Carbenoid-mediated nucleophilic “hydrolysis” of 2-(dichloromethylidene)-1,1,3,3-tetramethylindane with DMSO participation, affording access to one-sidedly overcrowded ketone and bromoalkene descendants^§

Introduction

The 1,1,3,3-tetramethylindan-2-yl(idene) fragments shown in the hydrocarbon parts of formulae 4–8 (Scheme 1) are preferable to the corresponding acyclic di-tert-butylmethylidene moiety (t-Bu₂C in 1–3) as the shielding substituent in static and dynamic model systems for several reasons. (i) With respect to repulsive strain, an attempted protonation of the alkoxide 2 immediately after its generation [2] at −70 °C failed to provide 1, because 2 cyclized too rapidly with formation of the chlorooxirane 3. On the other hand, the somewhat alleviated internal repulsion in alkoxide 4 allowed it to be trapped by protonation below −10 °C before the cyclization could interfere [3], so that the resultant alcohol 5 could be isolated (crude yield 90% from 1,1,3,3-tetramethylindan-2-one) and dehydrated to give 2-(dichloromethylidene)-1,1,3,3-tetramethylindane (6) as the only product (97% yield). (ii) In X-ray diffraction analyses [4-9], the 1,1,3,3-tetramethylindan-2-ylidene parts turned out to be rather rigid, except for an occasional folding along the C-1/C-3 axis, and they did not exhibit the structural disorder problems and vexing angular flexibility which can arise with the t-Bu₂C groups exemplified in Scheme 1. (iii) Depending on the substituents at the exocyclic C-α atom in 7, all four methyl groups in the 1,1,3,3-tetramethylindan-2-ylidene parts [10] and in their truncated analogue 1,1,3,3-tetramethylcyclopent-2-ylidene [11] may be nonequivalent and provide useful stereochemical and stereodynamic NMR information that would not be available from models containing the t-Bu₂C moiety with free rotation about the t-Bu–C bonds. (iv) Vinylic nucleophilic substitution (S_NV) of the chloride anion from 6 by even a very strong nucleophile R–Mt (Mt = alkali metal) to give 7 may appear problematic, because the first intermediate 9 expected with the usual ARE (addition–rotation–elimination) [12] mechanism would suffer from poor stabilization of the negative charge at C-2 which is flanked by two tert-alkyl groups. Instead, the substitution products 7 were obtained from 6 in THF via the cyclic Li,Cl-carbenoid 8 at room temperature (rt) by the carbenoid chain mechanism [3], as indicated in the bottom line of Scheme 1. These S_NV reactions proceed properly because 8 has a reduced (albeit not vanishing) inclination toward cycloalkyne formation through the Fritsch–Buttenberg–Wiechell (FBW) rearrangement [13], whereas S_NV reactions of acyclic Mt,Hal-carbenoids often have to compete with FBW processes forming acyclic alkynes.

The α,α-dibromo analogue of 6 (available [14] from 6 in two steps) was found to undergo corresponding but more intricate carbenoid chain reactions. Therefore, it was planned to prepare α-bromo analogues of 7 from related ketones which should be accessible via the carboxylic acid 10 to be expected from a hydrolysis of 6.

Results and Discussion

“Hydrolysis” of the α,α-dichloroalkene 6

The carboxylic acid 10 (Scheme 2) would normally (and more expediently than before [15]) be accessible through a simple hydrolysis reaction [16,17] of 6 with concentrated (80–100%) sulfuric acid. Our in situ ¹H NMR spectra showed that 6 was insoluble in D₂SO₄ (97%) at rt and that two promising, equally intense methyl singlet signals, as expected for 10, appeared after seven hours at 100 °C. However, the usual (Et₂O/water) work-up procedure did not afford any (acidic or nonacidic) organic product, which suggests that 6 or 10 were converted to water-soluble, unserviceable arenesulfonic acids that disappeared with the aqueous phases.

