Competition between Polar and Centrosymmetric Packings inMolecular Crystals: Analysis of Actual and Virtual StructuresRoberto Centore,*,† Sandra Fusco,† Fabio Capone,‡ and Mauro Causa ‡
†Department of Chemical Sciences, University of Naples ”Federico II”, Via Cintia, I-80126 Naples, Italy‡Department of Chemical, Materials and Production Engineering, University of Naples “Federico II”, Piazzale Tecchio, I-80125Naples, Italy
*W Web-Enhanced Feature *S Supporting Information
ABSTRACT: Imines obtained by condensation of 4-hydroxy-benzohydrazide with aliphatic ketones are a rare example of a classof compounds showing a remarkable tendency to crystallize inacentric polar space groups (Pna21 or Cc). In fact, all of the(seven) compounds studied up to now show at least one polarpolymorph. In some cases, polymorphism was detected, and anonpolar centrosymmetric phase was also identified (P21/c orP21/n space group). With the aim to disclose the conditions thatcan favor the formation of acentric structures in molecular crystals,we report, in this paper, a theoretical analysis (ab initio densityfunctional theory with periodic boundary) of the lattice energyand density of all the packing modes observed in the whole set ofimines. The computational analysis has been performed byoptimizing each compound in its own experimental packings (actual crystal structures) and also in the packings of the othercompounds of the class (virtual structures). The experimental crystallographic data and the theoretical analysis suggest that twoconformers, basically differing for the orientation of the phenolic H atom in the plane of the phenyl ring, compete, in solution, forthe formation of polar or centrosymmetric packings. The transitions between polar and centrosymmetric polymorphs are ofdiffusive type, and single crystals are not preserved, while the transitions between different polar polymorphs can be of single-crystal-to-single-crystal type.
■ INTRODUCTION
In a polar crystal, there is a direction that is not transformed inthe opposite direction by any symmetry operation of the crystalclass. That direction is called the polar axis of the crystal.1
There is a general interest in polar crystals, because somephysical properties of materials, represented by odd ranktensors and highly desired for advanced applications, such aspyroelectricity, piezoelectricity, ferroelectricity, second harmon-ic generation, electrooptic effect, are only allowed or they arestrongly enhanced in polar space groups.1 The center ofsymmetry transforms each direction in the opposite one, so thatcentrosymmetric crystals are not polar. Only 10, out of the 21classes lacking the center of symmetry, are polar. As a matter offact, polar symmetry is rare. It is well-known that a largefraction of organic nonchiral compounds crystallize incentrosymmetric space groups, mainly P21/c and P1 ,2−4 andalso in crystals of enantiomerically pure chiral compounds, themost frequent space group is P212121 that is acentric but notpolar.1−4 Actually, the bias for centrosymmetric over acentriccrystals is a problem lying at the very heart of crystallography,addressed since the beginning of the discipline. It is still a longdebated, challenging problem of the structural science.5−8
In view of all this, it is noteworthy that we have found in theimines obtained by condensation of 4-hydroxybenzohydrazide
with aliphatic ketones, Chart 1,9,10 a class of compounds with apersistent tendency to form acentric polar crystal structures(Pna21 or Cc space groups).11−16
That tendency was checked against change of the ketone,spanning from acyclic (acetone, methylethylketone) to cyclicketones (cyclobutanone, cyclohexanone);9 it was confirmed indifferent polymorphs of the same compound (out of the fourpolymorphs identified up to now for imine 1, three are polar,with the transitions between the polar phases being of single-crystal-to-single-crystal type);9,17 finally, the tendency wasconfirmed in the case of the racemic mixture of a chiralimine.10 In the latter case (rac-3-methylcyclopentanone as theketone in imine 5), two different crystal phases were observed:a polar phase (Pna21) and a centrosymmetric phase (P21/n),with the second being thermodynamically stable at roomtemperature.In the present paper we report the solid state analysis of the
new compound 4, in which cyclopentanone was used as theketone reagent, and a comparative theoretical analysis (ab initiodensity functional theory (DFT) with periodic boundary) of
Received: January 12, 2016Revised: February 29, 2016Published: March 1, 2016
the lattice energy and density of all the different modes ofpacking observed for the whole set of imines studied up to now.The aim is to find a rationale for the observed tendency to formnon-centrosymmetric structures.
