In this work the two individual detector systems of the HPGe detector with its massive shielding and the anti cosmics veto have been combined to a low back-ground γ ray spectrometry system. Furthermore, a data analysis chain was devel-oped to allow for reliable low background material characterisation with respect to radioactivity.
An enormous background suppression was achieved reflecting the great efforts put into the construction of the DLB detector system. The already remarkable low background level of the γ ray spectrometry system without the active veto is even further reduced by nearly a factor of 6 by applying the anti coincidence sys-tem. By this means, a very low resulting integral count rate of 4.06 cts/(kg min) is achieved. Nevertheless, several small peaks caused by radioimpurities in the detector crystal and all surrounding materials can be identified in the spectrum.
Their origin has been investigated by means of MC simulation and the deter-mined activity values have been cross checked with other data. In addition to the peaks caused by radioimpurities, only three small peaks caused by neutron cap-ture induced metastable states of germanium isotopes can be identified. Due to the integrated neutron shielding these peaks are much smaller than in the spec-tra of other γ ray spectrometry systems. Based on a formula given in [ŠAK+92], a remarkably low thermal neutron flux of 4.4(23)×10−5cm−2s−1 at the loca-tion of the HPGe detector was determined, which is nearly two orders of mag-nitude below the value of conventional detector systems and proves the effec-tiveness of the neutron shielding. The thermal neutron induced peaks can not effectively be suppressed further by the anti cosmics veto, since the lifetimes of the metastable states of germanium are longer than the applied rejection time of the veto. The similar determined upper limit on the flux of fast neutrons is with 2.2×10−4cm−2s−1nearly equal to the upper limit for a system with 500 mwe of overburden (compare [WMA+96]).
Based on the DIN ISO 11929:2011 [DIN11] standard a data analysis software was developed to allow the reliable determination of nuclide activities from the low background γ ray spectra acquired with the DLB. This software provides dif-ferent methods to the analyst, including efficiency determination based on MC simulations, and allows to present the obtained results in a standardised evalua-tion report to the requesting user.
The MC simulations necessary for the efficiency determination rely on some extensions to the COBRA simulation framework VENOM allowing the read-in of externally defined GDML geometries. In the scope of the present work
fur-ther methods have been implemented to make the convenient modification of simulation cuts possible, which puts the focus on important regions within the simulated detector set-up as well as reduces the computational effort.
Using the MC geometry of the detector system based on the data sheet of the HPGe detector and other mechanical drawings, a comparison between measure-ments and simulations revealed discrepancies of up to 28 %. While the efficiency in the low energy region was vastly underestimated in the simulation, the effi-ciency for γ rays with energy higher than 200 keV was overestimated by a nearly constant fraction. Therefore, complex and detailed simulation studies have been carried out to determine modified detector parameters enabling the reliable re-production of the measured detection characteristics. Based on the measured peak ratio of two low energetic γ ray peaks, it was possible to determine the thick-ness of the outer detector contact, which especially influences the efficiency in the energy region below 200 keV, by MC simulation. In this manner, a thickness of the outer contact was determined that is with 437 µm about 37 % smaller than the value given in the manufacturers data sheet. The vertical distance of the detector crystal to the inside of the endcap was deduced by comparing the ratio of the same low energetic γ ray peak in two spectra recorded with the same radioactive source placed at different, but precisely known heights above the detector end-cap. From these spectra a distance more than twice as large as the nominal value was determined, which is necessary to adjust an overall bias of the detection ef-ficiencies in the MC simulation to the ones seen in the measurements (compare Table 5.5). The detection efficiency in the high energy region does not depend sig-nificantly on the thickness of the outer contact, but the overall detector volume determines the probability for multi-site interactions, which dominate the full en-ergy deposition in this enen-ergy region. By a slight adjustment of the dimensions of the inner contact, an overall good agreement with deviations in the range of
±3 % between simulation and measurement has been obtained. Except from the γlines from the used152Eu source, which was in general problematic, the range of deviations holds true for all γ lines used in the MC optimisation.
Based on the established MC model of the detector system, the analysis of dif-ferent material samples was conducted. A comparison of the results determined for two IAEA proficiency tests using the new MC model reveals a general good agreement with the target values. Nevertheless, some values show significant deviations, but cross-checks with the results of M. Laubenstein, LNGS, suggest that the indicated target values could be biased. After proving the reliability of the developed analysis methods, several material samples were examined for the COBRA experiment. In the assessment of a conductive silver sample with a mass of 200 g upper limits as low as a few mBq/kg were determined for some radionuclides, what emphasises the detection potential of the constructed γ ray spectrometry system.
Besides the efforts put in the commissioning of the DLB detector system as well
as the development of the analysis chain, a huge contribution to the redesign and relocation of the COBRA R&D set-up at LNGS was made. In addition to the mod-ification of the neutron shielding a completely new EMI shield was designed and its construction supervised. The new parts of the lead and copper shielding, nec-essary for the modification as well, were also designed. Only due to the in detail planned time schedule to coordinate the delivery of machined parts and ordered equipment, it was possible to relocate the whole detector set-up at LNGS within three weeks. In combination with the DAQ electronics developed by O. Schulz, T. Köttig and J. Tebrügge the value of the modifications is reflected by the high quality data taken with the COBRA set-up since then.
The discrepancies observed for the152Eu γ lines are most likely caused by an inaccurately calibrated activity of the radioactive source as well as a primary par-ticle generator in Geant4 not able to correctly reproduce the decay characteristics of152Eu. The insufficiencies of the data that is used by the decay generator have also been seen in the comparison of the simulated decay characteristics with liter-ature values for different nuclides of the decay chains. Since the decay databases have been updated to the ENSDF 2012 data set in Geant4 10.0, efforts should be put in the migration of VENOM to this new Geant4 release. To further in-crease the reliability of the MC model, the simulation should be compared to measurements of additional calibrated radioactive sources. As far as possible these sources should be voluminous and at least one nuclide with a low ener-getic γ ray emission should be contained for a check of the thickness of the outer contact. To a certain extent this could be realised by producing a source with a mixture of different chemicals containing known amounts of naturally occurring radioactive isotopes of lutetium, lanthanum and potassium (compare [IHM+97]).
The results of this work show that the DLB has already a remarkable high sensi-tivity for the detection of traces of radioacsensi-tivity despite its location above ground.
As it is recommended in this work, the extension of the anti cosmics veto de-tector is currently implemented. This will provide a further suppression of the remaining background and allow to achieve lower detection limits in the same acquisition time.