A Combined Computational and Experimental Approach to Evaluating Ionizing Radiation Produced in Accelerators for Ion Implantation Devices for Radiation Shielding

Ion implantation for advanced semiconductor manufacturing is increasing beam energy to achieve deeper doping in silicon wafers, opening new nuclear reaction channels that can lead to undesirable radiation dose fields in chip fabrication plants. For example, helium doping with energies above 1.7 MeV exceeds the 29Si(alpha,n)32S reaction threshold and results in the generation of fast neutrons. Gamma and neutron production cross section data and computational models for predictive design for ion implantation in the multi-MeV regime are, respectively, not comprehensively measured and validated, complicating efforts to implement effective radiation shielding, eliminate problematic materials and isotopes, and take other dose-reduction measures. We present initial results characterizing the gamma and neutron fields present during helium implantation at 2.2 MeV in ultrapure silicon using an Axcelis Firefly ion implanter; the results are being used to validate Geant4 neutron-production reaction models - the QBBC HP physics list using TENDL and ENDF cross sections within the ParticleHP package - towards predictive capability for dose minimization engineering. Experimental energy spectra obtained with a 4"x6"x16" NaI(Tl) detector and LaBr3(Ce) gamma detector show detailed energy peak structures, corresponding to identifiable nuclear reactions with materials in the silicon target and accelerator beamline. An EJ309 liquid organic scintillator with digital pulse processing detected fast neutrons with peak energies of 1.1 MeV, which roughly agrees with initial Geant4 simulations showing a broad distribution from 0 to 1000 keV with a large peak between 200 and 1000 keV. This preliminary investigation indicates that the Geant4 ParticleHP package predicts the neutron flux and energy spectra moderately well from the 29Si(a,n)32S reaction but that more accurate experimental cross section measurements, model improvement, and validation, particularly for neutron-production reactions, is required for predictive modeling to inform dose reduction strategies.