■ EXPERIMENTAL SECTIONMaterials and Methods. Differential scanning calorimetric (DSC)
analysis was performed using a PerkinElmer Pyris instrument, underflowing nitrogen, at 10 K/min scanning rate. Temperature controlledoptical microscopy was performed with a Zeiss Axioskop polarizingmicroscope equipped with a Mettler FP90 heating stage. 1H NMRspectra were recorded with a Varian spectrometer operating at 200MHz. All chemicals were obtained commercially and used as received,except for 4-hydroxybenzohydrazide, which was prepared as describedin the literature.18
Synthesis of 4. The compound was obtained by refluxing, inabsolute ethanol, 4-hydroxybenzohydrazide and cyclopentanone (instoichiometric ratio 1:1.30 by mol) for 1 h (20 mL of ethanol wereused for 0.5 g of 4-hydroxybenzohydrazide). After that period, abouthalf of the solvent was removed by gentle boiling, and, on cooling toroom temperature, the product was obtained as a white crystallinesolid, which was recovered by filtration and dried in oven at 100 °C for2 h. Single crystals for X-ray analysis were grown from ethanol solutionby evaporation.4. N′-Cyclopentylidene-4-hydroxybenzohydrazide. C12H14N2O2.
Yield 88%; mp 268 °C (from EtOH, dec). δH (200 MHz; DMSO-d6, 25 °C) 1.73 (4 H, m), 2.39 (4 H, m), 6.80 (2 H, d, 3J = 8.2 Hz),7.69 (2 H, d, 3J = 8.8 Hz), 10.01 (2 H, s, broad) ppm.X-ray Analysis. All data for crystal structure determinations were
measured on a Bruker-Nonius KappaCCD diffractometer equippedwith Oxford Cryostream 700 apparatus, using graphite monochro-mated MoKa radiation (0.71073 Å). Reduction of data andsemiempirical absorption correction were done using SADABSprogram.19 The structures were solved by direct methods (SIR97program20) and refined by the full-matrix least-squares method on F2
using SHELXL-97 program21 with the aid of the program WinGX.22 Hatoms bonded to C were generated stereochemically and refined bythe riding model; those bonded to O and N were found in differenceFourier maps and their coordinates were refined. To all H atoms, Uisoequal to 1.2 times Ueq of the carrier atom was given. Crystal andrefinement data are summarized in Table 1. The analysis of the crystal
packing was performed using the program Mercury,23 which was alsoused for the calculation of the powder diffraction patterns. Hirshfeldsurface analysis24 was performed using the program CrystalExplorer.25
Computational Details. Ab initio calculations with periodicboundary conditions were performed using the Hybrid DensityFunctional method: the Becke’s B3LYP functional was applied.26 Wehave used a 6-31G(dp) Gaussian basis set,27 which gives reasonableresults for organic molecular crystals.28,29 The dispersion forces weretreated by the Grimme’s method,30,31 adopting the parametersoptimized for molecular crystals.28,29 The treatment of the dispersionforces using a damped Lennard-Jones potential as proposed byGrimme30,31 is absolutely mandatory in molecular crystals for gettingrealistic optimized geometries. The lattice energy (Ulat) was calculatedas the difference between the DFT molar energy of the optimizedcrystal and the DFT molar energy of the free molecule. The crystalstructures were optimized with respect to both the atomic fractionalcoordinates and the lattice parameters, keeping the experimental spacegroup symmetry. The generation of the starting virtual structures to beoptimized has been performed as already described32 and is detailed inthe Supporting Information (henceforth SI). All calculations wereperformed using the CRYSTAL program.33
■ RESULTS AND DISCUSSIONPolymorphism of Imine 4. Crystallization of 4 produces
samples of different morphologies, and this can suggest solidstate polymorphism.34,35 As it will be clear later, the twomorphologies do correspond to two different polymorphs: apolar orthorhombic modification (Pna21, Figure 1a) and acentrosymmetric monoclinic modification (P21/c, Figure 1b).The polymorphism was confirmed by DSC analysis and opticalobservations. In the DSC heating curve of a sample onlycontaining prisms like that of Figure 1a, melting at 268 °C isthe only transition recorded (see SI). When a lozenge-shapedcrystal as that of Figure 1b is heated under the polarizingmicroscope, an irreversible solid−solid transition is observed at175 °C, and the solid phase obtained melts at 268 °C. So, for 4,we deduce that the polar orthorhombic phase is thermodynami-cally stable in the whole investigated temperature range up tomelting, while the monoclinic centrosymmetric phase ismetastable. Moreover, the solid state transition from themonoclinic to the orthorhombic phase is not topotactic, andsingle crystals are not preserved during the transition (see SI fora detailed analysis; a movie of the transition in avi format is alsoavailable). This suggests that the transition is of diffusive type
Chart 1. Chemical Formulae of the Studied Compounds andthe Space Groups of the Observed Crystal Phasesa
aIn the case of 5, the racemic mixture is considered.
