Problem 2195

Created attachment 593 [details]
neutron production on Deuterium with electron beam

We saw at Jefferson Lab an increase in the rates of neutron production from a switch from liquid Hydrogen target to a Deuterium target: This was not well described by GEANT4 (more than one order of magnitude in discrepancy).
We then went and investigate the issue, seeing that the problem has been there since GEANT4.9.6 (see attachment): Version previous (4.9.5 and earlier ones) better describe the increase seen in the detectors at Jefferson Lab.
After presenting this issue at the last GEANT4 technical forum, the problem was discussed.
Follows the e-mail exchanged with Dennis H. Wright, where he points out better the issue:
" I believe there are two problems, both in the Bertini cascade model.
The electro-nuclear reaction in Geant4 converts the virtual gamma to a real one, then uses the Bertini code to perform the gamma-nuclear reaction on the target nucleus.
    The first problem is that Bertini assumes a nucleus with A > 3.   So it tries to initiate a cascade and then a precompound deexcitation of the residual 
nucleus, neither of which is appropriate for deuterium.  The second problem 
is related to the first: one of the processes that occur after the cascade phase is coalescence.  Nucleons ejected during the cascade are given the chance to  recombine into light ions as long as their relative momenta are not too far 
apart.   This would account for at least some of the disappearance of neutrons 
that you see.
   Years ago we split off the case of 1H targets to treat it differently, but we somehow omitted doing the same for D. "

Lorenzo Zana