Detailed Ion Temperature Information from the Neutron Emission Spectrum

Abstract

Neutron emission recorded with a spectrometer comes from the plasma within the region defined by a collimator and is a line integral along the chord viewed. The resulting spectrum, S(E_n), is, for a thermal plasma, a superposition of Gaussian distributions (G(E_n,T)) each specified by the temperature of the plasma region of emission. The superposition is weighted by the emission brightness distribution, B(T), reflecting the integrated emissivity over the regions with temperature T, y_n(T). This gives a heavy dominance to the high temperature regions (normally the core) but only as long as the peak temperature is not too high, say T_M<20 keV. S(E_n) can be significantly narrower than G(T_M) so that from its width one deduces an apparent temperature T_A<T_M which is not a well defined plasma parameter. Various prescriptions have been presented relating T_A to T_M (see e. g. ref.¹). It has also been noted² that much of the ambiguity in the conventional (width based) analysis of neutron spectra can be removed by an elemental decomposition of the spectrum given that data of high statistical accuracy can be recorded. This should be possible in neutron spectrometry diagnostics of DT plasmas on ITER³. Here we will therefore illustrate the temperature information that can be obtained from a detailed shape analysis of the neutron spectrum using the slope from the peak at E₀ ≃ 14 MeV towards the high energy tail where the hot plasma components dominate as one approaches E_n≈15.5 MeV; the latter is the estimated limit due to the onset of α-particle kinetic effects^2,4,5