Table 1. Crystal and Refinement Data for the TwoPolymorphs of 4
(nucleation and growth process) with relevant differences inthe crystal packing of the two phases.Discussion of the Crystal Structures of 4. The X-ray
molecular structures of the orthorhombic and monoclinicmodifications of 4 (henceforth 4ort and 4mon respectively) areshown in Figure 2.
The two molecules basically differ for the orientation of thephenolic hydrogen in the plane of the phenyl ring: in Figure 2ait is placed on the same side of the lone pair on N2, while inFigure 2b it is on the opposite side. It is important to note thatthe orientation shown in Figure 2a is common to all polarpolymorphs (space group Pna21) of the imines of Chart 1,while the orientation of Figure 2b is found in all thecentrosymmetric polymorphs (P21/c or P21/n) and in thepolar polymorph Cc. So, it is as if the two conformations fitteddifferent packings. Of course, the sole observation of theconformational polymorphism of 4 implies that both theconformers are present in solution.36
The packing of 4ort is completely analogous to the imines ofChart 1 with the same space group, described in refs 9 and 10.Amide chains running along a are formed by H-bondingbetween the N−H donor and the carbonyl acceptor.37 Thechains formed through H-bonding between the O−H donor
and imino N acceptor are wrapped around the polar c axis. Interms of symmetry, the two orthogonal chains, shown in Figure3, are generated by a glide plane and by a 21 axis parallel to theglide plane. This fixes the crystal class as mm2, non-centrosymmetric and polar (see SI for a more detaileddiscussion based on topological and symmetry considerations).The H bonding patterns of 4mon are similar to 4ort, but the
H-bonded chains are no longer orthogonal to each other.Amide chains parallel to c are formed by H bonding betweenthe N−H donor and the carbonyl oxygen acceptor of gliderelated molecules. Chains of H bonded molecules parallel to (a+ c/2) are formed by H bonding between O−H donors andimino N acceptors of glide related molecules. The two patternsare shown in Figure 4. In this case, the glide plane and the 21axis are perpendicular, and the crystal class is 2/mcentrosymmetric. The packing of 4mon is completelyanalogous to the monoclinic P21/c structure of 1.17 The twocrystal structures can be considered isomorphous.So, notwithstanding the coincidence of the H bonding
synthons in the two structures and the formation of chainsbetween the same set of donor and acceptor groups, the crystalpacking is basically different in the two cases.The packing diagrams reported in Figure 5, in particular,
clearly evidence the polarity of the packing of 4ort, and thecentrosymmetric, non polar packing of 4mon.
Theoretical Calculations, Actual and Virtual Struc-tures. From the present and previous works,9,10,17 altogetherfour different packings have been identified for the compoundsof Chart 1: two polar packings (Pna21 and Cc space groups)and two centrosymmetric packings (P21/c and P21/n spacegroups). Following a recently proposed approach,32 we haveoptimized, by ab initio DFT calculations with periodicboundary, each imine in its own experimental packings (actualcrystal structures) and also in the different crystal packings ofthe other imines (as an example, for 4, also in the packing Cc ofimine 3 and P21/n of racemic imine 5). We will call virtualcrystal structures these latter, since they have not (yet) beenfound experimentally.32 In this way, we have access tocalculated thermodynamic properties (lattice energy anddensity) of each imine in all the crystal packings observedexperimentally for the whole set of compounds.38 Imines 2 and5 have not been considered in this computational study becausetheir crystal structures show some disorder.9 The results areshown in Tables 2−5.A survey of Tables 2−5 reveals that the polar orthorhombic
structure has always a low lattice energy in all the studiedcompounds, within 0.6 kcal/mol of the absolute minimum, andits lattice energy monotonically decreases with increasing thesize of the R group (Chart 1). The polar monoclinic packing(Cc) has instead a high lattice energy in all the imines (≥2.4kcal/mol of the absolute minimum) but 3, in which its latticeenergy is the lowest (actual structure). The trend of the latticedensity with the size of the R group is remarkable. If weconsider the centrosymmetric P21/c packing, the lattice densitymonotonically decreases by increasing the size of R, so that themaximum density is for 1 and the minimum for 6. In the case ofthe polar packing Pna21, and of the centrosymmetric packingP21/n, the trend is exactly reversed, as the density substantiallyincreases by increasing the size of R. In the case of the polarpacking Cc the trend is not monotonic, and ρlat has themaximum just for 3. So, it seems that in the centrosymmetricpackings it is possible to optimize the lattice density both for Rsmall and large. The analysis of Tables 2−5 also shows that, in
Figure 1. (a) Prismatic crystal of the orthorhombic modification of 4;(b) lozenge shaped crystal of the monoclinic modification of 4. In bothcases the scale bar is 200 μm.