Without electron-withdrawing substituents in both 6 and the prospective ARE [12] intermediate 9, several strongly caustic methods of hydrolysis with KOH failed to consume 6 in diethylene glycol [18] (11 hours at 135 °C), in triglyme (six hours, 150 °C), in HMPA [tris(dimethylamino)phosphinoxide, five hours at 150 °C], and in acetonitrile (23 hours, 70 °C). Reisolation of pure 6 from hot formic acid (44 hours, 90 °C) or from dimethyl sulfoxide solution (DMSO, five hours at 156 °C without a base) showed that C–Cl bond heterolysis (vinylic S_N1 reaction) did not occur in these polar solvents. In DMSO with potassium tert-butoxide (KOt-Bu, four equiv, >22 hours at 140 °C) or better with solid KOH (at least 29 equiv, ≥ six hours at 100 °C or 60 min at 154 °C), however, 6 was slowly transformed into 10. This may be reminiscent of a 10⁹-fold increased kinetic basicity [19,20] of KOCH₃ in DMSO as a solvent. Due to the strongly enhanced thermodynamic basicity of solid KOH in DMSO [21-23], this system will provide and maintain a small concentration of the potassium salt 11 of DMSO (“dimsyl-K”, Scheme 2). On the other hand, a more special involvement of DMSO as an oxidant in the present “hydrolysis” of 6 became evident when pure dimethyl sulfide (Me₂S, boiling point 37 °C) distilled from the reaction vessel into a cold trap during such a preparation of 10. In situ ¹H NMR spectra revealed the obligatory formation of ca. one equivalent of Me₂S (δ_H = 2.06 ppm). Small portions of Me₂S stemmed from the slow destruction of DMSO by 11, as confirmed in a faster run with 11 alone in DMSO during four hours at 150 °C, and presumably also from generation of the side-product potassium formate as formulated further below. Clearly, Me₂S could not have been formed from 6 in a simple hydrolytic ARE [12] process with an intermediate such as 9 as a precursor of 10. Instead, we propose a carbenoid pathway in Scheme 3 and justify the specified steps in the sequel.

In a Cl/K interchange reaction that generates the Cl,K-carbenoid 12, the transfer of a chlorine particle from 6 to KOH cannot be excluded at the outset in view of the early protocols [24-27] which described the application of dry NaOEt or KOt-Bu at ca. 190 °C for converting aryl₂C=CCl₂ into aryl₂CH–CO₂H along with aryl–C≡C–aryl (the latter in a Fritsch–Buttenberg–Wiechell (FBW) rearrangement [13]). However, dimsyl-K (11) in DMSO without KOH consumed 6 immediately already at rt, albeit without formation of the acid 10. This suggests KOH to be essential for creating product 10; it also suggests that KOH might be a poor competitor of 11 in a nucleophilic attack on 6, perhaps due partially to the low solubility (ca. 0.001 M [21]) of KOH in DMSO. Granting preference to the chlorine transfer from 6 to 11, the byproduct 13 of 12 would hardly be traceable: even if 13 reacted with 11, for example, the resultant water-soluble [28] bis(sulfoxide) 16 might get lost in the usual procedure of aqueous work-up. As an important confirmation of the Cl,K-carbenoid 12, however, its conjugated CH-acid 14 [29] was observed during the early period of a running transformation and vanished slowly with regeneration of 12. This interpretation was corroborated through independent generations of 12 from 14: Using suspensions of KOt-Bu as the base in warm THF or cyclohexane for deprotonating 14, the generated 12 was found [30] to undergo expansion (FBW) of its five-membered ring rather than the intended formation of the acid 10. As two demonstrations that FBW ring expansion is not an inevitable fate of bona-fide 12, we deprotonated 14 also with the freely soluble base KN(SiMe₃)₂ (in place of the hardly soluble KOt-Bu) in t-BuOMe or in THF as the solvents and observed [30] in situ merely the nonexpanded enamine (the expected S_NV product). Alternatively, we generated 12 from the dichloroalkene 6 with a THF solution of benzyl potassium, finding only the carbenoid chain (S_NV) product and its descendants but again no ring expansion [30]. Thus, 12 behaves like a normal unsaturated carbenoid whose S_NV reaction requires a more abundant, soluble nucleophile than solid KOt-Bu (or, by analogy, solid KOH). These observations pointed to a possible S_NV reaction of 12 with DMSO and/or dimsyl-K (11), as proposed in Scheme 3 and later on as follows.

