The BNL-AGS E906 experiment was the first fully electronic experiment to produce and study double hypernuclei with large statistics. Two dominant structures were observed in the correlated $\pi^-$--$\pi^-$ momentum matrix at (p$_{ \pi-H}$,p$_{ \pi-L}$) = (133,114) MeV/c and at (114,104) MeV/c. In this work we argue that the interpretation of the structure at (133,114) MeV/c in terms of $^3_{ \Lambda}$H+$^4_{ \Lambda}$H pairs is questionable. We show, that neither a scenario where the hypernuclei are produced after capture of a stopped $\Xi^-$ by a $^9$Be nucleus nor interactions of energetic $\Xi^-$ with $^9$Be nuclei in the target material can produce a sufficient amount of such twin pairs. We have therefore explored the conjecture of Avraham Gal that decays of the $_{ \Lambda\Lambda}^{ 4}$n may be responsible for the observed structure. Indeed, the inclusion of $_{ \Lambda\Lambda}^{ 4}$n with a two-body $\pi^-$ branching ratio of 50 % in a statistical multifragmentation model allows to describe the E906 data remarkably well. On the other hand, a bound $^{ 3}_{ \Lambda}$n nucleus would cause a striking structure in the momentum correlation matrix which is clearly inconsistent with the observation of E906.
Categories: nucl-ex nucl-th