J.M. Perez

A new approach to automatic radiation spectrum analysis

The application of adaptive methods to the solution of the automatic radioisotope identification problem using the energy spectrum is described. The identification is carried out by means of neural networks, which allow the use of relatively reduced computational structures, while keeping high pattern recognition capability. In this context, it has been found that one of these simple structures, once adequately trained, is quite suitable to identify a given isotope present in a mixture of elements as well as the relative proportions of each identified substance. Preliminary results are presented, and are deemed good enough to consider these adaptive structures as powerful and simple tools in the automatic spectrum analysis. >

Application of neural network techniques in gamma spectroscopy

Abstract The usual methods of automatic radiation spectra analysis, based on fittings of peaks and background to exact mathematical curves, are valid for high resolution detectors. However, these methods are less successful for lower resolution detectors, such as the common scintillators or new room temperature semiconductors. Trying to solve some of the problems inherent in the application of complex fittings to the response of these detectors, we test here a new and less strict approach, based on the use of a neural network algorithm known as “associative memory”. This method appears useful in those cases in which a simple operation and a fast response are needed, together with a reasonable (and not extreme) accuracy. Furthermore, as the pattern recognition is carried out through the rough shape of the whole spectrum, instead of each individual peak, it can be used with advantage for low resolution detectors. With the idea of comparing the behavior of this method with the “classical” ones, the response of the network in the analysis of several spectra, taken with a NaI spectrometer, is presented.

Energy deposition model for low-energy electrons (10–10 000 eV) in air

An energy deposition model for electrons in air that can be useful in microdosimetric applications is presented in this study. The model is based on a Monte Carlo simulation of the single electron scattering processes that can take place with the molecular constituents of the air in the energy range 10‐10 000 eV. The input parameters for this procedure have been the electron scattering cross sections, both differential and integral. These parameters were calculated using a model potential method which describes the electron scattering with the molecular constituent of air. The reliability of the calculated integral cross section values has been evaluated by comparison with direct total electron scattering cross-section measurements performed by us in a transmission beam experiment. Experimental energy loss spectra for electrons in air have been used as probability distribution functions to define the electron energy loss in single collision events. The resulting model has been applied to simulate the electron transport through a gas cell containing air at different pressures and the results have been compared with those observed in the experiments. Finally, as an example of its applicability to dosimetric issues, the energy deposition of 10 000 eV by means of successive collisions in a free air chamber has been simulated.

Drift problems in the automatic analysis of gamma-ray spectra using associative memory algorithms

Perturbations affecting nuclear radiation spectrometers during their operation frequently spoil the accuracy of automatic analysis methods. One of the problems usually found in practice refers to fluctuations in the spectrum gain and zero, produced by drifts in the detector and nuclear electronics. The pattern acquired in these conditions may be significantly different from that expected with stable instrumentation, thus complicating the identification and quantification of the radionuclides present in it. The performance of Associative Memory algorithms when dealing with spectra affected by drifts is explored considering a linear energy-calibration function. The formulation of the extended algorithm, constructed to quantify the possible presence of drifts in the spectrometer, is deduced and the results obtained from its application to several practical cases are commented. >

Production rate of proton-induced isotopes in different materials

Abstract High background counting rates are the main limitation on sensitivity of satellite borne gamma-ray instrumentations. The observed background comes from different sources: cosmic diffuse, charged particles and high energetic photons. Among the different background components, that due to the activation of the telescope and satellite passive materials by cosmic protons is the most difficult component to evaluate. In the framework of background studies and sensitivity estimations for the INTEGRAL and MINISAT-01 projects, a wide range of materials has been irradiated with proton beams at different energies to identify the induced unstable isotopes and their production cross-sections. In this paper we present experimental results obtained from the analysis of such irradiation experiments. These values are compared to those obtained by means of the two frequently used prediction methods: GEANT/GCALOR Monte-Carlo code and Silberberg and Tsao semiempirical expressions.

Energy Deposition Models at the Molecular Level in Biological Systems

Abstract In this study we have developed a model to describe the electron interaction of intermediate and high energy electrons (10–10000 eV) with some molecules of biological interest. Differential and integral electron scattering cross sections have been calculated with an optical potential method following an independent atom representation. Important improvement related to relativistic corrections, many-body effects, local velocity considerations and a screening correction procedure which take into account the overlapping of the constituent atoms in the molecule have been introduced to improve the accuracy and applicability of the method for a high variety of molecular targets. The accuracy of these calculations has been checked by comparison with total electron scattering cross section data we have measured in a transmission beam experiment with experimental errors within 5%. Finally, we have developed a Monte Carlo simulation program, based on the general tools of GEANT4, which uses as input parameters our calculated cross sectional data and the energy loss distribution functions based on the experimental energy loss spectra. This simulation procedure allows energy deposition models at the molecular level that could be very useful in biological and medical applications when microscopic energy deposition patterns are required.

