Lithium diphenylphosphide and arsenide have been investigated as reagents for metal-halogen exchange. Reactions involving two molar equivalents of the anions with 1,2-dibromoethylenes lead to moderate to good yields of acetylenes and smaller yields of phosphorus-containing by-products. Similar reactions with 2,3-dichlorobuta-l,3-diene and 2-chlorobuta-l,3-diene give SN2' rather than direct substitution products. Evidence is presented that adamantene is an initial product in the reaction of lithium diphenylphosphide with 1,Z-di-iodo-and 1,2-dibromo-adarnantane, but not with 1,2-dichIoroadamantane, although 1 -and 2-adamantyldiphenylphosphine oxides are formed in every case. Finally reactions of phosphide anion with geminal dihalides were briefly investigated ; 1 ,I -dibromo-2,2-diphenylethylene gave diphenylacetylene as the major product.Attack on halogen by neutral tervalent phosphorus has been commonly observed and through the work of Appe12 and Castro has been developed into a series of general synthetic procedures. These compounds have also been used as debrominating reagent^.^ By comparison there have been few reports of attack on halogen by phosphide anions and we are aware of only one example of dehalogenation, that of 1,2-dibromoethane to give ethylene.6 However, we were encouraged to investigate group 5 anions as metallating, and hence dehalogenat ing, reagents since organometallic compounds, especially organolithium, have been extensively used in this way' and the greater polarisability of group 5 nucleophiles suggests that attack on halogen would be more likely in these cases than for carbon nucleophiles.Our choice of lithium compounds as reagents was on the basis of greatest polarisability and this is supported by the preferential attack on halogen by lithium phosphides and carbon by sodium phosphides.6 On the same basis the larger arsenide anion should be a more satisfactory reagent than phosphide.
Results and Discussion(a) Reactions of 1,2-Dihalogenoa1kenes.-Aguiar and his co-workers have extensively investigated the reactions of vicinal vinyl dichlorides with phosphides * and have shown that nucleophilic displacement at carbon is the predominant pathway; t however, the softer halogen should encourage attack on bromine in the corresponding dibromide.The addition of two molar equivalents of lithium diphenylphosphide or lithium diphenylarsenide to 1 ,Zdibromoethylene (la) (64 : 36, cis : trans), a$-dibromostyrene (lb) (29 : 71, cis : trans), and a,P-dibromostilbene (lc) (40 : 60, cis : trans) in refluxing tetrahydrofuran gave in each case the corresponding acetylene ( Table 1). Determination of any preferred stereochemistry for the elimination was precluded by the rapid isomerisation of the dibromides under the reaction condit i o n~.~ In the phosphide reactions, after oxidative work-up, diphenylphosphinic acid was also formed, while small yields of the oxides (2), (3), and (4) were isolated from reactions with the dibromides (la), (lb), and (lc), respectively.The formation of acetylenes in these react...