摘要:
Intermolecular interactions of the halogen atom, namely halogen⋯halogen and D–H⋯halogen (halogen⊕=⊕F, Cl, Br; D⊕=⊕O, N, C), continue to attract significant attention in chemical, crystallographic and crystal engineering literature. In a recent communication,1Orpen, Brammer and coworkers have analysed the geometry of O–H and N–H donors to Cl–M (M⊕=⊕transition metal), Cl−(chloride ion) and Cl–C (organic chlorine) acceptor groups. Their crystallographic analysis shows that the anisotropy at acceptor chlorine (M–Cl⋯H⊕=⊕90–140°) and the large number of short H⋯Cl contacts (<2.5 Å) with Cl–M and Cl−are significantly diminished or absent with the Cl–C moiety. The picture that emerges from recentstudies1,2is that metalbound chlorine and fluorine as well as halide ions are good hydrogen bond acceptors with O–H and N–H donors. On the other hand, the exact nature of C–H⋯Cl interactions is still not clearly understood. They have been observed in some structures,3aascribed to determine the molecular conformation3band also the arrangement of molecules and ions in the crystal.3cPolymorphism and twinning in crystals of 1,2,4,5tetrabromobenzene have been explained through Br⋯Br and Br⋯H interactions.4Resolution of racemic 1,2dibromohexafluoropropane with (–)sparteine hydrobromide occurs through robust halogen bond helices in the cocrystal.5Intermolecular interactions involving halogen atomsneed not always be structure determining. Because of their weak nature, a balance of forces and interplay of halogen and hydrogen bonding drive selfassembly during crystallisation when strong hydrogen bonding groups, such as OH, are present.6Statistical analysis of crystallography databases is well suited for the study of weak halogen⋯halogen and D–H⋯halogen interactions.7C–H⋯Cl interactions are even weaker compared with the weak O–H⋯Cl and N–H⋯Cl hydrogen bonds because the donor atom is much less electronegative.8Yet, the phrase ‘C–H⋯Cl hydrogen bond’ has appeared in the title of many recent papers,3b,3c,9,10of which one is in this journal.9bWhile there appears to be a general consensus that the weak O–H⋯Cl and N–H⋯Cl interactions are hydrogen bonds1,2the nature of even weaker C–H⋯Cl interactions is still not fully understood – are they hydrogen bonds or mere van der Waals interactions? Aakeröy,Seddon and coworkers10have argued that the traditional van der Waals cutoff be dropped for the analysis of C–H⋯Cl interactions and instead be replaced by a softer distance–angle criterion, determined empirically. We show herein through a crystallographic survey that the weak C–H⋯Cl−and C–H⋯Cl–M interactions exhibit the characteristics of conventional hydrogen bonds and hence could be significant in molecular recognition and crystal engineering. On the other hand, the less polarisable C–H⋯Cl–C interaction has no specific directionality and is mostly a van der Waals contact. These trends are in agreement with an earlier study10and are corroborated in subsets of data with activated donor atoms, when bifurcation is removed, and when competing donor/acceptor groups are excluded.The Cambridge Structural Database11(CSD, version 5.19, April 2000 update, 215 403 entries) was searched for C–H⋯Cl contacts in the range 2.0⊕<⊕H⋯Cl⊕(d/Å)⊕<⊕3.3, 90⊕<⊕C–H⋯Cl⊕(&thetas;/°)⊕<⊕180 from three different subdatabases of the Cl acceptor moiety: Cl−(chloride ion, 1858 hits), Cl–M (metalbound chlorine, M⊕=⊕transition metal, 11 861 hits) and Cl–C (organic chlorine 5304 hits). The C–H distance was neutronnormalised to 1.083 Å. Liberal distance (ΣvdWH 1.20 Å and Cl 1.75 Å⊕+⊕10%⊕=⊕3.3 Å) and angle (bent geometry up to 90°) cutoff criteria wereused. Screens 33 (errorfree), 35 (no disorder), 85 (chemical/crystallographic connectivity match), 88 (Rfactor⊕≤⊕0.10), and 153 (atom coordinates present) were applied. Only organic compounds were considered (screen 57) with Cl–C and Cl−acceptors; screen 57 was not applied with Cl–M. Duplicate refcodes were