Figure 2. (a) Ortep diagram of 4ort. (b) Ortep diagram of 4mon. Inboth cases, thermal ellipsoids are drawn at 30% probability level.
general, the packing with the lowest lattice energy also exhibitsa high density; the reverse, however, is not true because in
many cases the highest density is shown by the packing Cc thathas a high lattice energy.Concerning the polymorphism of 1 and 4, a relevant
difference comes out from the inspection of Tables 2 and 4. Infact, in 4 the two actual structures have lattice energy anddensity similar to each other (ΔUlat = 1.1 kcal/mol, Δρlat =0.015 g/cm3). In the case of 1, on the contrary, the minimum ofUlat is reached in two different packings (both actual structures)
Figure 3. Different H-bonded chains in the crystal structure of 4ort. Structural data of the H-bonds: 37N1−H···O2i: 0.81(4), 2.31(4), 3.122(4) Å,178(4)°, i = 1/2 + x, 1.5 − y, z; O1−H···N2ii: 0.97(5), 1.86(5), 2.827(4) Å, 169(4)°, ii = −x, 2 − y, −1/2 + z.
Figure 4. Different H-bonded chains in the crystal structure of 4mon. Structural data of the H-bonds: 371−H···O2i: 0.83(3), 2.29(3), 3.115(3) Å,172(3)°, i = x, 1/2 − y, −1/2 + z; O1−H···N2ii: 1.00(3), 1.82(3), 2.781(3) Å, 159(2)°, ii = −1 + x, 1/2 − y, −1/2 + z.
Figure 5. (a) Partial crystal packing of 4ort viewed down a; (b) partialcrystal packing of 4mon viewed down c.
Table 2. Calculated Lattice Energy (kcal/mol) and LatticeDensity (g/cm3) for Actual and Virtual Crystal Structures of1 at 0 Ka
having quite different lattice density (Δρlat = 0.18 g/cm3). That“softness” of the othorhombic packing of 1 can be related withits rich polymorphism and single-crystal-to-single-crystaltransitions.9,17
The pattern emerging from the analysis of the actual andvirtual crystal structures is intriguing and shows apparentcontradictions that can be summarized as follows:(1) In some cases, two crystal structures with slightly
differing thermodynamic parameters are not both observedexperimentally, but only one.39 This is the case, for instance, ofthe two centrosymmetric structures of 4 (ΔUlat ≅ 0.2 kcal/mol), but also of the two centrosymmetric structures of 1(ΔUlat ≅ 0.6 kcal/mol);(2) In other cases, two crystal structures with slightly
differing lattice energy are both observed experimentallynotwithstanding a relevant difference in lattice density. This isthe case, for instance, of the polar and centrosymmetricstructures of 1 (ΔUlat ≅ 0 kcal/mol, (Δρlat≅ 0.2 g/cm3);(3) In other cases, two crystal structures with thermody-
namic parameters differing by a greater extent are bothobserved experimentally. This is the case of the Pna21 andP21/c structures of 4 (ΔUlat ≅ 1.1 kcal/mol).In the search of a possible rationalization of these results, we
note that case 1 corresponds to crystal structures that are bothcentrosymmetric and in which the molecular conformation isthe same. Cases 2 and 3 corresponds to crystal structuresprofoundly different in their symmetry (i.e., polar versuscentrosymmetric) and molecular conformation (see Figure 2).So, the reason could be related with the balance betweenkinetic and thermodynamic factors. In the first stages ofnucleation, before critical clusters are formed, with reference to1, the activation energy for the interconversion is low, andtherefore the structure with lower (free) energy is selected.34,35