In a rapid oxygen-transfer reaction that is known [31,32] for saturated carbenes or carbenoids, 12 will attack the solvent DMSO to generate the K,O-carbenoid 15 which decays in an E1cb-like expulsion of the observed Me₂S. The resultant, still unknown di-tert-alkyl ketene 18 appears to have only limited options in this milieu: Taking t-Bu₂C=C=O [2] as a model of the slightly less congested ketene 18, a Pummerer-type acylation of DMSO by 18 should require the assistance of a carboxylic acid [33]; this is excluded in our strongly basic system. On the other hand, KOH consumed t-Bu₂C=C=O at rt readily [34]; thus, the certainly faster addition of KOH to 18 can give the observed (in situ ¹H NMR) potassium salt 17 of 10. Of course, [2-D]17 was detected and [2-D]10 isolated [30] (with ca. 80% D, for example) from runs of 6 with KOH in [D₆]DMSO due to the H/D interchange reaction affording KOD; a separate experiment established that [2-D]17 was not formed in a subsequent step from unlabeled 17 with 11 in [D₆]DMSO in the course of 11 hours at 120 °C. In the outlined mechanism, the formal “hydrolysis” of 6 to produce the acid 10 can be seen to consist of an initial reduction of 6 generating the carbenoid 12, followed by the oxidation of 12 by DMSO that affords the precursors of 10.

Although the isolated yield of acid 10 never exceeded 60%, our preparation is convenient because 10 was obtained as the only acid and in a suitably clean state, expecially when the acidified alkaline layers of aqueous work-up were extracted with pentane (in place of Et₂O) which can more efficiently be cleaned from the last traces of DMSO through washing with water. The procedure can also be made colleague-friendly by an early deodorizing of the Me₂S contamination through a short treatment of the aqueous solution of 17 with NaOCl (but not KMnO₄), to be terminated soon through the addition of NaHSO₃. Gratifyingly, all side-products did not contaminate the acid 10 because they assembled in the nonacidic fraction as described in the sequel.

The structure of the main side-product 23 (Scheme 4) suggests that it might result from a base-catalyzed condensation of DMSO with 1,1,3,3-tetramethylindan-2-one (24) [35]; this assumption seemed confirmed through the isolation of 23 (57%) from 24 with KOH (28 equiv) in DMSO after 16 hours at 100 °C. However, the idea that 24 might have been generated from 6 via 20 (and the dichloro alcohol 5) had to be dismissed: Although an authentic [3] sample of 5 was quickly consumed under the above conditions of hydrolyzing 6, it furnished none of 24, 23 or 10. Nevertheless, 24 accumulated in the last period of transforming 6 into 10 (>15 hours at or above 100 °C); such a late appearance of 24 may be caused by a slowly emerging precursor (other than 5). While 24 obviously did not hasten to form 23 with DMSO, an alternative route from 6 with 11 to 23 appears reasonable: the first intermediate 19 would isomerize to 22 and then dissociate into KCHCl₂ (21) and product 23. As a Cl,K-carbenoid, 21 will fall a victim to the above-mentioned type of oxygen-transfer reaction [31,32] with DMSO, via 25 for example, to furnish Me₂S and potassium formate (26) which was actually detected (δ_H = 8.54 ppm) in situ.

