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§

Summary 2-(Dichloromethylidene)-1,1,3,3-tetramethylindane was “hydrolyzed” by solid KOH in DMSO as the solvent at ≥100 °C through an initial chlorine particle transfer to give a Cl,K-carbenoid. This short-lived intermediate disclosed its occurrence through a reversible proton transfer which competed with an oxygen transfer from DMSO that created dimethyl sulfide. The presumably resultant transitory ketene incorporated KOH to afford the potassium salt of 1,1,3,3-tetramethylindan-2-carboxylic acid (the product of a formal hydrolysis). The lithium salt of this key acid is able to acylate aryllithium compounds, furnishing one-sidedly overcrowded ketones along with the corresponding tertiary alcohols. The latter side-products (ca. 10%) were formed against a substantially increasing repulsive resistance, as testified through the diminished rotational mobility of their aryl groups. As a less troublesome further side-product, the dianion of the above key acid was recognized through carboxylation which afforded 1,1,3,3-tetramethylindan-2,2-dicarboxylic acid. Brominative deoxygenation of the ketones furnished two one-sidedly overcrowded bromoalkenes. Some presently relevant properties of the above Cl,K-carbenoid are provided in Supporting Information File 1.


An alternative route to ketone 38b
Epoxidation of the known [S1] olefin S1 with peracetic acid furnished the oxirane S2 (Scheme S1) whose chiral nature became immediately evident through the 1 H NMR nonequivalence of all four methyl groups. We did not succeed in the following two methods of ring-opening nucleophilic bromination of S2 with the intention to arrive at the bromoalkene 42b. Treatment with Li 2 NiBr 4 in THF at rt [S2] led to the quantitative recovery of S2, whereas ring-opening by acetyl bromide plus Et 4 N + Br − in CHCl 3 or by pyridinium bromide in 1,2-dichloroethane at 100 °C furnished 1,1,2,3tetramethylindene (S3) and benzaldehyde; this provided further examples of the imminent rearrangement [S3] in the 1,1,3,3-tetramethylindane system. Fortunately, deprotonation at C-2´ of S2 by n-BuLi occurred readily in THF as the solvent at rt (but very slowly in hexane with a first half-reaction time of >50 hours). The resultant lithium enolate S4 was trapped with ClSiMe 3 to give the -OSiMe 3 derivative S5.
Alternatively, the same procedure of generating S4 but trapping by protonation afforded the ketone 38b. Scheme S1: An alternative synthesis of ketone 38b.

1,1,3,3-Tetramethylspiro[2´-phenyloxirane-3´,2-indane] (S2): [S4]
A suspension of the olefin S1 (1.80 g, 6.86 mmol) [S1], suspended in glacial acetic acid (8 mL), was stirred at rt during the slow addition of peracetic acid (freshly prepared from H 2 O 2 with acetanhydride). This batchwise addition was continued until the olefin S1 had dissolved completely and a peroxide test (KI/starch) remained positive, which required several hours. The mixture was diluted with water (40 mL) and stirred for four hours to hydrolyze residual acetanhydride, then rinsed with more water (80 mL) and hours with exclusion of air and moisture, the yellow mixture containing a white precipitate was poured onto solid CO 2 , warmed up, and dissolved in aqueous NaOH (1 M, 20 mL). The aqueous layer was shaken with Et 2 O (3 × 20 mL) and the combined four Et 2 O layers were washed with distilled water until neutral, dried over Na 2 SO 4 , and concentrated. This crude, non-acidic fraction (481 mg) contained the ketone 38b, n-butylbenzene, and the alcohol 39b in a 9:4:1 ratio. (The pure ketone 38b was prepared by the alternative route described above.) The above aqueous NaOH layer was cooled in ice and acidified with conc. hydrochloric acid (white precipitate), then shaken with Et 2 O (3 × 20 mL). These latter Et 2 O extracts were combined and washed with distilled water until neutral, dried over Na 2 SO 4 , and evaporated to yield the acidic product fraction (222 mg) containing 10, benzoic acid, and diacid 40 in roughly equal amounts.  -(1,1,3,3-tetramethylindan-2-yl) the signals of the enantiotopic nuclei (p-CH 3 , C-p, C-ipso, and p-CH 3 ) did not split.
This established a restricted mobility at C- with impeded rotation about the C-/Cipso single bonds and implies that the formation of 39a and 39b was retarded by a substantially increasing repulsive resistance.