With reference to 2 and 3, the profound difference in symmetryand molecular conformation results in a higher activationenergy for the interconversion and thereby both crystalstructures can develop independently. So, the presence insolution of both conformers is a prerequisite for polymorphism,because the two conformers fit different sets of packings (polaror centrosymmetric). The actual observation of it relies on therelative (free) energies of the two sets of packings.As a final remark, we note that data of Tables 2−5 provide a
confirmation, on a completely different class of compounds andon a set of space groups including acentric ones, of thehypothesis of virtual isomorphism that we have recentlyproposed (see also the SI): in a class of similar compounds,all the different modes of packing observed experimentally forsingle members of the class correspond to minima of the latticeenergy and to acceptable lattice densities for every member ofthe class.32
■ CONCLUSIONSThe tendency to form acentric polar crystal structures in iminesobtained by condensation of 4-hydroxy-benzohydrazide with
aliphatic ketones has been further confirmed experimentallyand is quite remarkable if compared with other literatureexamples.12−16 In fact, all the seven imines investigated up tonow have a polar crystal structure, either as the thermodynamicstable phase or as a metastable phase. This confirms thattransverse molecules9,40 are real candidates to yield acentricpolar crystals in high score.A general look at the experimental and theoretical results
reported in this paper would also suggest the provocativeobservation that, in the case study, the bias for acentric overcentric structures (or the reversed bias) is only apparent andstrongly dependent on the identification of the polymorphs.After all, several imines of Chart 1 show, at the present level ofinvestigation, both an acentric and a centric polymorph (all inspace groups that, according to Kitaigorodskii, allow the closepacking) and, in some cases in which only one has been found,there are reasons to believe that also the other could beobtained.41 In many cases, one of the polymorphs wasidentified by chance, or after several trials and in a scarcelyreproducible way or it was long sought, and eventually found,because it was supposed to exist.42
■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.cgd.6b00054.
Analysis of the polymorphism of 4. Symmetry andtopological analysis of H bonding patterns in relationwith the different conformation of the imine molecules.Analysis of Hirshfeld fingerprint plots of 4ort and 4mon.Discussion of the optimized actual and virtual structures.1H NMR spectrum of 4 (PDF)
*W Web-Enhanced FeatureMovie of the phase transition from the monoclinic to theorthorhombic phase of 4, recorded on a single crystal.Accession CodesCCDC 1417087−1417088 contains the supplementary crys-tallographic data for this paper. These data can be obtained freeof charge via www.ccdc.cam.ac.uk/data_request/cif, or byemailing [email protected], or by contacting TheCambridge Crystallographic Data Centre, 12, Union Road,Cambridge CB2 1EZ, UK; fax: +44 1223 336033.
■ ACKNOWLEDGMENTSThanks are due to the CRdC NTAP of Regione Campania(Italy) for the X-ray facility. R.C. thanks the COST Associationfor support and critical discussions within the COST ActionCM-1402- Crystallize.
■ REFERENCES(1) Hahn, Th.; Klapper, H. Point Groups and Crystal Classes. InInternational Tables for Crystallography, Vol. A, Hahn, Th., Ed.; D.Reidel Publishing Company: Dordrecht (Holland), 1983; pp 746−785.(2) Mighell, A. D.; Himes, V. L.; Rodgers, J. R. Acta Crystallogr., Sect.A: Found. Crystallogr. 1983, 39, 737−740.
