Attempts to prepare an all-carbon indigoid system

Introduction

Cross-conjugated organic molecules are defined as unsaturated systems containing two π-electron systems (or lone pairs) that are in direct conjugation, whereby a third such system is excluded from interaction [3]. Typical examples are 2-vinyl-buta-1,3-diene ([3]dendralene, 3-methylene-penta-1,4-diene), benzophenone or urea.

Whereas the hydrocarbon parent systems, the [n]dendralenes, have long been a neglected class of oligoenes [4], the recent preparative accomplishments of the Sherburn group have changed the situation fundamentally [5,6]. These cross-conjugated hydrocarbons are now known up to [13]dendralene, and many of these potentially very valuable compounds are available in gram quantities, allowing, often for the first time, comprehensive chemical studies [6].

Notwithstanding modern progress, the phenomenon of cross-conjugation is as old as (scientific) organic chemistry. Many of the organic compounds that played an important role in the dawn of (industrial) organic chemistry are cross-conjugated systems or were converted into cross-conjugated organic salts during the color-generating process.

This is illustrated by indigo (1, Scheme 1) and its derivatives (e.g., Tyrian purple) and the triphenylmethane-derived carbocations.

Starting from the generalized indigoid structure 2, the role of the heteroatoms X and Y can be played not only by O and NH as in 1 itself, but by many other equivalent atoms and/or structural elements. In fact, the anellated benzene rings could be omitted altogether, resulting in what Lüttke, a pioneer in this area of dyestuff chemistry, has called urindigo (3, primordial indigo): representing the basic cross-conjugated π-electron system of the indigoid compounds [7-9].

Whereas the replacement of Y in 2 (X = O) by sulfur and selenium [10,11] or even tellurium [12,13] has been known for some time, derivatives of “phosphaindigo” (Y = various substituted phosphorus derivatives) have only been described more recently [14].

As far as we are aware, however, no attempt to prepare an all-carbon equivalent of indigo has ever been described. A system that could qualify as such an all-carbon analog of 1 is the bis-anion 5, which itself should be obtainable by anionization of the linearly conjugated triene 4.

Although we have to accept for the time being that our different approaches to preparing 4 (and 5) have so far been unsuccessful (see below), we think that our initial efforts to attain this goal are worth publication. Furthermore, we are convinced that 5 will eventually become available.

Results and Discussion

Our first attempt to prepare hydrocarbon 4 started from indene (6, Scheme 2). Epoxidation with m-chloroperbenzoic acid (MCPBA) according to a literature method [15] yielded the epoxide 7 in meager yields (Scheme 2).

The methylation of 7 to 8 was achieved by the treatment with trimethylsulfonium iodide in the presence of n-butyllithium in THF [16].

The spectroscopic data of these intermediates are incomplete in the chemical literature and are hence given in full detail in the experimental section (see Supporting Information File 1).

Unfortunately, all attempts to oxidize 8 to the α-methylene ketone 12, failed (Dess–Martin reaction, IBX, Swern oxidation etc.); the original plan was to dimerize this intermediate to 4 by, e.g., a McMurry coupling reaction. Also the second route, starting with the reaction of 2-indanone (9) with the Eschenmoser salt 10 according to [17] to give the iodide 11, was unsuccessful, since the attempted Hofmann elimination to 12 failed. Of course 12, if formed, might not have survived the isolation process.

Since the target molecule 12 was expected to be a reactive compound, we decided to increase its stability by the introduction of two methyl substituents at its exocyclic double bond. Indeed, when 9 was first metalated with LDA and the resulting enolate quenched with acetone, the resulting ketol could be dehydrated in situ by treatment with diethyl chlorophosphate to yield 13 [18].

Derivative 13 is a crystalline solid that can be kept in the refrigerator for longer periods of time without decomposition. Slow evaporation of the solvent of a chloroform solution provided single crystals of 13 that were suitable for X-ray structure analysis. The resulting structure is shown in Figure 1.

Compound 13 crystallizes with two independent but closely similar molecules (rmsd 0.05 Å) in the asymmetric unit; these occupy independent alternating layers parallel to (). The ring systems are almost exactly planar (mean deviations <0.03 Å) with the atoms of the isopropylidene group lying approximately 0.2 Å to one side of the plane and the keto oxygen approximately 0.2 Å to the other side. The exocyclic C=C bonds are slightly lengthened, at 1.352(3), 1.353(3) Å.

