Microdosimetric measurements and the variance-covariance method. Some experimental experience.

Systematic and statistical uncertainties in the variance-covariance method have been investigated. Two spherical wall-less detectors have been used to determine the dose mean lineal energy (yD) in a neutron beam of 5.7 MeV produced by a Van de Graaff accelerator. It is shown that certain systematic uncertainties influenced the mean yD of the two detectors much less than yD from only one of them. A statistical uncertainty of 6% (95% confidence level) was achieved if yD was calculated from 2000 measurements. In this particular experiment insufficient shieldings of the preamplifiers positioned in the beam turned out to limit the possibility to measure below 20 nm.