FBW ring expansion of 12
Potassium tert-butoxide (KOt-Bu), but not LiOt-Bu, is a sufficiently active base to deprotonate the monochloride 14 (Scheme S2) slowly at 70 °C. The resultant Cl,Kcarbenoid 12 did not add to cyclohexene in THF as the solvent and was not trapped (at least not irreversibly) by di-tert-butyl ketone (t-Bu 2 C=O) in cyclohexane. In both of these solvents, 12 expanded its five-membered ring to generate the hitherto unknown cycloalkyne S6 in analogy with the earlier [S9] examples involving unsaturated Br,K-and Cl,Li-carbenoids. A run in heptane as the solvent (50 hours at 90 °C) provided evidence for the unknown enolate S7 through quenching with solid CO 2 , which furnished the -ketoacid S9 along with the known [S10] ketone S8. The chiral constitution of S9 followed from its 1 H NMR spectrum in CDCl 3 which exhibited four nonequivalent methyl groups ( H = 1.29, 1.49, 1.55, and 1.61 ppm) and for the center of chirality a one-proton signal ( H = 3.98 ppm). The thermal lability of S9 prevented its isolation and further characterization: The complete decarboxylation in CDCl 3 solution within four days at rt afforded the ketone S8, whose constitution [S10] established the ring expansion of carbenoid 12. Because tert-butyl ethers deriving from 12 or S6 were never detected, it remains unknown whether the enolate S7 arose through hydration of S6 by adventitious moisture or through a base-induced decay of a tert-butyl ether derived from S6.

The S N V reaction of benzyl potassium (S11) with carbenoid 12
The easily prepared (Scheme S3) and purified [S11] benzyl potassium (S11) guides the dichloroalkene 6 into and through an efficient carbenoid chain reaction: S11 acts not only as a chlorine acceptor in step 1 and as a nucleophile in step 2; it is also consumed by the coproduct PhCH 2 Cl of step 1 (giving dibenzyl), by the primary chain S12 product S13 to produce the allene S14, and by deprotonating S14 to afford S15 which was recognized through carboxylation that generated the acid S16.  6 12 S15 S16 S12 S13 S14 S13 Scheme S3: Carbenoid chain S N V of the Cl,K-carbenoid 12.

No ring expansion during deprotonation and S N V of monochloride 14 by the potassium amide KHMDS
It remained to demonstrate that a less reactive potassium nucleophile than S11, provided that it is more soluble than KOt-Bu, is also able to perform the S N V reaction before the ring expansion becomes perceptible. We observed by in situ 1 H NMR spectroscopy how mixtures of KN(SiMe 3 ) 2 (KHMDS, potassium 1,1,1,3,3,3-hexamethyldisilazide) and HN(SiM 3 ) 2 (HMDS, 1,1,1,3,3,3-hexamethyldisilazane) [S13] deprotonated the monochloride 14 (Scheme S4). The consumption of 14 required six days at rt in toluene with t-BuOMe (6:5 vol/vol) but only three hours in THF (where LiHMDS did not react over days at rt). In t-BuOMe as the solvent, 14 (0.07 M) was consumed by KHMDS (0.13 M) and HMDS (0.14 M) at rt with a first half-reaction time of ca. seven hours. In all cases, the enamine S18 was the main (and the only identified) product; ring expansion generating the cycloalkyne S6 did not take place.
Final evidence for the S N V reaction was obtained through hydrolysis of the crude material containing the enamine S18 which furnished the known [S14] aldehyde S19 as the only descendant.

KN(SiMe 3 ) 2 as a base and nucleophile:
A weighed, dry NMR tube (5 mm) was charged with potassium hydride in mineral oil (49 mg) and pentane (0.3 mL). The suspension was whirled up through gentle shaking, and the turbid supernatant was withdrawn by syringe from the heavy precipitate of KH. After twofold repetition of such leaching, the residual pentane was removed in a soft stream of dry argon gas emanating from a long pipette for at least 5-15 seconds, leaving dry KH powder (29