Table 5. Calculated Lattice Energy (kcal/mol) and LatticeDensity (g/cm3) for Actual and Virtual Crystal Structures of6 at 0 Ka
(3) Padmaja, N.; Ramakumar, S.; Viswamitra, M. A. Acta Crystallogr.,Sect. A: Found. Crystallogr. 1990, 46, 725−730.(4) The most recent figures released by the Cambridge StructuralDatabase, on February 16, 2015 (www.ccdc.cam.ac.uk/SupportandResources/Resources/pages/Resources.aspx), are 78.1%of structures in centrosymmetric space groups (58.9% in P21/c andP1 ), 21.9% in non-centrosymmetric space groups and 16.5% in chiralspace groups.(5) The fundamental work on this problem was delivered byKitaigorodskii who, in his book (Kitaigorodskii, A. I. Organic ChemicalCrystallography; Consultant Bureau: New York, NY, 1961) ranked thespace groups in terms of suitability for the close packing of moleculesof arbitrary shape. His study ruled out many of the 230 space groups asunfit for the close packing. It resulted, in particular, that the two mostfrequent space groups, P21/c and P1 (both centrosymmetric), areparticularly suited for close packing of molecules. However, accordingto his analysis, other, far less frequent, non-centrosymmetric (andpolar) space groups, are also suited for close packing (e.g., P21, Pna21,Pca21). More recent contributions on this topic are given in refs 6−8.(6) Brock, C. P.; Dunitz, J. D. Chem. Mater. 1994, 6, 1118−1127.(7) Dunitz, J. D.; Filippini, G.; Gavezzotti, A. Helv. Chim. Acta 2000,83, 2317−2335 (“The centre of symmetry is the best packing operatorfor molecules with awkward shape”)..(8) Kelley, S. P.; Fabian, L.; Brock, C. P. Acta Crystallogr., Sect. B:Struct. Sci. 2011, 67, 79−93.(9) Centore, R.; Jazbinsek, M.; Tuzi, A.; Roviello, A.; Capobianco, A.;Peluso, A. CrystEngComm 2012, 14, 2645−2653.(10) Centore, R.; Fusco, S.; Jazbinsek, M.; Capobianco, A.; Peluso, A.CrystEngComm 2013, 15, 3318−3325.(11) Non trivial examples of classes of compounds with tendency toform acentric crystals are rare. References 12−16 contain a list withpertinent citations.(12) Nitroanilines: Panunto, T. W.; Urbanczyk-Lipkowska, Z.;Johnson, R.; Etter, M. C. J. Am. Chem. Soc. 1987, 109, 7786−7797.(13) Ortho-substituted benzoic acids: Frankenbach, G. M.; Etter, M.C. Chem. Mater. 1992, 4, 272−278.(14) Hydrazone derivatives: Serbutoviez, C.; Bosshard, C.; Knopfle,G.; Wyss, P.; Pretre, P.; Gunter, P.; Schenk, K.; Solari, E.; Chapuis, G.Chem. Mater. 1995, 7, 1198−1206.(15) Phenolic polyene compounds: Kwon, O.-P.; Jazbinsek, M.; Yun,H.; Seo, J.-I.; Seo, J.-Y.; Kwon, S.-J.; Lee, Y. S.; Gunter, P.CrystEngComm 2009, 11, 1541−1544.(16) N-substituted stilbazolium compounds: Kim, P. J.; Jeong, J.-H.;Jazbinsek, M.; Kwon, S.-J.; Yun, H.; Kim, J.-T.; Lee, Y. S.; Baek, I.-H.;Rotermund, F.; Gunter, P.; Kwon, O.-P. CrystEngComm 2011, 13,444−451.(17) Sahoo, S. C.; Panda, M. K.; Nath, N. K.; Naumov, P. J. Am.Chem. Soc. 2013, 135, 12241−12251.(18) Centore, R.; Carella, A.; Tuzi, A.; Capobianco, A.; Peluso, A.CrystEngComm 2010, 12, 1186−1193.(19) SADABS, Bruker-Nonius: Delft, The Netherlands, 2002.(20) Altomare, A.; Burla, M. C.; Camalli, M.; Cascarano, G. L.;Giacovazzo, C.; Guagliardi, A.; Moliterni, G. G.; Polidori, G.; Spagna,R. J. Appl. Crystallogr. 1999, 32, 115−119.(21) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr.2008, 64, 112−122.(22) Farrugia, L. J. J. Appl. Crystallogr. 2012, 45, 849−854.(23) Macrae, C. F.; Bruno, I. J.; Chisholm, J. A.; Edgington, P. R.;McCabe, P.; Pidcock, E.; Rodriguez-Monge, L.; Taylor, R.; van deStreek, J.; Wood, P. A. J. Appl. Crystallogr. 2008, 41, 466−470.(24) Spackman, M. A.; Jayatilaka, D. CrystEngComm 2009, 11, 19−32.(25) CrystalExplorer (Version 3.1); Wolff, S. K.; Grimwood, D. J.;McKinnon, J. J.; Turner, M. J.; Jayatilaka, D.; Spackman, M. A.University of Western Australia: Perth, 2012.(26) Becke, A. D. J. Chem. Phys. 1996, 104, 1040−1046.(27) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon,M. S.; DeFrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654−3665.
