Synthesis of trans-1,2-dimetalloalkenes through reductive anti-dimagnesiation and dialumination of alkynes

Polar reactive organometallic species have been key reagents in synthesis for more than a century. Stereodefined 1,2-dimetallated alkenes offer promising synthetic utility; however, few methods are available for their preparation due to their relatively low stability. Here we report the reductive anti-1,2-dimetallation of alkynes to stereoselectively generate trans-1,2-dimagnesio- and 1,2-dialuminoalkenes, which are stable and have been demonstrated in organic synthesis. These stereodefined 1,2-dimetallated alkenes are prepared through the use of a sodium dispersion as a reducing agent, and organomagnesium and organoaluminium halides as reduction-resistant electrophiles. Highly nucleophilic 1,2-dimagnesioalkenes serve as dual Grignard reagents and have been demonstrated to react with various electrophiles to afford anti-difunctionalized alkenes. The 1,2-dialuminoalkenes react with paraformaldehyde with dearomatization of the aryl moieties to form the corresponding dearomatized 1,4-diols, with the overall reaction being regarded as alkynyl-directed dearomatization of arenes. X-ray crystallographic analysis further supports the formation of trans-1,2-dimagnesio- and 1,2-dialuminoalkenes, with computational studies providing insight into the mechanism of dearomative difunctionalization. Stereodefined 1,2-dimetallated alkenes are underexplored in organic synthesis due to their relatively low stability. Now, the reductive anti-1,2-dimetallation of alkynes provides access to trans-1,2-dimagnesio- and 1,2-dialuminoalkenes. The process uses sodium dispersion as a reducing agent with organomagnesium and organoaluminium halides as reduction-resistant electrophiles.

Organometallic species bearing a polar carbon-metal bond occupy an indispensable position in organic chemistry due to their high reactivity 1 . As a representative organomagnesium species, Grignard reagents have been widely used for the synthesis of useful molecules ranging from bioactive compounds to organic semiconductors since their discovery in 1900 2,3 . Nowadays, a variety of polar organometallic species, such as organolithium, organozinc and organoaluminium compounds, are readily available and have significantly contributed to the development of organic chemistry.
The preparation of such polar organometallics is often dependent on deprotonation at acidic C-H bonds or on metallation of organic halides via a reductive process or halogen-metal exchange (Fig. 1a) [4][5][6] . Although these methods are reliable, special classes of precursors, acidic hydrocarbons and halogenated compounds are necessary.
Unsaturated hydrocarbons such as alkenes and alkynes represent another important class of precursor (Fig. 1b). Hydrometallation and carbometallation across unsaturated bonds provide highly useful reactive organometallic species such as alkyl and alkenyl metals 7-12 . A key advantage of metallating unsaturated bonds is the increase in molecular complexity and values by starting from simple unsaturated compounds such as alkynes, which are among the most prevalent and primitive structures in organic molecules. The subsequent conversion of the resulting carbon-metal bond then offers straightforward routes to target molecules.
Given the unparalleled importance of monometallation such as hydrometallation and carbometallation, 1,2-dimetallation of unsaturated bonds would be a fascinating transformation. For example, twoelectron reduction of alkynes should ideally lead to 1,2-dimetallation Article https://doi.org/10.1038/s44160-022-00189-z We have been interested in the development of alkali-metalpromoted reductive transformation of unsaturated compounds [24][25][26][27][28] using reduction-resistant electrophiles as key reagents [29][30][31] . On one hand, this class of electrophiles, which includes trialkoxyboranes, are resistant to single-electron reduction and hence can coexist in the same pot where the reduction of unsaturated substrates takes place. On the other hand, they are sufficiently electrophilic to immediately trap the unstable anions thus formed. We have developed sodium-mediated 1,2-syn-diboration of alkynes with trimethoxyborane 30 . This success encouraged us to develop a new method to generate more reactive and polar 1,2-dimetalloalkenes that are much more difficult to prepare and use. Here we report that the sodium-mediated reductive dimetallation of alkynes proceeds in the presence of magnesium-and aluminium-based reduction-resistant electrophiles (Fig. 1f). The reaction efficiently affords the corresponding reactive but stable 1,2-dimetalloalkenes with, more importantly, high anti-selectivity that is hard to achieve. The resulting trans-1,2-dimagnesioalkenes react with various electrophiles to afford the corresponding trans-difunctionalized alkenes. Their aluminium equivalents were found to induce an unexpected dearomatization [32][33][34][35][36][37][38] at the terminal arene upon treatment with an aldehyde. Crystallographic analysis of the intermediates and density functional theory (DFT) calculations of the dearomatization are also described herein.

