; Gavezzotti, A.; Hofmann, D. W. M.; Leusen, F. J. J.; Lommerse, J. P. M.; Mooij, W. T. M.; Price, S. L.; Scheraga, H.; Schweizer, B.; Schmidt, M. U.; van Eijck, B. P.; Verwer, P.; Williams, D. E. Acta Crystallogr. 2002, B58, 647-661. (4) Desiraju, G. R. Nat. Mater. 2002, 1, 77-79. (5) Blagden, N.; Davey, R. J. Cryst. Growth Des. 2003, 3, 873885. (6) Sijbesma, R. P.; Meijer, E. W. Chem. Commun. 2003, 5-16. (7) Tuda, K.; Izijawa, Z.; So, D. J. Pharm. Soc. Jpn. 1939, 59, 213-215. (8) There have been other reports that crystallization from nitromethane yields the thermodynamic modification for hydrogen-bonded crystals. The reason for this is not clear, but it could have to do with the fact that nitromethane also exists as another tautomer, which can block hydrogen bond sites in a molecule via two point recognition. We thank Prof. David Grant (University of Minnesota) for helpful discussions on this point. (9) SHELXTL, Version 6.12; Bruker AXS Inc.: Madison, WI, 2000. (10) (a) Pranata, J.; Wierschke, S. G.; Jorgensen, W. L. J. Am. Chem. Soc. 1991, 113, 2810-2819. (b) Biradha, K.; Nangia, A.; Desiraju, G. R.; Carrell, C. J.; Carrell, H. L. J. Mater. Chem. 1997, 1111-1122. (c) Mele´ndez, R. E.; Hamilton, A. D. Hydrogen-bonded ribbons, tapes and sheets as motifs for crystal engineering. In Design of Organic Solids; Weber, E., Ed.; Springer: Berlin, 1998; pp 97-129. (d) Whitesides, G. M.; Simanek, E. E.; Mathias, J. P.; Seto, C. T.; Chin, D. N.; Mammen, M.; Gordon, D. M. Acc. Chem. Res. 1995, 28, 3744. (e) Whitesides, G. M.; Mathias, J. P.; Seto, C. T. Science 1991, 254, 1312-1319. (f) Seto, C. T.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, 905-916. (g) Zerkowski, J. A.; Seto, C. T.; Whitesides, G. M. J. Am. Chem. Soc. 1990, 112, 9025-9026. (h) Ranganathan, A.; Peddireddi, V. R.; Rao, C. N. R. J. Am. Chem. Soc. 1999, 121, 1752-1753. (11) To our knowledge, this is the first reported such example of a short F‚‚‚S contact. The CSD contains only 10 contacts of this type less than 3.5 Å.

CG049912T

Structural Variations and Polymorphism of Some ... - ACS Publications

Recommend Documents