not removed. Bifurcated hits were not removed except in one case of C–H⋯Cl–C contacts to compare distance–angle scatter plots of single and bifurcated motifs. Thed–&thetas;plot, with or without bifurcation, did not show much difference. However, by removing bifurcated motifs the number of hits dropped to a mere 10% of when bifurcation was included. Moreover, bifurcation will perturb all types of contacts uniformly and so the chemical conclusions should not be altered in a comparative study. In order thatthe number of fragments is about the same and also statistically significant in different distributions, theRfactor was varied such that the scatter plot or histogram containedca.1000 fragments. TheRfactor cutoff for each search is mentioned in the figure caption.d–&thetas;Plots of hydrogen bond distribution are displayed using VISTA, a visualisation program distributed with the CSD package. Distance and spatial distributions were normalised:12x⊕=⊕(RH–Cl)3whereRH–Cl⊕=⊕d⊕/⊕2.95 (ΣvdW);y⊕=⊕&thetas;norm⊕=⊕1⊕−⊕cos (180⊕−⊕&thetas;). The corrected (RH–Cl)3vs.&thetas;normplots are justified when&thetas;is close to 180°. Since many C–H⋯Cl contacts deviate from linearity and the conventionald–&thetas;plots are a more familiar representation to chemists, they are used in this paper.Fig. 1shows thed–&thetas;scatter plot for C–H⋯Cl interactions from any Catom (sp3, sp2, sp) to Cl–M, Cl−and Cl–C acceptors. The inverse length–angle correlation characteristic of a hydrogen bond with significant electrostatic contribution8that is shortd–linear&thetas;and longd–bent&thetas;is clear inFig. 1(a)and1(b). There are a few contacts in the offdiagonal region (longd, linear&thetas;), but this is not surprising because soft, weak interactions have a variable geometry in crystals. InFig. 1(c), however, the van der Waals region (upper right) is heavily populated suggestingthat in the C–H⋯Cl–C interaction there is very little electrostatic component. A comparison of histograms inFig. 2is instructive. We define a C–H⋯Cl contact as either short, medium or long using the criteria <2.6 Å (d⊕≪⊕ΣvdW), 2.6–3.0 Å (d⊕≤⊕ΣvdW), and >3.0 Å (d⊕>⊕ΣvdW), respectively. The proportion of short and medium contacts – that is, those below the van der Waals radius sum – decreases as the acceptor changes from Cl−(56/1059 and 574/1059; 5.3% and 54.2%) to Cl–M (35/1424 and 745/1424; 2.5% and 52.3%) to Cl–C (1/1395 and 495/1395; <0.1% and 35.5%). This is not surprising becausethe strength of a hydrogen bond increases not only with donor acidity but also with acceptor basicity.13For the C–H⋯Cl–C contact [Fig. 3(c)] there is only one interaction with a short geometry (refcode BOJJOZ, 2.54 Å, 160.5°).Distance–angle (d–&thetas;) scatter plot: (a) C–H⋯Cl–M,Rfactor⊕≤⊕0.02; (b) C–H⋯Cl−,Rfactor⊕≤⊕0.03; (c) C–H⋯Cl–C,Rfactor⊕≤⊕0.03. Notice that the upper-right van der Waals region is more populated in scatter plot (c) compared with (a) and (b).Distance histogram ofFig. 1: (a) C–H⋯Cl–M; (b) C–H⋯Cl−; (c) C–H⋯Cl–C. Contacts⊕<⊕2.6 Å are shaded black (≪ΣvdW), 2.6–3.0 Å grey (≤ΣvdW) and >3.0 Å white (>ΣvdW). Notice that there are a few to more short contacts in (a) and (b) but only one in (c).d–&thetas;Scatter plot: (a) (sp2)C–H⋯Cl–M,Rfactor⊕≤⊕0.02; (b) (sp2)C–H⋯Cl−,Rfactor⊕≤⊕0.035; (c) (sp2)C–H⋯Cl–C,Rfactor⊕≤⊕0.035. Compare withFig. 1.Subsets of data with activated acidic C–H donors were considered next.Table 1shows H⋯Cl distances with different categories of donors, namely CHCl3, (sp)C–H, CH2Cl2, (CH3)2S&z.dbd6;O,etc.The shortness ofdfor more acidic donors is apparent with Cl–M and Cl−acceptors (entries 1 and 2vs.5 and 6) compared with Cl–C where the difference is negligible (within esd). The correlation of hydrogen bond distance with C–H pKashows that C–H⋯Cl–M and C–H⋯Cl−interactions have a significant electrostatic component whereas C–H⋯Cl–C is an isotropic interaction that is insensitive to donor acidity. Thed–&thetas;scatter plots of H⋯Cl contacts with (sp2)C–Hdonors are displayed inFig. 