Optimization and reaction scope of dimagnesiation
We started our investigation by optimizing the reaction conditions for 1,2-dimagnesiation of diphenylacetylene (1a) ( Table 1). The corresponding organomagnesium species 2a should be sensitive to air and moisture, and burdensome to isolate; therefore, the reaction efficiency was evaluated by the yield of stable diborylated product 3a after treatment of 2a with a boron electrophile. Following the conditions of the diboration of alkynes 30 , we initially employed magnesium alkoxides and halides as reduction-resistant magnesium electrophiles. However, magnesium ethoxide and related alkoxides were not readily soluble in THF, so that 3a was not obtained at all. Magnesium dichloride was not resistant to reduction and underwent preferential reduction to Rieke-type magnesium 39 without forming 3a. After further trials, it was ( Fig. 1c). The resulting 1,2-dimetalloalkenes should engage in further bond formations at the two reactive carbon-metal bonds to provide a wide variety of multisubstituted alkenes that are otherwise difficult to synthesize. Despite this promising utility, few methods for generating polar reactive 1,2-dimetalloalkenes have been reported [13][14][15][16][17][18] . The limited research is considered to be due to the following reactivitystability trade-off issues. (1) Electron injection into alkynes requires strong reductants such as alkali metals and solvated electrons. (2) After electron injection from alkali metal occurs, the resulting radical anion intermediates and the 1,2-dimetalloalkene products are too unstable for use in organic synthesis. When the electron injection to alkynes is too slow, as in the reduction of diphenylacetylene with lithium in Et 2 O, the generated radical anion intermediates readily dimerize to form 1,4-dimetallo-1,3-butadienes 19,20 . While 1,2-dilithioand 1,2-disodioalkenes can be generated from diphenylacetylene and alkali metal in THF 13 , they decompose via protonation by the THF solvent even at −78 °C and cannot be used for organic synthesis (Fig. 1d). Another notable example is that the dissolving metal reduction of alkynes always ends up with the formation of trans-alkenes via smooth protonations of anionic intermediates by a protic solvent (liquid ammonia) with isolation of any vinylic metal intermediates being elusive. (3) As examples of very limited successful cases, engineered bulky dinuclear Mg(I) and Al(II) complexes [14][15][16] are known to undergo anti-dimetallation and yield stable trans-1,2-dimetalloalkenes (Fig. 1e). Bulky diiminatoaluminium chloride complexes are also known to undergo anti-dialumination in the presence of potassium metal via the addition of an aluminium-centred radical to alkynes 17,18 . However, these complexes are ingeniously designed and decorated with special ligands, and are thus not readily accessible. Furthermore, the generated 1,2-dimetalloalkenes were not used for organic synthesis but were examined to undergo simple protonolysis, iodonolysis and transmetallation to zinc. (4) While a different approach to 1,2-dimetalloalkenes could be dimetallation of 1,2-dihaloalkenes, β-elimination from the 1-metallo-2-haloalkene intermediates takes place more rapidly than the second metallation [21][22][23] . Therefore, there remains ample room to develop facile synthesis and applications of potentially useful 1,2-dimetalloalkenes.  Article https://doi.org/10.1038/s44160-022-00189-z discovered that the Grignard reagents are suitable reduction-resistant magnesium electrophiles in the presence of alkynes although they are always considered as nucleophiles in organic synthesis.