The carbenoid chain [3] option has not yet been addressed in the above proposals, because the results outlined in Scheme 3 and Scheme 4 were not compatible with that variant. However, a further side-product 33 was encountered whose structure may reasonably be explained using the carbenoid chain mechanism in the following manner. As a competitor of the Me₂S-producing decay of the K,O-carbenoid 15 to give 18 (Scheme 5), a chlorine particle transfer from 6 to 15 will close a carbenoid chain cycle and give rise to the chain carrier 12 together with the primary product 28. In analogy with the base-induced O–S bond cleavage during the Pummerer acylation with t-Bu₂C=C=O [36], a base-induced decay of 28 can be expected to generate the 2-thioniapropene 29 along with the α-chloroenolate 30 that may produce the ketene 18 (or the unknown acyl choride) and finally (as in Scheme 3) the precursor 17 of acid 10. This alternative route to acid 10 does not create Me₂S and may be a minor contribution, therefore. The expelled cation 29 may be consumed by the liberated chloride anion to give H₃C–S–CH₂Cl (32) or by dimsyl-K (11) or by formation of 33 along the following lines (Scheme 5).

In competition with the abundant DMSO to give 15, dimsyl-K (11) may displace a chloride anion from the Cl,K-carbenoid 12 to generate the vinylpotassium derivative 27. Proton transfer steps should isomerize 27 into the energetically more stable allyl anion derivative 31. Either 29 or 32 may intercept 31 with formation of the side-product 33, whose structure was recognized through its NMR data [30] as follows. The precursor 31 of 33 cannot have reacted at its C-β center, since this would have led to an isomer of 33 with only one proton-bearing olefinic ¹³C center (C-α), whereas two olefinic CH centers (C-α and C-β) were detected in 33 while its C-2 atom was found to be aliphatic rather than olefinic. Although 33 is chiral, its only CH₂ signal exhibits merely weak line-broadening instead of the expected AB splitting pattern; this indicated the center of chirality (sulfur) to be rather distant from CH₂. The NOESY correlation of this quasi-singlet CH₂ signal with the SCH₃ resonance established the presence of a CH₂–S–CH₃ instead of a CH₂–S(O)–CH₃ fragment. The =CH–S(O)–CH₃ group and its proximity to the CH₂ protons were recognized through the NOESY correlations of the olefinic β-proton. This completes the mechanistic considerations for the unusual “hydrolysis” reaction of the unactivated α,α-dichloroalkene 6 and its side-products.

Toward one-sidedly overcrowded ketone and bromoalkene descendants

The α,α-(di-tert-alkyl)methyl aryl ketones 38 (Scheme 6) would normally be considered to be accessible through Friedel–Crafts acylation of arenes by the acid chloride of 10. This possibility was repudiated, however, because the similar acyl chloride t-Bu₂CH–CO–Cl did not acylate benzene [37,38]. Instead, the technique [39] of acylating an aryllithium compound such as 34 by a lithium carboxylate such as 36 to form 35 proved effective in Et₂O as the solvent, in spite of a dissuasive message [40] that was based on the failure with 2,6-dimethylbenzoic acid at ice temperature. Admittedly, our protocol requires prolonged stirring in Et₂O at or above rt due to steric shielding in 36, which enabled at least two yield-reducing side-reactions to consume the aryllithium reagents 34a,b to some extent, as delineated below. Therefore, acid 10 was treated with an excess of 34a,b which were generated from aryl bromides, using either 2.5 equivalents thereof together with tert-butyllithium (t-BuLi, five equiv) or four equivalents of both aryl bromide and n-BuLi. The preformed sodium carboxylate of 10 in place of 36 was apparently too insoluble in Et₂O to react with 34.

The primary products 35 tend to eliminate Li₂O slowly with formation of ketones 38 in the presence of aryllithium reagents 34 which will rapidly add to 38, creating the tertiary alcohols 39. The unwanted side-products 39 might also arise during simple aqueous work-up where hydrolytic ketone formation from primary products such as 35 was often [40,41] able to compete with the fast protolysis of residual aryllithium reagents. To avoid this suspected mischief, we applied nonaqueous quenching of residual 34, using either carboxylation by solid CO₂ or the commended treatments with aniline [41] as the proton donor or with Me₃SiCl [40] prior to the aqueous work-up. However, none of these three precautionary measures improved the 38/39 ratios (ca. 9:1 after ca. 20 hours at rt) significantly, which means that the side-products 39 arose essentially before the work-up procedure. Actually, the unwelcome portions of 39 increased in parallel with extending reaction periods and made especially the isolation of pure ketone 38b somewhat tedious. (An alternative route [30] can provide 38b without side-products.) The overcrowded nature of 39a and 39b became evident [30] through NMR line broadening effects as caused by impeded rotations about the C-α/C-ipso single bonds.