3.Mean H⋯Cl distance,d(Å), with activated C–H donors to Cl−, Cl–M and Cl–C acceptor moieties. The number of observations is given in square bracketsEntryDonorCl−Cl–MCl–C1CHCl32.38(3) [16]2.66(3) [170]2.98(5) [16]2(sp)C–H2.56(3) [16]2.82(13) [6]3.07(6) [7]3CH2Cl22.64(4) [25]2.86(1) [683]3.06(2) [51]4CH3CN2.89(8) [10]2.95(1) [654]3.03(4) [17]5(CH3)2S&z.dbd6;O2.81(3) [15]2.98(1) [559]3.05(5) [8]6(CH3)2C&z.dbd6;O3.04(4) [3]3.03(2) [94]2.87 [1]Lastly, the C–H⋯Cl–C interaction was analysed in the absence of interfering and competing effects – strong OH and NH groups were excluded, electronegative atoms in the vicinity of Cl were removed, and only single interactions were considered – so that even a very weak hydrogen bond type character will be manifested. These results are displayed inFig. 4. There is no significant improvement in thed–&thetas;distribution when interference from OH and NH groups is removed [Fig. 4(a)] or when competing electronegative O, N, Cl atoms in the vicinity of the Cl acceptor are excluded up to a radius of 4.0 Å [Fig. 4(b)]. Refcodes with a bifurcated motif were removed from the subdatabase of 2854 (sp2)C–H⋯Cl–C contacts with theRfactor⊕≤⊕0.10 toprovide the scatter plot ofFig. 4(c)with 299 contacts. Notwithstanding that the donor is moderately activated and that these are single interactions, there is little improvement in the quality of thed–&thetas;plot suggesting that C–H⋯Cl–C is a van der Waals interaction. In a study of the weak C–H⋯F–C interactions, similar CSD distributions were observed and the inversed–&thetas;correlation without the offdiagonal ‘noise’ could be obtained only with a small subset of carefully selected fluorobenzene crystal structures.14d–&thetas;Scatter plot of C–H⋯Cl–C interaction: (a) all C–H donors, no OH and NH groups present,Rfactor⊕≤⊕0.03; (b) all C–H donors, N, O and Cl atoms excluded up to 4 Å distance from acceptor Cl atom,Rfactor⊕≤⊕0.05; (c) (sp2)C–H donors, single contacts (hits with bifurcation removed),Rfactor⊕≤⊕0.10. Notice that these distributions are qualitatively not very different fromFig. 1(c)and3(c), although the number of points is different in each plot.It is difficult to establish conclusively the nature of halogen⋯halogen and X–H⋯halogen interactions from this database study – are they hydrogen bonds with electrostatic and polarisation contributions or are they mere space-filling van der Waals interactions? These issues have been the subject of intense debate for about two decades now.15Our analysis of C–H⋯Cl interactions parallels the observations with stronger OH and NH donors,1keeping in mind that as the interaction becomes weaker the distribution is more diffuse.10We note that acidic C–H donors form shorter contacts with electronegative Cl–M and Cl−acceptors. Thus, the structural significance ascribed to C–H⋯Cl–M hydrogen bonds in recent crystallographic studies (M⊕=⊕Au, Ti, Sb)3,9isconfirmed by this database analysis. A statistical validation for weak hydrogen bonds is necessary because they are subject to crystal forces and exhibit a smear of distance–angle features. On the other hand, C–H⋯Cl–C interactions with organic chlorine are more a result of van der Waals close packing. There is little improvement in the scatter plot correlation coefficient in subsets of selected crystal structures. To summarise, the electrostatic nature of C–H⋯Cl–M and C–H⋯Cl−hydrogen bonds is revealed through chemical probes whereas C–H⋯Cl–C interactions continue to exhibit isotropic behaviour even after competing and interfering effects are removed. Even so, the contribution of C–H⋯Cl interactions towards selfassembly in a crystal structure must be assessed by a combination of the following parameters: (1) the distance–angle characteristics; (2) the natureof the Cl acceptor; and (3) the presence of other donors and acceptors.
ISSN:1466-8033
DOI:10.1039/b102780h
出版商:RSC
年代:2001
数据来源: RSC