In the presence of 2 equiv. of isopropylmagnesium bromide, 1a was reduced by 2 equiv. of a sodium dispersion (in mineral oil) in THF at 0 °C before treatment with methoxypinacolborane (MeOBpin) at 60 °C for 1 h (entry 1). The expected dimagnesiation and diborylation proceeded to afford 3a in 90% yield, although without stereoselectivity. Other boron electrophiles such as ethoxypinacolborane (EtOBpin) and isopropoxypinacolborane ( i PrOBpin) were screened, which identified i PrOBpin as the best for stereoretentive diboration (entries 2 and 3). The effect of the substituent on the organomagnesium species was investigated next. When methylmagnesium bromide (MeMgBr) was used, the conversion of 1a was relatively low, possibly because the reduction of MeMgBr competed with that of 1a (entry 4). The use of tert-butylmagnesium bromide or phenylmagnesium bromide afforded 3a with low stereoselectivities, although with good yields (entries 5 and 6). Finally, cyclopentylmagnesium bromide ( c Pent-MgBr) was found to be optimal in terms of both yield and selectivity (entry 7). When sodium lumps (diameter, ∼2 mm) or lithium granules (diameter, ∼2 mm) were used as the reductant, the efficiency of the electron transfer was so low because of their small surface area that the conversion of 1a was insufficient (entries 8 and 9). The use of lithium powder (diameter, ∼0.12-0.25 mm) improved the yield of 3a to 59%, although a lower yield than that in entry 7 (entry 10). Employing sodium naphthalenide resulted in the formation of 3a in 74% yield with good stereoselectivity (entry 11).
Having optimized reaction conditions (Table 1, entry 7), we investigated the scope of this dimagnesiation-diborylation sequence with respect to diarylacetylenes (Fig. 2a). It is worth noting that yields of the pure E isomers of 3 are shown here since the major isomers (E)-3 could be easily separated from the minor isomers (Z)-3 (E:Z = 77:23-91:9) by means of routine silica-gel chromatography (see Supplementary Information for the yields and isomeric ratios of the products in crude reaction mixtures). A variety of diarylacetylenes underwent the reduction to yield the corresponding (E)-diborylalkenes 3a-j in high yields, which represents a rare example of formal anti-diboration of alkynes [40][41][42][43][44] . In spite of the strongly reducing and basic conditions, ether (3c and 3f), thioether (3d), fluoro (3f) and silyl (3g) moieties were well tolerated during the reaction. Moreover, sterically demanding ortho-substituents in diarylacetylenes 1i and 1j did not hamper the reaction. In addition, the present anti-diboration also accommodates arylacetylene with a π-extended naphthyl group (1h). Instead of boron electrophiles, carbonyl compounds such as paraformaldehyde reacted with 2 to yield 1,4-diols 4aa and 4fa in high yields (Fig. 2b).

Dialumination and subsequent dearomatization reaction
Encouraged by the success of the dimagnesiation, dialumination was attempted using an aluminium-based electrophile. When dimethylaluminium chloride was used as the electrophile, similar anti-dimetallation took place (Fig. 3a). Surprisingly, the subsequent reaction of the dialuminoalkene 5a with paraformaldehyde resulted in the formation of not 4aa but an unexpected product, dearomatized diol 6a, with regeneration of the alkyne unit and with exclusive anti-selectivity. Further optimization (Supplementary Table 2) of the conditions for the dialumination (in 4-methyltetrahydropyran (4-MeTHP) at −78 °C), as shown in Fig. 3b, afforded 6a in 86% yield without the formation of 4aa and its stereoisomer 4aa′.
With the optimal conditions established, the scope with respect to arylacetylenes was explored (Fig. 3b). Diarylacetylenes 1a, 1b, 1h and 1m underwent the dearomatization to afford the corresponding 1,4-diols 6 in good yields. The naphthyl moiety of 1h was selectively dearomatized, probably due to its lower aromaticity than that of the phenyl moiety. It is noteworthy that sterically demanding 1m participated in the reaction to give a mixture of two regioisomers 6m and 6m′, regardless of the ortho-methyl substituent. The reaction also accommodates alkylarylacetylenes 1l and 1n-p. Biphenyl substrate 1o reacted selectively on the phenyl ring with an alkynyl substituent to yield 6o. Arylsilylacetylenes 1q and 1r were converted while the frangible silyl groups were untouched. These transformations are regarded as a novel class of dearomatization reaction of arylacetylenes directed by a carbon-carbon triple bond. As an additional note, these are rare examples of alkyne-directed reactions with the alkyne unit left in the products, while reported transformations directed by a carbon-carbon triple bond have always accompanied irreversible conversions of the reactive triple bond into a double bond 46,47 .

