A stable enol from a 6-substituted benzanthrone and its unexpected behaviour under acidic conditions

Treatment of benzanthrone (1) with biphenyl-2-yl lithium leads to the surprisingly stable enol 4, which is converted by dehydrogenation into the benzanthrone derivative 7. Under acidic conditions 4 isomerises to the spiro compound 11 and the bicyclo[4.3.1]decane derivative 12. Furthermore, the formation of 7 and the hydrogenated compound 13 is observed. A mechanism for the formation of the reaction products is proposed and supported by DFT calculations.


Introduction
Compounds for optoelectronic applications with electroluminescent (e.g. organic light-emitting diodes, OLEDs) or lightharvesting properties (e.g. organic solar cells) are receiving more and more attention [1]. In this respect benzanthrone (1), with its luminescent and photosensitizing properties, is an interesting candidate for the construction of these systems. Recently, aminobenzanthrone derivatives have been shown to be efficient emitters for OLED applications [2]. In these devices, the benzanthrone moiety acts as an electron accepting group, whereas the diarylamine group functions as an electron donor.
The reaction of 1 with various organometallic reagents was studied by Allen in the 1970s [3]. It was shown that an attack of phenylmagnesium chloride or phenyl sodium after 1,4-addition leads to the 6-substituted benzanthrone derivative 3 (Scheme 1). On changing the solvent from ether-benzene to tetrahydrofuran the ketone could also be isolated in high yields, but additionally a labile enol was produced that was hard to separate. To this compound, obviously an intermediate in the addition process, the authors assigned structure 2, a compound that under the reaction condition is dehydrogenated to 3.  Here, we present the first isolable enol derived from a benzanthrone and the unexpected behaviour of this adduct under acidic conditions.

Results and Discussion
Syntheses Benzanthrone (1) was treated with biphenyl-2-yl lithium (Scheme 1). After work-up and chromatography the surprisingly stable enol 4 was obtained in 56% yield. However, no formation of the tertiary alcohol 5 could be observed, a compound type which is produced (derivative 6) when benzan-throne was treated with phenyl lithium [3]. The yield of 4 was not improved by addition of a copper(I) salt in catalytic amounts [4]. This procedure should have favoured the ratio of a 1,4-to a 1,2-addition product [5].
The enol 4 is stable as a solid and also in deuterated dimethyl sulfoxide, since an NMR solution in this solvent was unchanged after one week. In contrast, a solution of 4 in chloroform showed quantitative conversion to the 6-substituted benzanthrone 7 after approximately one week; a process that was subsequently monitored by 1 H NMR spectroscopy ( Figure 1) in Scheme 2: Proposed mechanism for the formation of 4 and its oxidation to 7. CDCl 3 . This conversion is much slower when the CDCl 3 is filtered through an alumina plug before use. The reaction constitutes a formal dehydrogenation of 4 (Scheme 2).
As shown in Scheme 2, we propose that the formation of 4 and 7 starts as a 1,4-addition process as discussed above via the enolate 8 as an intermediate. From this, the enol 4 is generated under the influence of the added acid. Further protonation provides the oxonium ion 9 which is set up for a retro-[2+4]cycloaddition (see transition state 10) to lose hydrogen and finally become deprotonated to yield the isolated 7. Since at this stage of our study we were not interested in mechanistic investigations we did not look for the production of hydrogen. Considering the small amount of substrate we were working with (0.7 mM concn of 4) and the slow process of the conversion, it is not surprising that we could not see any gas formation (hydrogen bubbles). However, what makes this rationalisation attractive is the production of both an aromatic system as well as a carbonyl group, so the process is thermodynamically favourable. Furthermore, the formation of quinomethides from ortho-substituted phenols is a well known phenomenon in mass spectrometry (the "ortho-effect" see [6]). Next, the enol 4 was treated deliberately under acidic conditions by heating it with phosphoric acid in toluene under reflux. Silica gel was added to the two-phase mixture in order to effect a better contact between the layers. The progress of the reaction was monitored by TLC, which indicated that, surprisingly, three different new compounds were produced besides 7 (21% yield). After workup and chromatography, these new products were identified as 11, 12, and 13 (Scheme 3). The ketone 7 itself is stable under these reaction conditions. Spiro compound 11 (11% yield) was characterised by NMR spectroscopy, mass spectrometry and single crystal X-ray crystallography (see below). The 1 H NMR spectrum (600 MHz) of 11 shows two aliphatic triplets at δ = 2.16 and 3.42 ppm (J = 6.2 Hz) which are assigned to the four methylene protons.
In the 13 C NMR spectrum (151 MHz) the corresponding carbon atoms cause signals at 28.4 and 37.3 ppm, respectively. The spiro carbon atom is represented by a singlet at 53.4 ppm. All other spectroscopic data correspond to expectations and are recorded in the Experimental section.   Compound 13 was characterised by NMR spectroscopy, mass spectrometry and by single crystal X-ray crystallography (see below). The mass spectrum of compound 13 shows a signal with m/z = 386, which exceeds the molar mass of the starting material 4 by 2 Da. The ethylene moiety is represented in two groups of multiplets in the 1 H NMR spectrum (400 MHz; δ = 1.68-1.77 and 2.79-2.84 ppm).
The formation of these three new compounds can be explained as follows. For the production of 11 and 12 we propose the mechanism summarised in Scheme 4 [8].
Both 11 and 12 have the same molecular mass as the starting material 4, so the processes leading to these two products are isomerisations. The protonation that initiates the rearrangements can take place at C-4 or C-5 of the starting material 4. In the former case the secondary cation 14 results, which by proton loss is converted into hydrocarbon 15; in other words, 4 has undergone an acid-catalyzed allylic rearrangement. Renewed protonation leads to the tertiary cation 16, which by an internal Friedel-Crafts alkylation provides the spiro compound 11. Alternatively, protonation of 4 at C-5 generates the benzylic cation 17, which by intramolecular electrophilic attack leads to the bicyclo[4.3.1]decane derivative 12.
Finally, the formation of 13 is a formal hydrogenation of the starting material 4. In the absence of a catalytically active layer that promotes a hydrogen-transfer reduction [9,10], we propose an acid-catalysed hydride transfer of the type reported by Carlson and Hill [11]. Thereby, a carbenium ion such as 14, 16 or 17 (only the case of 16 is discussed in the following) can abstract hydride from another molecule that itself forms a stable cation (Scheme 5).