(28) Civalleri, B.; Zicovich-Wilson, C. M.; Valenzano, L.; Ugliengo,P. CrystEngComm 2008, 10, 405−410.(29) Centore, R.; Causa, M.; Fusco, S.; Carella, A. Cryst. Growth Des.2013, 13, 3255−3260.(30) Grimme, S. J. Comput. Chem. 2004, 25, 1463−1473.(31) Grimme, S. J. Comput. Chem. 2006, 27, 1787−1799.(32) Centore, R.; Causa, M.; Cerciello, F.; Capone, F.; Fusco, S.CrystEngComm 2014, 16, 9168−9175.(33) Dovesi, R.; Orlando, R.; Erba, A.; Zicovich-Wilson, C. M.;Civalleri, B.; Casassa, S.; Maschio, L.; Ferrabone, M.; De La Pierre, M.;D’Arco, P.; Noel, Y.; Causa, M.; Rerat, M.; Kirtman, B. Int. J. QuantumChem. 2014, 114, 1287−1317.(34) Bernstein, J. Polymorphism in Molecular Crystals; ClarendonPress: Oxford, 2002.(35) Bernstein, J. Cryst. Growth Des. 2011, 11, 632−650.(36) It is worthy of great reflection that a structural feature veryminimal in relation to the single molecule, as it is the orientation of thephenolic H atom in the plane of the phenyl ring, has far reachingconsequences in terms of molecular aggregation (polar versuscentrosymmetric packings) and thereby of physicochemical propertiesof the final materials.(37) The geometric parameters of the hydrogen bond D−H···A aregiven, here and throughout the paper, in the following order: D−H(Å), H···A (Å), D···A (Å), D−H···A (deg), symmetry code of theacceptor atom.(38) Actually, in the case of imine 1, there are three differentpolymorphs having the space group Pna21 and very similar packing:they are named I, II, and III after ref 9. Among those three, thepolymorph stable at room temperature is III and it corresponds to thePna21 structures observed in the other imines of Chart 1. Thetransitions I−II and II−III are all single-crystal-to-single-crystal, andthe I−II is also thermosalient, ref 9. Polymorphs I and II have not beenconsidered in the present theoretical analysis.(39) The caveat to this statement is that, in principle, we can never besure that all the crystal polymorphs of a given compound have beenidentified.(40) In ref 9 a transverse molecule was defined as a molecule inwhich the ground state dipole moment is transverse to the head-to-tailmolecular vector.(41) This is the case of imine 2 for which the P21/c polymorph,although not yet discovered, is very likely to exist. There remains thecase of imine 3. However, this is a particular case, because themolecular conformation in the known polar packing Cc (see ref 9) isthe same as found in the centrosymmetric (P21/c and P21/n) packingsof the other imines, so, in this case, based on the data of Table 3, theexpected conformational polymorph should be that in the polarorthorhombic space group Pna21.(42) The chronological development of the research on the imines ofChart 1 may be instructive in this regard. We started with theexperimental observation that all the studied imines only formed polarstructures (ref 9 for imines 1, 2, 3 and 6). After one year, Prof.Naumov (in Abu Dhabi) identified, by chance, the centrosymmetricpolymorph of imine 1 (ref 17), but he admitted (personalcommunication to R.C.) that the reproducibility in the preparationof that polymorph was very poor. Then, we passed to imine 5, forwhich we isolated the expected polar polymorph at first (ref 10). Inorder to collect better X-ray data, we recrystallized imine 5, butsurprisingly and frustratingly, we were no longer able to get the polarpolymorph, but only the (new) centrosymmetric P21/n one. In thecase of 4 (this work), we first identified the centrosymmetricpolymorph, and after, driven by the belief that a polar polymorphshould exist, we found, by recrystallization, the orthorhombic one.Subsequent trials to get again the centrosymmetric monoclinic phasewere (of course!) all pointless. So, it seems that imines 1, 4, and 5provide new examples of Ostwald’s rule of stages and of themysterious phenomenon of disappearing polymorphs (refs 34 and 35).