X-ray crystallography of trans-1,2-dimetalloalkenes
The structures of key dimetalloalkene intermediates 2 and 5 were unambiguously determined by X-ray crystallography (Fig. 4). A dimagnesioalkene for X-ray diffraction analysis was synthesized by a reaction of 1a with mineral-oil-free sodium lumps in the presence of isopropylmagnesium chloride for a long reaction time of 16 h. After removal   4 as yellowish orange crystals in 47% yield. Following a similar procedure, trans-1,2-dialuminoethene 5a Et (THF) 2 was prepared from diethylaluminium chloride as pale yellow crystals in 12% yield. The verification of the trans stereochemistry of 5 is important to eliminate the possibility of the preferential formation of its cis isomer 51 . X-ray diffraction analysis of single crystals of both 2a iPr (THF) 4 and 5a Et (THF) 2 [15][16][17] . Akin to the C i symmetric system in the crystal structures, NMR spectral analysis of 2a iPr (THF) 4 and 5a Et (THF) 2 also revealed symmetric features in solution. Notably, the 13 C NMR spectra of 2a iPr (THF) 4 and 5a Et (THF) 2 showed highly downfield-shifted 13 4 and 5a Et (THF) 2 were observed in the typical aromatic region.

Computational investigation of dearomatization
To investigate the mechanism of the unique dearomatization of arylacetylenes (Fig. 3a,b), DFT calculations were performed using the reaction of 1,2-dialumino-1,2-diphenylethene 5a Me with monomeric formaldehyde in THF solvent as a model reaction (Fig. 5).
In the case of dearomatization (Fig. 5a), the reaction begins with the coordination of a formaldehyde molecule to one of the aluminium atoms of 5a Me  The first C-C bond-forming step via TS-d1 tAl with dearomatization is rate-determining (ΔG ‡ = 18.8 kcal mol −1 ) while the second C-C bondforming step proceeds more smoothly (ΔG ‡ = 8.3 kcal mol −1 ), which is consistent with the experimental result that the allenylic by-products derived from INT-d3 tAl were not observed at all. Benzylic metal species, especially benzylmagnesiums, are known to undergo nucleophilic addition to aldehyde with dearomatization despite low efficiency and regioselectivity 52,53 . Our reaction is much more efficient and regioselective, probably because the second nucleophilic attack drives the reaction forward and suppresses the collapse back to the original dialuminoalkene 5a Me and formaldehyde. The low activation energy for each step allows the entire process to proceed smoothly even at room temperature. Although the reaction between 5 and paraformaldehyde was performed at 60 °C, high temperature is necessary for the decomposition of paraformaldehyde to the monomeric species H 2 C=O (Fig. 3b).
In the case of the conceivable twofold ipso-alkylation (Fig. 5b), the first and second C-C bond formations occur via four-membered cyclic transition states TS-i1 tAl and TS-i2 tAl , respectively. Unlike the dearomatization in Fig. 5a, the rate-determining step is the second C-C bond-forming step (INT-i3 tAl → TS-i2 tAl ; ΔG ‡ = 22.0 kcal mol −1 versus 21.3 kcal mol −1 for the first step), which is in good agreement with the experimental fact that a trace amount of monohydroxymethylated alkene derived from INT-i3 tAl was observed as a side product. The difference between the activation energies for the first steps in Fig. 5a,b was calculated to be 2.6 kcal mol −1 (18.8 kcal mol versus 21.3 kcal mol −1 ) in favour of the dearomatization process.