As a second side-reaction, deprotonation of 2-H in 36 by 34 competed with the retarded acylation of 34, so that the dianion 37 of acid 10 was generated and contributed to the recovery of 10 on aqueous work-up. The intermediacy of 37 was recognized through quenching either with D₂O to afford [2-D]10 or with solid CO₂ and subsequent isolation of the highly symmetric diacid 40 (only seven ¹³C NMR signals). The constitution of 40 followed from its ready decarboxylation in CDCl₃ solution at rt to regenerate acid 10 within four days.

Conversion of the ketone 38a into bromoalkene 42a through brominative deoxygenation [42,43] with tribromide 41 was slow in hot chloroform but almost complete within 17 hours at 80 °C in 1,2-dichloroethane as the solvent. The purification of 42a by simple recrystallization was easy, unless the employed ketone 38a was contaminated with its side-product 39a whose p-methyl groups were brominated by reagent 41 with formation of 43 [30]. The pure ketone 38b and reagent 41 furnished the known [44,45] bromoalkene 42b as the only product [30,46].

Conclusion

The observed features of the “hydrolysis” reaction of the unactivated α,α-dichloroalkene 6 with solid KOH in DMSO are incompatible with the key intermediate 9 of the ARE [12] mechanism. Instead, a Cl,K-carbenoid 12 (indicated by the observation of its reversible protonation) is compatible and appropriate as the first intermediate in the route from 6 to acid 10. This formal “hydrolysis” of 6 makes use of the impressively enhanced electrophilicity of “unsaturated” carbenoids [13] such as 12, which property enables 12 to react with the moderately nucleophilic but abundant solvent DMSO that brings along the possibility of an oxygen transfer: The necessary cleavage of the O–S bond in the resultant K,O-carbenoid 15 may occur either directly (the Me₂S producing pathway in Scheme 3) or upon formation of the carbenoid chain product 28 (Scheme 5). Obviously, DMSO is a unique choice for oxidizing 12 in both the chain and the nonchain mechanistic options. The nonoxidizing nucleophiles PhCH₂K and KN(SiMe₃)₂ can convert 12 to the expected nonacidic S_NV products, whereas 12 prefers to expand its five-membered ring if a suitable nucleophile is not available [30].

The easily isolated and purified acid 10 is able to transfer its bulky di-tert-alkylacetyl body onto a sufficiently reactive nucleophile. The first nucleophilic attack is already sterically impeded, as suggested by its ability to compete with the generation of dianion 37 of acid 10 (Scheme 6). Nevertheless, such one-sided shielding in the product ketone does not prevent the incorporation of a second nucleophile, affording a tertiary alcohol with a substantially impeded internal mobility. Final brominative deoxygenation of the ketones can yield bromoalkenes as the only products with moderately decreased speed.