Subjecting 13 to a McMurry coupling reaction (TiCl₄/HgCl₂ in THF) did not result in the formation of the expected dimer 14, a tetramethyl derivative of 4, but furnished dimer 17 in low yield (10%, Scheme 3).

To rationalize its formation we propose that the deoxygenation of 13 does indeed take place, but provides a carbenoid intermediate 15 rather than the vicinal diol complex that is usually postulated to be formed during the McMurry dimerization. Possibly the dimerization of the substrate molecule cannot take place because of steric hindrance by the gem-dimethyl group. Intermediate 15 is a vinylcarbene that, in principle, has several options to react further. It could dimerize to the intended product 14 (or its diastereomer), cyclize to a cyclopropene derivative or react via its resonance structure 16 to the isolated dimer 17. Clearly, among these alternatives, the last route is preferred. In 17 the two benzene rings are anellated in anti-orientation, i.e., as far apart as possible. There exists an alternative structure, however, in which the two aromatic rings point in the same direction. To distinguish between these two possibilities based on spectroscopic evidence alone would not be easy. Fortunately, however, single crystals of the isolated dimer, suitable for X-ray analysis, were obtained, showing unambiguously that the anti-orientation is preferred (Figure 2).

The molecule of 17 exhibits crystallographic inversion symmetry, but the true symmetry is close to C_2h (rmsd 0.02 Å). The ring system is planar, with a mean deviation of only 0.02 Å. The molecular packing involves a herringbone pattern in layers perpendicular to the a-axis. There are no noticeably short intermolecular interactions.

Our next approach towards 4 (or a derivative) started from indan-1-one (18, Scheme 4).

Enolization of 18 was carried out with sodium hydride in THF and on subsequent oxidative dimerization the expected bis-ketone 19 was obtained. This compound is produced as a mixture of two diastereomers in roughly 1:1 ratio (NMR analysis), which can be separated by column chromatography. Since, however, in the next step the connecting single bond is transformed into a double bond, we used the diastereomeric mixture for the subsequent oxidation. This was carried out by metalating 19 with sodium hydride again, and then oxidizing the presumably resulting carbanion to 21 (a known compound [19]) with Cu(OTf)₂, which gave cleaner results than the chloride employed previously.

Unfortunately, all efforts to convert 21 into 4 failed. Thus neither the Wittig reaction of 21 (MeP(Ph)₃Br/n-BuLi, THF) nor its treatment with Oshima–Lombardo reagent (Zn, CH₂Br₂, TiCl₄) [20,21] yielded a trace of 4.

Likewise, the exposure of 21 to Tebbe’s reagent (trimethylaluminum with titanocene dichloride, pyridine, toluene) led to no sign of reaction in the desired sense. Nor was the tetramethyl derivative of 4, hydrocarbon 14, obtained when 21 was subjected to a crossed McMurry coupling with excess acetone. In a stepwise approach, 21 was treated with either methyllithium in ether or methylmagnesium bromide in the hope of either preparing a mono- or the bis-tertiary alcohol derivative 20, which subsequently could be subjected to a dehydration reaction. In both cases only minute amounts of products could be obtained. When trying to purify these by column chromatography on silica gel, the stationary phase turned blue, but no defined products could be isolated.

Since it might have been the central conjugated butenedione core that caused all these preparative difficulties, we next decided to investigate the behavior of diketone 19, in which this conjugation is interrupted. Although its methylenation under Wittig conditions was again unsuccessful, the reaction with the Tebbe’s reagent provided a product. This, however, was not the hoped-for diene 22, but an isomer, the hydrocarbon 23 (Scheme 5).

The structure of hydrocarbon 23 was established by the usual spectroscopic data (see experimental section) and also by an X-ray crystal structure determination (Figure 3).

The molecule of 23 is planar (mean deviation 0.03 Å) and displays crystallographic inversion symmetry, although the true symmetry is close to C_2h (rmsd 0.03 Å). The central C–C bond seems slightly short at 1.520(2) Å, but a similar value of 1.523(5) Å was observed in the only other known system with two analogous five-membered rings joined by a –CH₂–CH₂– moiety [22,23]. The molecular packing is devoid of striking features.

As far as the mode of formation of 23 is concerned, the isomerization formally is a [3.3]sigmatropic rearrangement (Cope rearrangement). Since the rearrangement of structurally similar compounds [24], including the parent system hexa-1,5-diene, requires much higher temperatures than those given in Scheme 5, we assume that the metalorganic reagent (or products derived therefrom) play a role in the isomerization.

