Unexpected isolation of a cyclohexenone derivative

1 Introduction

The bicyclic diketone 1 exhibits electronic interactions between the two non-conjugated carbonyl groups as Knott and Mellor [1] have shown by UV/VIS spectroscopy.

This effect is also evident from the EPR spectrum of the corresponding radical anion 1˙⁻. We determined a complete set of proton hyperfine structure (hfs) coupling constants [a1H=0.4830 mT (8 H, 2 CH₃, H¹, H⁵), a2H=0.0416 mT (2 H, H³, H⁷), a3H=0.0167 mT (2 H, H⁹, H^9′)] [2] from an EPR spectrum which was better resolved compared with our earlier result [3]. This clearly demonstrates the spin density to be distributed over the whole molecule irrespective of the lack of conjugation between the two moieties; i.e. intramolecular electron transfer (“spin jumping”) takes place. To determine the proton hfs coupling constants in the complex multi-line EPR spectrum of the asymmetric thioketone radical anion 2˙⁻ [2] more precisely, and assign the coupling constants more convincingly, we wanted to prepare a related diketone and the corresponding thioxoketone with tert-butyl substituents instead of the methyl groups present in 1 and 2, which would simplify the EPR spectra.

2 Results and discussion

Knoevenagel [4] was the first to prepare the diketone 1 by base-catalyzed condensation of acetylacetone with formaldehyde and subsequent acid-catalyzed intramolecular aldol condensation of the intermediate tetraketone 3. Knott and Mellor later on [1] confirmed Knoevenagel’s results by NMR spectroscopy and identified the cyclohexenone 4 as an intermediate. Furthermore, Suzuki et al. [5] showed by careful interpretation of the NMR spectra that 4 predominantly exists as the tautomeric enol 5.

To replace the methyl groups of 1 by tert-butyl substituents, we performed a Koevenagel reaction of formaldehyde with pivaloylacetone (5,5-dimethylhexane-2,4-dione) in the presence of piperidine as a base and obtained the tetraketone 6 as a mixture of the racemic (6a) and the meso-diastereoisomer (6b) which we separated by chromatography. Their IR and NMR spectra are very similar but clearly different. An assignment of the spectra to the two isomers was, however, not possible.

Fortunately, one of the two diastereoisomers was crystalline. X-ray structure determination on single crystals allowed us to assign the meso-configuration 6b to this isomer (see the molecular structure in Fig. 1).

Fig. 1:

Ortep plot of the meso-4,6-diacetyl-2,2,8,8-tetramethylnonane-3,7-dione (6b) molecule. Displacement ellipsoids are drawn at the 50% probability level, H atoms as spheres with arbitrary radii.

Unexpectedly, a corresponding enol tautomer was not detected in the NMR spectra of the product mixture. Accordingly, all C=O bonds and C(O)–C bond lengths of the crystalline 6b clearly correspond to double and single bonds, respectively, with bond angles at the carbonyl centers of ca. 120° (cf. Table 1). The product mixture contained, however, 3-ethoxymethyl-5,5-dimethylhexane-2,4-dione (7) as a by-product.

Attempts to prepare the bicyclic diketone 8 by a condensation reaction of 6 were unsuccessful. Only the starting compound 6 was recovered under catalysis with diethyl amine in contrast to the reported formation of 1 from 3 [1].

A reaction took however place when the crude product obtained from pivaloylacetone and formaldehyde was directly treated with an excess of polyphosphoric acid (PPA) without isolation and purification of the intermediate 6. A low yield of only one pure condensation product was isolated by column chromatography. Further components of the complex mixture could not be identified. But, according to the parent peak at m/z=278 (calculated for C₁₇H₂₆O₃: m/z=278.4) in its mass spectrum, this product was not the desired bicyclic diketone 8 (C₁₇H₂₄O₂: m/z=260.4).

It exhibits three different carbonyl groups according to the IR spectrum. Its constitution could only be completely asymmetric as it showed seven different ¹H NMR and 13 different ¹³C NMR signals. Fortunately we succeeded to grow a single crystal suitable for X-ray diffraction. It showed the compound to be racemic (4S*,6R*)-3-methyl-4,6-dipivaloylcyclohex-2-enone (9) (Fig. 2).

Fig. 2:

Ortep plot of the molecular structure of (4S*,6R*)-3-methyl-4,6-dipivaloylcyclohex-2-enone (9) in the crystal. Displacement ellipsoids are drawn at the 50% probability level, H atoms as spheres with arbitrary radii.

The cis-orientation of the two pivaloyl substituents is more obvious from Fig. 3 showing clearly the half-chair of the cyclohexene ring from a different point of view. We could not detect the corresponding trans-diastereoisomer in the product mixture.

Fig. 3:

Side view of the molecular structure of 9 in the crystal (Ortep; displacement ellipsoids are drawn at the 50% probability level, H atoms were omitted for clarity).

Selected bond parameters of 6b and 9 are collected in Tables 1 and 2.

Obviously, steric hindrance prevents a twofold condensation reaction between the acetyl and the pivaloyl groups of the intermediate tetraketone 6 which adopts a different orientation of the substituents compared with 3. The condensation then occurs only between the two acetyl groups and ends up with the monocyclic triketone 9 instead of the bicyclic diketone 8 (see Scheme 1).

Scheme 1

Formation of the monocyclic 9 instead of the bicyclic 8.

3 Experimental section

4 Crystal structure determinations

The crystal data of 6b and 9 and a summary of experimental details are given in Table 2. The structure of 6b was solved and refined using the programs Shelxt [8] and Shelxl [9]. The structure of 9 was solved with Multan [10] and completed with differential Fourier syntheses [11]. Structure refinement of 9 was performed by least-squares methods [12]. The crystallographic data of 6b and 9 have been deposited [13].