Scheme 5:
Proposed mechanism for the formation of 13.
In order to make the above mechanistic speculations more than simple "electron pushing", we decided to apply the following computational methods.

Reaction mechanisms by computational methods
The gas phase global minima of the relevant molecules 4, 7 and 9-18 were obtained by first applying an extended conformational analysis using the OPLS2005 force field [12] together with a Monte Carlo torsional sampling as implemented in the Macromodel 9.5 program [13]. Each lowest energy conformation of 4, 7 and 9-18, respectively, was then optimised by applying density functional theory. The M05-2X hybrid functional [14] was employed, and all atoms were described by a standard triple zeta all electron basis set augmented with one set of polarization functions (6-311G(d,p)). After the relevant stationary points were localised on the energy surface, they were further characterised as minima states by normal mode analysis based on the analytical energy second derivatives. Enthalpic and entropic contributions were estimated from the partition functions calculated at room temperature (298 K) under a pressure of 1 atm using Boltzmann thermostatistics and the rigid rotor harmonic oscillator approximation as implemented in the Gaussian03 set of programs [15]. Table 1 summarises the reaction energies/enthalpies of the different reaction steps. Although all calculations were carried out in the gas phase, it can be assumed that solvation effects will not counterbalance such high energetic differences.
The proposed reaction pathway from 4 to 7 proceeds via a protonation of the OH group to form 9, releasing In order to explain the formation of the bicyclo[4.3.1]decane derivative 12 the enol 4 is protonated first at C-5 which releases 212.87 kcal mol −1 . The benzyl cation thus generated can undergo a similar electrophilic substitution to produce the final product 12 at an effort of 206.32 kcal mol −1 , resulting in an overall exothermicity of −6.55 kcal mol −1 for the reaction 4 → 12.