Experimental

1,1,3,3-Tetramethylindan-2-carboxylic acid (10): The following procedure should be carried out in an efficient hood and/or with a cold trap that can collect and retain malodorant volatile products such as Me₂S. The dichloroalkene 6 (2.80 g, 11.0 mmol), dimethyl sulfoxide (45 mL), and a magnetic stirring bar were placed in a round-bottomed flask carrying either a wide-bore connection to a cooled trap or a T-shaped glass-tube with a bubbler and gas inlet. This suspension was stirred at 60 °C until 6 was completely dissolved. Freshly pulverized KOH (18.0 g, 321 mmol) was added and the brown mixture stirred at 100 °C under inert-gas cover for at least six hours, then cooled to rt, poured into distilled water (250 mL), and shaken with Et₂O (3 × 50 mL). The malodorant aqueous layer containing the potassium salt 17 was treated batchwise with NaOCl solution in an amount that sufficed for showing a positive potassium iodide/starch test for 30 min, at which time this residual NaOCl was forthwith destroyed by solid NaHSO₃ that had to be added until the KI/starch test became negative. (KMnO₄ in place of NaOCl was found to decompose 17.) The solution was thoroughly stirred with charcoal, then filtered, cooled in ice, and acidified with concd. hydrochloric acid. The precipitated acid 10 was extracted with Et₂O or (better) pentane (3 × 70 mL). These combined extracts were repeatedly washed with distilled water to remove traces of DMSO, dried over Na₂SO₄, and evaporated to leave the almost pure acid 10 as a white powder (1.31 g, 54%) with mp 188–190 °C (ref [15]: 189–190.5 °C); ¹H NMR (CDCl₃, 400 MHz) δ 1.41 (s, 6H, 1-/3-CH₃ syn to CO₂H), 1.52 (s, 6H, 1-/3-CH₃ anti to CO₂H), 2.91 (s, 1H, 2-H), 7.16 (AA´ part of an AA´BB´ system, 2H, 4-/7-H), 7.24 (BB´ part, 2H, 5-/6-H) ppm, assigned through the NOESY correlations 2-H ↔ anti-CH₃ ↔ syn-CH₃ ↔ 4-/7-H ↔ anti-CH₃; ¹H NMR ([D₆]acetone, 400 MHz) δ 1.35, 1.48, 2.85, 7.12, 7.18 ppm; ¹³C NMR (CDCl₃, 100.6 MHz) δ 27.4 (1-/3-CH₃ syn to CO₂H), 30.2 (1-/3-CH₃ anti to CO₂H), 45.6 (C-1/-3), 64.8 (C-2), 122.3 (C-4/-7), 127.3 (C-5/-6), 149.4 (C-3a/7a), 179.1 (CO₂H) ppm, assigned through HSQC; ¹³C NMR ([D₆]acetone, 100.6 MHz) δ 27.7, 30.5, 45.9, 65.2, 123.1, 128.0, 150.5, 173.4 ppm.

1,1,3,3-Tetramethyl-2-(methylsulfinylmethylidene)indane (23). The combined Et₂O extracts containing the nonacidic side-products, as obtained in the above preparation of 10 and separated from the alkaline aqueous layer, were washed with distilled water until neutral, dried over Na₂SO₄, and evaporated. The remaining brown solid (<977 mg) contained mainly the sulfoxides 23 and 33 (3:1) together with a little of 1,1,3,3-tetramethylindan-2-one (24). The pure sample of 23 (284 mg, 10%) was isolated through extraction into hot, low-boiling petroleum ether (60 mL), filtration, and concentration. Recrystallization afforded almost colorless needles with mp 134–135 °C; ¹H NMR (CDCl₃, 400 MHz) δ 1.41 and 1.45 (2 s, 2 × 3H, 2 × 3-CH₃), 1.51 and 1.73 (2 s, 2 × 3H, 2 × 1-CH₃), 2.67 (s, 3H, OS-CH₃), 6.25 (s, 1H, α-H), 7.17 (AA´ part of an AA´BB´ system, 2H, 4-/7-H), 7.27 (BB´ part, 2H, 5-/6-H) ppm, assigned through comparison with the phenylsulfinylmethylidene analogue [47]; ¹H NMR (DMSO, 200 MHz) δ 1.38 (2 × 3-CH₃), 1.46 and 1.63 (2 × 1-CH₃), 6.47 (α-H), 7.24 (quasi-s, 4-/5-/6-/7-H); ¹³C NMR (CDCl₃, 100.6 MHz) δ 30.12 and 31.94 (2 × 3-CH₃), 31.54 and 32.95 (2 × 1-CH₃), 40.73 (OS–CH₃), 48.54 (C-3), 48.76 (C-1), 122.41 and 122.51 (C-4/-7), 127.70 and 127.82 (C-5/-6), 128.78 (C-α), 147.04 (C-3a), 148.23 (C-7a), 174.13 (C-2) ppm, assigned as above in accord with the deuterium-induced shifts ²Δ = –0.068 for C-2, ³Δ = −0.021 for C-3, and ³Δ = −0.045 ppm for C-1 as caused by the =C^αD–S(O)–CD₃ group incorporated during a run in [D₆]DMSO; IR (KBr) ν: 2971, 2959, 2920, 2860, 1627 (w), 1484, 1363, 1032, 1022, 968, 748, 677, 505 cm⁻¹; anal. calcd for C₁₅H₂₀OS (248.39): C, 72.53; H, 8.12; S, 12.91; found: C, 72.75; H, 8.25; S, 12.94.

