Technical Comments

Comment on “Single-Crystal X-ray Structure of 1,3-Dimethylcyclobutadiene by Confinement in a Crystalline Matrix”

See allHide authors and affiliations

Science  19 Nov 2010:
Vol. 330, Issue 6007, pp. 1047
DOI: 10.1126/science.1195752


Legrand et al. (Reports, 16 July 2010, p. 299) reported on the photolytic reaction of an α-pyrone confined in a crystalline matrix. Their structural analysis invoked four products: activated precursor, isomeric Dewar β-lactone, and square and rectangular isomers of 1,3-dimethylcyclobutadiene. The reported x-ray data, however, suggest that all observed structures correspond to only one distinct species, the Dewar β-lactone.

The report by Legrand et al. (1) on the structure of 1,3-dimethylcyclobutadiene (Me2CBD) in a crystalline guanidinium-sulfonate-calixarene (G4C) matrix caught my attention, given a long-standing interest in phenomena associated with aromaticity and anti-aromaticity (25). The observation of the rectangular minimum and the square-planar transition state (or triplet state) of the highly reactive Me2CBD as two distinguishable species inside the G4C host promised major insights regarding the structure and bonding of cyclobutadiene, the archetypal anti-aromatic compound. Despite a recent report on bond length equalization in anti-aromatic pentaphenylborole (6), it was surprising that Me2CBD in its square-planar form, that is, the bond-length–equalized triplet state [(7)], should be stable. Even more intriguingly, it was reported to coexist in close vicinity with a strongly bent CO2 molecule, which had been retained in the G4C host. Although hydrogen bonding to the guanidinium cations of G4C may possibly affect the geometry of the CO2 moiety, this seemed a rather remote possibility: In HF-CO2 adducts, for instance, the CO2 unit does not deviate from linearity (8). I therefore analyzed the x-ray diffraction studies reported by Legrand et al. in more depth.

Legrand et al. (1) interpreted the solid-state structures obtained after irradiation of a single crystal of 4,6-dimethyl-α-pyrone (Me21) in a G4C host (all species referred to hereafter are confined in G4C) as due to the rearrangement of Me21 via activated Me21′ to Dewar β-lactone Me23. Further irradiation was claimed to result in decarboxylation of Me23 to rectangular Me2CBDR and square-planar Me2CBDS (see Scheme 1). The relatively small carbon-oxygen distances between the CO2 moiety and the CBD unit from 1.502 to 1.612 Å were attributed to van der Waals interactions even though they are fully in line with typical covalent single bonds. In particular, β-lactones exhibit C–O bonds of up to 1.553 Å (9). Notably, Legrand et al. do not discuss the even shorter distance of 1.41 Å between the carbon atoms of CO2 and Me2CBDR after prolonged irradiation. I show here that the reported data are consistent with the photolytic formation and enantiomerization of Me23 (10) without the need to invoke any other species (Scheme 1).

Scheme 1

Reaction proposed by Legrand et al. (1) (top) and reinterpretation presented here (bottom). The host, guanidinium-sulfonate-calixarene (G4C), is not displayed.

For the data obtained after the first irradiation of Me21 (Fig. 1, A and B), Legrand et al. (1) applied a split model between Me21 (50%) and a species referred to as Me21′ (50%), a more or less satisfactory description of the structure. Conversely, the split model including Me2CBDR-1 (22.7%) and Me23 (77.3%) used for the data set obtained after a second period of irradiation (Fig. 1, C and D) is problematic. Atom C3 of Me23 features a very large cigar-shaped ellipsoid that can only be due to two unmodeled positions of this atom. Its orientation toward the plane of the remaining ring atoms suggests that the use of a three-way split model including Me21 could have resolved the issue. Therefore, the long C3–C6 distance in Me23 (Fig. 2) is an artifact of this unresolved disorder involving Me21, which has no C3–C6 bond at all. Because the main distinguishing feature between Me21′ and Me23 is the C3–C6 bond length, it can be concluded that they both represent the Dewar β-lactone isomer. The shortening of the C2–C(3) and C3–C4 bonds of Me23 is a typical librational effect of a large orthogonal displacement of an atom (cigar-shaped thermal ellipsoid) and thus an artifact.

Fig. 1

Thermal ellipsoid plots of Me21 (A), Me21′ (B), Me2CBDR-1 (C), Me23 (D), Me2CBDR-2 (E), and Me2CBDS (F). Molecules are rotated to show topological similarities.

Fig. 2

Distances between endocyclic carbon atoms [data from CSD 764866 to 764868 (1)]. Error bars, 4 SD.

There is also little topological difference between Me2CBDR-1 and Me21′. The diagram in Fig. 2 shows barely significant deviations between the allegedly different compounds (<0.2 Å). To put these deviations into perspective, the two structures both assigned as Me2CBDR by Legrand et al. show C2–C3 bonds differing by 0.19 Å. Within the accuracy limits of the model, Me2CBDR-1 can therefore be regarded as the enantiomer of Me21′ (or Me23).

The authors’ interpretation of the third data set raises similar issues (Fig. 1, E and F). In this case, they applied a split model using Me2CBDR-2 (62.7%) and Me2CBDS (37.3%). Inspection of Me2CBDS confirms that it is almost identical to Me21′ (and thus with Me23), now unperturbed due to the absence of the completely consumed starting material Me21. Following the previous reasoning, Me2CBDR-2 is a somewhat distorted mirror image of Me21′ (and Me23) as well.

Finally, the bond lengths of Me2CBDS shown in figure 4 in (1) are inconsistent with those deposited at the Cambridge Crystallographic Data Centre (CCSD 764868). Aside from the fact that it is impossible to determine reliable bond lengths involving atom positions that are shared by two disordered molecules (C4 and C6), the bond lengths of Me2CBDS (Fig. 2) are not identical, but rather vary from 1.423 to 1.493 Å. The geometry of Me2CBDS is thus much closer to a rectangular arrangement than Legrand et al. (1) imply. Perhaps the strongest argument, however, is that the CO2 moiety is hardly perturbed by the alleged changes of the CBD molecules. The carbon-oxygen bond lengths in all structures range from 1.319 to 1.347 Å for C2–O1 and from 1.163 to 1.178 Å for C2–O7, comparing favorably with the single (1.378 Å) and double bond (1.118 Å) of a bicyclic β-lactone closely related to Me23 (11).

In conclusion, Legrand et al. (1) are encouraged to thoroughly reinvestigate their crystallographic studies, maybe at lower temperatures. It can be anticipated that the formation of Dewar β-lactone, Me23, and its photolytically produced enantiomer will thus be confirmed rather than the presence of any cyclobutadiene derivative.

References and Notes

  1. Financial support by the Aventis Foundation (Karl Winnacker fellowship) is gratefully acknowledged. I thank A. Berndt and W. Massa (Universität Marburg, Germany) for helpful discussions.
View Abstract

Navigate This Article