Magnus Schlösser, Simone Rupp, Hendrik Seitz, Sebastian Fischer, Beate Bornschein, Tim M. James, Helmut H. Telle 

Journal of Molecular Structure, in press, Available online, doi:10.1016/j.molstruc.2012.11.022,

The Karlsruhe Tritium Neutrino (KATRIN) experiment aims to measure the neutrino mass via high-precision electron spectroscopy of the tritium β-decay with a sensitivity of mν = 200 meV/c2 (90% C.L.). This can only be achieved if systematic uncertainties are minimized. An important parameter is the isotopic composition of the tritium gas used as the gaseous β-electron source, which is measured inline by Raman spectroscopy. The KATRIN experiment requires a measurement trueness of better than 10% of said composition; to achieve this, accurate calibration of the Raman system for all hydrogen isotopologues (H2, HD, D2, HT, DT, T2) is required. Here we present two independent calibration methods, namely (i) a gas sampling technique, which promises high accuracy, but which is difficult to apply to tritiated species; and (ii) an approach via theoretical Raman signals (theoretical intensities plus spectral sensitivity), which in principle includes all six isotopologues. For the latter method we incorporated ab initio off-diagonal matrix elements of the polarizability from the literature; these have been verified by depolarization measurements. The system’s spectral sensitivity was determined by a NIST-traceable SRM2242 luminescence standard. Both methods exhibited their individual merits and difficulties, but in cross calibration proved to be successful: a comparison for the non-radioactive isotopologues (H2, HD, D2) yielded agreement to better than 2% for the relative Raman response function. This is within the estimated (dominant) uncertainty of the theoretical Raman signal approach of about 3%. Therefore, one can be confident that, when using this approach, the trueness requirement of 10% for the KATRIN-relevant species (T2, DT, D2 and HT) will in all likelihood be exceeded..