2-(p-Methylbenzoyl)-1,1,3,3-tetramethylindane (38a): A round-bottomed Schlenk flask (50 mL) was charged with 4-bromotoluene (0.563 mL, 4.58 mmol), anhydrous Et₂O (10 mL), and a magnetic stirring bar. The contents were stirred and cooled at −78 °C under argon gas cover during the dropwise addition of t-BuLi (9.16 mmol) in pentane (6.10 mL), then stirred without cooling for 30 min. After the dropwise addition of acid 10 (400 mg, 1.83 mmol) in anhydrous Et₂O (10 mL) to this solution of p-methylphenyllithium (34a) and further stirring at rt for 18 hours, the mixture was poured onto solid CO₂, warmed up, and diluted with aqueous NaOH (1 M, 20 mL). The aqueous layer was shaken with Et₂O (3 × 20 mL) and the combined four Et₂O layers were washed with distilled water until neutral, dried over Na₂SO₄, and concentrated to leave the crude nonacidic material (455 mg) consisting mainly of 38a, 39a, and toluene (9:1:9). Repeated crystallizations from pentane afforded white needles of 38a (isolated yield up to 35%); mp 95.5–96.5 °C (methanol); ¹H NMR (CDCl₃, 400 MHz) δ 1.34 and 1.39 (2 s, 2 × 6H, 2 × 1-/3-CH₃), 2.42 (s, 3H, p-CH₃), 4.10 (s, 1H, 2-H), 7.17 and 7.24 (AA´BB´ system, 2 × 2H, 4-/5-/6-/7-H), 7.28 (broadened d, ³J = 8.3 Hz, 2H, 2 × m-H), 7.88 (dm, ³J = 8.3 Hz, 2H, 2 × o-H) ppm; ¹H NMR (CCl₄, 80 MHz) δ 1.29, 1.35, 2.39, 3.99, 7.06 (s, 4H), 7.18 (d), 7.80 (d) ppm; ¹³C NMR (CDCl₃, 100.6 MHz) δ 21.56 (p-CH₃), 28.02 and 30.98 (2 × 1-/3-CH₃), 47.17 (C-1/-3), 64.32 (C-2), 122.34 (C-4/-7), 127.12 (C-5/-6), 128.28 and 129.34 (2 × C-m and 2 × C-o), 138.36 (C-ipso), 143.50 (C-p), 149.99 (C-3a/7a), 201.25 (C=O) ppm; IR (KBr) ν: 2967, 2925, 2862, 1664 (s), 1606, 1480, 1368, 1228, 1209, 1186, 869, 759 cm⁻¹; anal. calcd for C₂₁H₂₄O (292.42): C, 86.26; H, 8.27; found: C, 86.50; H, 8.41.

The above aqueous NaOH layer was acidified with concd. hydrochloric acid and was shaken with Et₂O (3 × 10 mL). These combined Et₂O extracts were washed with distilled water until neutral, dried over Na₂SO₄, and evaporated to leave a white powder containing acid 10, p-methylbenzoic acid, p-cresol, and diacid 40 (ca. 4:2:2:1).