Similar calculations for the reaction of the magnesium counterpart 2a Me with two molecules of formaldehyde were next conducted to identify similar pathways for the dearomatization and the twofold ipsohydroxymethylations ( Supplementary Figs. 7 and 8). In contrast to the aluminium case, the activation barrier of the first C-C bond formation step for the ipso-hydroxymethylation is much lower than that for the dearomatization (ΔΔG ‡ = 7.6 kcal mol −1 ), which coincides with the exclusive ipso-hydroxymethylation without the dearomatization as shown in Fig. 2b. Based on distortion/interaction analyses 54,55 ( Supplementary  Figs. 15−17), we assume that the shorter C-Al bond (2.08 Å) in TS-i1 tAl would induce a larger ring strain in the four-membered transition state TS-i1 tAl than the C-Mg bond (2.22 Å) in TS-i1 tMg , and consequently render the activation energy at TS-i1 tAl higher.

Conclusion
We have uncovered reductive anti-1,2-dimetallation of arylacetylenes using sodium dispersion as a reducing agent and cyclopentylmagnesium bromide and dimethylaluminium chloride as reductionresistant electrophiles. The intermediates, trans-1,2-dimagnesio-and dialuminoalkenes, represent rare examples of trans-1,2-dimetalloalkenes that are difficult to prepare, reasonably stable and useful for organic synthesis. Highly nucleophilic 1,2-dimagnesioalkenes reacted with a wide variety of electrophiles to afford anti-difunctionalized alkenes in a stereoselective manner. Interestingly, the reaction of the 1,2-dialuminoalkenes with paraformaldehyde induced dearomatization of the aryl moieties to form the corresponding dearomatized 1,4-diols with recovery of the alkyne unit, which is regarded as alkynyldirected dearomatization of arenes. These represent the synthetic versatility of the dimagnesioalkenes and the unique reactivity of the dialuminoalkenes. The structures of 1,2-dimagnesio-and aluminoalkenes were unambiguously determined by X-ray crystallography to confirm their intermediacy. The mechanism of the dearomative difunctionalization was investigated using DFT calculations to clarify the unusual behaviour of unique organoaluminium species. Article https://doi.org/10.1038/s44160-022-00189-z 0.20 ml, 2.0 mmol) was added dropwise over 30 s to the tube, and the resulting suspension was stirred at 0 °C for 30 min. After the addition of CuCN·2LiCl (1.0 M in THF, 0.10 ml, 0.10 mmol) and an electrophile (4.0 mmol) to the tube, the reaction mixture was allowed to warm to room temperature and stirred for an additional 1 h. The reaction was then quenched by the addition of aqueous HCl (2 M, 2 ml) and H 2 O (2 ml), and the resulting biphasic solution was extracted with Et 2 O (4 ml × 3). The combined organic layer was dried over Na 2 SO 4 , filtered and concentrated under reduced pressure. Purification of the residue by column chromatography on silica gel (eluent: hexane/ EtOAc) provided 4. If necessary, further purification was done by gel permeation chromatography.
General procedure for synthesis of dearomatized diols 6 See Fig. 3b. An oven-dried 20 ml Schlenk tube was charged with alkyne 1 (1.0 mmol), Me 2 AlCl (1.0 M in hexane, 2.0 ml, 2.0 mmol) and 4-MeTHP (6.0 ml). After cooling the mixture to −78 °C, sodium dispersion (10 M, 0.22 ml, 2.2 mmol) was added dropwise over 30 s to the tube, and the resulting suspension was stirred at −78 °C for 30 min. After the addition of paraformaldehyde (4.0 mmol) to the tube, the reaction mixture was allowed to warm to room temperature over 10 min and then stirred at 60 °C for an additional 1 h. After cooling the mixture to room temperature, the reaction was quenched by the addition of aqueous HCl (2 M, 3.5 ml), and the resulting biphasic solution was extracted with Et 2 O (4 ml × 3). The combined organic layer was dried over Na 2 SO 4 , filtered and concentrated under reduced pressure. Purification of the residue by column chromatography on silica gel (eluent: hexane/ EtOAc) provided 6.
Copies of the data can be obtained free of charge via https://www.ccdc. cam.ac.uk/structures/. The data supporting the findings of this study are available within the Article and its Supplementary Information.