Hydrocarbon 23 is, in fact, a known compound. It has been prepared previously from indene (6) by other routes [25,26], but always in the form of mixtures also containing other isomers. As far as we are aware, the above route is the only one that provides isomerically pure 23, not surprising in view of its route of formation. These isomeric hydrocarbons are useful ligands for the preparation of bridged Ziegler–Natta catalysts employed for olefin polymerization [26,27].

Since compound 23 contains doubly activated methylene positions, it should be easy to alkylate or bis-alkylate it. Furthermore, use of a bis-electrophile such as 1,2-dibromoethane could lead to the [2.2]indenophane 26 or its isomer 27, both potential ligands for the preparation of novel metallocene derivatives and also so far unknown.

The other conceivable spiro isomer 28 seems to be a less likely product since its E-configurated double bond (within a seven-membered ring) should result in considerable strain (anti-Bredt hydrocarbon).

When 23 was treated with tert-BuLi in THF and the presumably resulting bis-anion quenched with 1,2-dibromoethane, a complex product mixture consisting largely of polymeric material was formed. From this mixture, however, trace amounts of two hydrocarbons (total yield 5%) could be separated by thin layer chromatography. Further (column) chromatographic attempts to obtain preparative amounts of the two isomers in analytically pure form failed, however. The spectroscopic and analytical data of this hydrocarbon mixture (see experimental part) agreed with the two structural proposals 24 and 25 shown in Scheme 5. An unambiguous structure determination had to await X-ray crystallographic analysis. The required single crystals were obtained from the different chromatographic fractions (which contained either isomer in enriched form only) by recrystallization from chloroform and dichloromethane/chloroform, respectively. As can be seen from the X-ray structures, the former is the anti-spiro compound 24 (Figure 4) and the latter its syn-isomer 25 (Figure 5), with the former possessing the longer chromatographic retention time. Unfortunately, because of lack of material, no high quality NMR spectra of these two hydrocarbons could be obtained.

The molecule 24 displays crystallographic inversion symmetry, although the true symmetry is close to C_2h (rmsd 0.02 Å). The molecular packing involves herringbone layers, parallel to the bc plane, in which two C–H···π contacts to the centroid of the aromatic six-membered ring are observed (H5···π 2.76 Å, H9A···π 2.83 Å). The structure determination of 25 was of limited accuracy because of twinning and disorder problems (indeed, there may be a small amount of contamination by 24) and we therefore do not discuss it in detail. Both independent molecules display non-crystallographic mirror symmetry (rmsd 0.04 Å).

Our final attempt for the present to prepare a precursor hydrocarbon for an all-carbon indigoid system rests on the observation that the Thiele condensation of indene (6) with acetone in pyrrolidine/methanol yields the benzofulvene 29 (Scheme 6).

A suitable methylene component for this type of condensation would be the tetra-substituted olefin 2-(2-indanylidene)indane (30), in which each methylene group is activated both by a benzene ring and a double bond. We prepared this known olefin [28] by a McMurry-type dimerization of 2-indanone (9). The structure determination by the usual spectroscopic methods was straightforward and the obtained spectra agreed with the reported data [28]. The assignment was corroborated by a single-crystal X-ray structure determination (Figure 6); single crystals were obtained by slow evaporation of a pentane solution.

The molecule of 30 is planar (mean deviation 0.02 Å), but the true symmetry is close to D_2h (rmsd 0.02 Å). The central C=C bond length is 1.327(2) Å. The molecular packing involves herringbone layers, parallel to the plane (102), in which C-H···π contacts from a methylene hydrogen to the centroid of the six-membered ring are observed (H1B···π 2.63 Å).

Unfortunately, all our condensation experiments with 30, under conditions that were successful for 6, failed and neither did we obtain the mono- (31) nor the (desired) bis-condensation product 14.

Conclusion

The preparation of a suitable precursor for the first all-carbon analog of an indigoid dyestuff has failed so far. It is conceivable that unsubstituted α-methylene ketones such as 12 are too reactive to be isolated under the reaction conditions. On the other hand, if the methylene group is protected by two methyl groups, the corresponding ketone, 13, becomes isolable, but fails to participate in a McMurry dimerization, possibly because of steric hindrance. If the double bond between the indane-derived “halves” in, e.g., 4 is replaced by a single bond (as in 22), a Cope reaction is the most favorable process yielding the ethano-bridged bis-indenyl derivative 23, which can be employed for the introduction of additional linker units, but is of no use for the preparation of the intended target molecule 4.