1,1,3,3-Tetramethylindan-2,2-dicarboxylic acid (40): The mixture of acids obtained above (see 38a) was leached with pentane, which left the insoluble diacid 40 behind (65 mg, 14% yield). One recrystallization from hot toluene afforded clean 40 (30 mg) but led to the decarboxylation of a portion that remained dissolved. The transparent needles of pure 40 had a mp of 195–197.5 °C (dec.), decomposed slowly on standing at rt in CDCl₃ solution, and were weakly soluble in CCl₄ only as long as they were a part of the original mixture with the other carboxylic acids. ¹H NMR ([D₆]acetone or CDCl₃, 400 MHz) δ 1.58 (s, 12H, 2 × 1-/3-CH₃), 7.21 (quasi-s, 4H, 4-/5-/6-/7-H) ppm; ¹³C NMR ([D₆]acetone, 100.6 MHz) δ 28.83 (2 × 1-/3-CH₃), 48.98 (C-1/-3), 122.28 (C-4/-7), 125.79 (C-2), 127.59 (C-5/-6), 150.30 (C-3a/7a), 171.73 (2 × CO₂H) ppm, assigned through comparison with the acid 10; IR (KBr) ν: 3400–2500 (very broad H–O), 1713 (s), 1291, 759 cm⁻¹; anal. calcd for C₁₅H₁₈O₄ (262.3): C, 68.69; H, 6.92; found: C, 69.23; H, 6.81.

40 was not formed from acid 10 with lithium N,N-diisopropylamide in THF as the solvent at rt: final quenching with solid CO₂ afforded only starting material 10.

2-(α-Bromo-p-methylbenzylidene)-1,1,3,3-tetramethylindane (42a): A dry NMR tube (5 mm) was charged under argon cover gas with 2-bromo-1,3,2-benzodioxaphosphole [48] (0.12 mL, 0.95 mmol) and anhydrous, EtOH-free 1,2-dichloroethane (0.5 mL). The solution was cooled in ice, treated with elemental bromine (0.040 mL, 0.78 mmol), and kept at rt for 15 min. This yellow solution of 2,2,2-tribromo-2,2-dihydro-1,3,2-benzodioxaphosphole (41) [49] was recooled in ice for the addition of the pure, solid ketone 38a (200 mg, 0.68 mmol), which caused an immediate color change to brown. The tightly stoppered tube was heated at 80 °C for at least 17 hours, then diluted with pentane (5 mL), and poured onto ice-cooled aqueous Na₂CO₃ solution (2 M, 5 mL), which was stirred for 15 min. The aqueous Na₂CO₃ layer was shaken with pentane (2 × 5 mL) and discarded. The combined pentane extracts were shaken with aqueous NaOH (2 M, 3 × 5 mL), washed with distilled water (10 mL), dried over CaCl₂, and evaporated to furnish the solid nonacidic fraction (256 mg) containing a little residual ketone 38a. Recrystallization from ethanol (12 mL) yielded pure needles of 42a (173 mg, 71%), mp 119.5–121 °C; ¹H NMR (CDCl₃, 400 MHz) δ 1.19 (s, 6H, 2 × 1-CH₃), 1.78 (s, 6H, 2 × 3-CH₃), 2.38 (s, 3H, p-CH₃), 7.03 (dm, ³J = 7.5 Hz, 1H, 7-H), 7.16 and 7.22 (quasi-AB system, ³J = 8 Hz, 2 × 2H, 2 × o-H and 2 × m-H of tolyl), 7.18–7.27 (m, 3H, 4-/5-/6-H) ppm, assigned through comparison with 42b [45]; ¹H NMR (CCl₄, 80 MHz) δ 1.17, 1.76, 2.38, 7.14 ppm; ¹³C NMR (CDCl₃, 100.6 MHz) δ 21.30 (p-CH₃), 28.05 (2 × 3-CH₃), 31.19 (2 ×1-CH₃), 50.55 (C-1/-3), 117.95 (C-α), 122.10 (C-4), 122.59 (C-7), 127.16 (C-5), 127.24 (C-6), 128.53 (2 × C-m), 129.61 (2 × C-o), 137.66 (C-p), 139.93 (C-ipso), 149.00 (C-3a), 149.86 (C-7a), 157.11 (C-2) ppm, assigned as above; IR (KBr) ν: 2987, 2959, 2924, 2865, 1505, 1487, 1456, 1363, 799, 757 cm⁻¹; anal. calcd for C₂₁H₂₃Br (355.32): C, 70.99; H, 6.52; Br, 22.49; found: C, 70.88, H, 6.66; Br, 23.17.