Unusual traits of cis and trans-2,3-dibromo-1,1-dimethylindane on the way from 1,1-dimethylindene to 2-bromo-, 3-bromo-, and 2,3-dibromo-1,1-dimethylindene

Do not rely on the widely accepted rule that vicinal, sp3-positioned protons in cyclopentene moieties should always have more positive 3J NMR coupling constants for the cis than for the trans arrangement: Unrecognized exceptions might misguide one to wrong stereochemical assignments and thence to erroneous mechanistic conclusions. We show here that two structurally innocent-looking 2,3-dibromo-1,1-dimethylindanes violate the rule by means of their values of 3J(cis) = 6.1 Hz and 3J(trans) = 8.4 Hz. The stereoselective formation of the trans diastereomer from 1,1-dimethylindene was improved with the tribromide anion (Br3−) as the brominating agent in place of elemental bromine; the ensuing, regiospecific HBr elimination afforded 3-bromo-1,1-dimethylindene. The addition of elemental bromine to the latter compound, followed by thermal HBr elimination, furnished 2,3-dibromo-1,1-dimethylindene, whose Br/Li interchange reaction, precipitation, and subsequent protolysis yielded only 2-bromo-1,1-dimethylindene.


Results and Discussion
The addition of elemental bromine to 1,1-dimethylindene (2, see Supporting Information File 1) in CCl 4 as the solvent afforded a 7:3 mixture of trans-1 and cis-1 (Scheme 1). A more useful 9:1 mixture was obtained through the slow titration of a well-stirred chloroform solution of equimolar amounts of 2 and tetraethylammonium bromide with elemental bromine in chloroform. Such a higher trans selectivity is typical of the very rapidly [1] formed tribromide anion (Br 3 − ) as the reactive species.
What kind of evidence supports our stereochemical assignments of trans-1 and cis-1? Although the three-bond NMR coupling constant 3  ) may obviously be ascribed to the two 1-Me groups [9]. In accord with the pseudoaxial C(3)-H bond, HMBC (hetero multiple bond correlation) cross peaks of 3-H in trans-1 were absent with both C-1 and C-7a. On the other hand, the corresponding cross peaks were strong in cis-1 between 3eq-H and both C-1 and C-7a, which indicated a significant 3 J C,H NMR coupling via the intervening single bonds in their roughly coplanar arrangement shown in Scheme 1. It may also be noticed that the small C(2)-H/C(3)-H torsional angles in either one of the two cis-1 conformations are of similar sizes and do not permit a conformational differentiation. On the other hand, the close to 90°t orsional angle between 3-H and 2-H in the 2eq-H,3eq-H conformation of trans-1 would imply an almost vanishing 3 J H,H value, in contrast with the observed value of 8.4 Hz that is explained by the predominant 2ax-H,3ax-H conformation.
Distillation of the trans/cis product mixture led to some enrichment of the thermally more stable diastereomer trans-1 due to the "base-free" HBr elimination from cis-1 with formation of 2-bromo-1,1-dimethylindene (4 in Scheme 2) [10]. We observed a less distinct kinetic preference in the corresponding base-induced processes: In di(2-methoxyethyl) ether (diglyme) as the solvent, a substoichiometric amount of KOt-Bu (potassium tert-butoxide) reacted faster with cis-1 than with trans-1 by a factor of roughly 9 at room temperature (rt). With an excess (>2 equiv) of KOt-Bu, both of these weakly exothermic HBr elimination reactions were completed within less than 30 min. With KOt-Bu (at rt) or KOEt (at 50 °C) but not with NEt 3 (no reaction at rt), the exclusive formation of 3-bromo-1,1-dimethylindene (3) from trans-1 and of 4 from cis-1 became evident through the similarity of the emerging 3/4 ratios as compared with the trans-1/cis-1 ratios in the employed mixtures [11]. This regiospecificity is readily understandable since trans-1 has no anti relationship of vicinal Br and H available (Scheme 1), whereas each cis-1 conformer holds a pseudodiaxial, vicinal Br/H anti relationship and hence is (presumably) able to react somewhat faster.
Information File 1), which decomposed already on distillation to yield 1,1-dimethylindene (2). With two equivalents of NBS, however, 9 afforded a mixture of 3 and 7 through the following sequence in the bottom part of Scheme 2: Under these conditions, the spontaneous HBr elimination from 8 had produced 3, which added Br 2 to generate the thermolabile tribromide 5, whose HBr elimination gave 7; the required Br 2 was visible in the weakly brownish gas phase and had been provided through the well-known reaction of HBr with the accompanying NBS. Consequently, three equivalents of NBS would be necessary for obtaining 7 in a maximum yield. This encouraged us to reflux 9 in CCl 4 with NBS (4 equiv), which furnished mainly 7 along with succinimide (3.6 equiv) and remnant NBS (0.4 equiv). As a side-reaction, the slower thermal HBr elimination from the intermediate 3-bromo-1,1-dimethylindane (see Supporting Information File 1) generated 1,1-dimethylindene (2), whose in situ bromination furnished 1 (ca. 1%) in a trans/cis ratio of ca. 3:2. Since both dibromides trans-1 and cis-1 were stable under the reaction conditions and would distil together with 7, they were destroyed through HBr elimination by KOt-Bu (or KOH in EtOH) to produce the monobromides 3 and 4. It may be noticed that 4 (from cis-1) cannot have been an intermediate in the initial step of the above "base-free" conversion of 9 to 7, since 4 would generate the thermally stable tribromide 6 that was not detected in the initial product mixture of 1 and 7.
For practical purposes, 7 may be useful as an alternative starting material in place of 2,3-diiodo-1,1-dimethylindene that had been employed [13,14] in cross-coupling studies. We actually used crude 7 as follows for a first specific route to 2-bromo-1,1dimethylindene (4). The rapid Br/Li interchange reaction of 7 in hexane as the solvent with n-butyllithium (n-BuLi) ocurred predominantly at the 3-position of 7 with formation of 2-bromo-3lithio-1,1-dimethylindene (10, Scheme 3). In the absence of cycloalkanes or benzene from the hydrocarbon solvent, rather concentrated mixtures of 7 and n-BuLi slowly deposited unsolvated 10, which opened the possibility of purifying 10 through simple washings with dry pentane under inert gas cover. Like the related 3-chloro-2-lithio-1,1-dimethylindene [15], 10 did not eliminate LiHal at rt, so that its final hydrolytic work-up provided clean 4 even from moderately contaminated 7. Due to a well-known mixing problem [16,17], this final protolysis will be successful only in the absence (or at least after an adequate washing-out) of residual n-BuLi: Since protonation of n-BuLi and 10 is comparably rapid, a local depletion of the added proton source will leave the generated portion of 4 together with remnant n-BuLi, so that a very rapid Br/Li interchange reaction of 4 with n-BuLi will produce 1,1-dimethyl-2-lithioindene, whose protolysis forms 1,1-dimethylindene (2). The alternative (incorrect) stereoassignment ("cis-19" in reference [8] and "cis-33" in reference [7] for the presently analyzed trans-1) would have misguided us to claim erroneously that we discovered a most unusual, highly cis selective olefin bromination.
What else may appear unusual with 1 and its congeners? Trans-1 and cis-1 undergo regiospecific HBr eliminations even in the absence of bases, cis-1 does so more rapidly than trans-1.

Experimental
General remark. 1 H and 13 C NMR chemical shifts δ (ppm) were referenced to internal tetramethylsilane.

Supporting Information
Supporting Information File 1