Spectrometric response of large volume CdZnTe coplanar detectors

In this paper, the spectroscopic performance of a new large volume coplanar detector is studied. Spectrometric results using classical techniques are presented. The new unit confirmed that this design of coplanar-grid anodes provided an acceptable balance between anode weighting potentials. The material of this detector is not as homogeneous as desirable. Nonetheless, the detector demonstrated a resolution of the order of 2.7% FWHM for 662 keV. The spectroscopic properties at different interaction depths were studied by making use of a multiparametric digital system. Depth sensing achievable resolution is quantified by making use of this setup. The interaction profile at different depths of the detector was compared with that expected in ideal detectors. An adaptation of the multiparametric digital system was applied to the study of the waveforms generated in the preamplifiers connected to the electrodes. Induced charge waveforms at selected interaction depths were analyzed in order to study the deviations in the interaction depth profiles found between simulation and theory. The causes are attributed to trapping effects in the detector bulk. Unexpected pulses are explained by modeling the charge drift in the detector. The mobility of electrons in these detectors was studied by digital signal analysis, which gave a value close to 950 cm/sup 2/ V/sup -1/ s/sup -1/ at room temperature. Results of the dependence of /spl mu//sub e/ with T are presented in the operation temperature range of these detectors.

Performance comparison of a large volume CZT semiconductor detector and a LaBr/sub 3/(Ce) scintillator detector

The development of portable nuclear instrumentation demands compact high sensitivity detectors operated at room temperature. The sensitivity of these detectors mainly depends on two parameters: absolute efficiency and energy resolution. In order to provide high efficiency, large volumes are needed. For semiconductor detectors able to operate at room temperature, the largest effective volumes with acceptable resolution are achieved with CdZnTe coplanar-grid and pixel detectors. On the other hand, new scintillation materials were recently developed with spectrometric capabilities only reachable some years ago with semiconductors. In this work we compare the performance of two state of the art detectors of different technologies with a relative large volume: a coplanar-grid CdZnTe detector with dimensions 15 mmtimes15 mmtimes10 mm and a LaBr3(Ce) crystal with volume 18 mmtimes18 mmtimes30 mm. The CdZnTe crystal was made by Yinnel Tech (USA) and the detector was manufactured by BSI (Latvia). The LaBr3(Ce) scintillator was grown and encapsulated by Saint-Gobain (France). The energy resolution for the CZT detector is 2.05% FWHM at 662 keV. The resolution for the scintillator at this energy was near 3%. The total efficiency was studied in a setup with calibrated point sources. The experimental spectra were compared with Monte-Carlo simulations performed with Geant4. The implications of the results of this comparison are discussed in the context of the practical use of these unitsThe development of portable nuclear instrumentation demands compact high sensitivity detectors operated at room temperature. The sensitivity of these detectors mainly depends on two parameters: absolute efficiency and energy resolution in the range 10-1500 keV. In order to provide high efficiency, large volumes are needed. For semiconductor detectors able to operate at room temperature, the largest effective volumes with acceptable resolution are achieved with CdZnTe coplanar-grid and pixel detectors. On the other hand, new scintillation materials have been recently developed with spectrometric capabilities only reachable, some years ago, with semiconductors. In this work we compare the performance of two states of the art detectors of different technologies with relative large volume: a coplanar-grid CdZnTe detector with dimensions 15 mm/spl times/15 mm/spl times/10 mm and a LaBr/sub 3/(Ce) crystal with volume 18 mm/spl times/18 mm/spl times/30 mm. The CdZnTe crystal was made by Yinnel Tech (USA) and the detector was manufactured by BSI (LV). The LaBr/sub 3/(Ce) scintillator was grown and made by Saint-Gobain (F). The energy resolution for the CZT detector is 2.05% FWHM at 662 keV. The resolution for the scintillator at this energy is near 3%. The total efficiency was studied in a setup with calibrated point sources. The experimental spectrum were compared with Monte-Carlo simulations performed with Geant4. Loss of total and photopeak efficiency due to detector defects was analyzed. The implications of the results of this comparison are discussed with regards to the practical use of these units.

Computer simulations of the behaviour of the partial charge collection model in thick HgI2 γ-detectors

Abstract The partial charge collection model has been reported as the most suitable technique to form good-resolution spectra with thick HgI 2 γ-detectors. This method is based on measuring the charge signal induced by the drift of the free carriers generated in the interaction, while the electrons have not reached the positive electrode, instead of taking the total pulse height as a measure of the energy. A computer simulation of the whole process, from the interaction of the γ-ray with the detector to the electronic signal treatment, has been developed. With the help of this tool, several situations concerning thick HgI 2 detectors have been considered, analysing different practical implementations of the partial collection method. In particular, two approaches, one based on shaping at short times and the other using a flash pulse digitation, have been extensively discussed and compared.

Use of thick HgI2 detectors as intelligent spectrometers

Abstract Mercuric iodide is a very attractive material to detect ionizing radiation due to its high stopping power and wide energy gap, which allows the use of a small and compact detectors at room temperature. However, the spectroscopic performances of these detectors are poor in comparison with other more popular semiconductors with better transport characteristics. This effect becomes dramatic when thick crystals are used. The partial charge-collection method is reported to be the most suitable one for enhancing the energy resolution achieved with thick detectors. A Monte Carlo simulation of the behavior of the model and its dependence with crystals and electronic parameters is presented, giving operating rules that optimize the system performance in each situation. Specially designed hardware has been developed to extract the maximum information of the charge pulse produced by photon-detector interaction, according with the results of the simulation. As a final step, an automatic isotope-identification process, based on the use of neural networks, is performed, the identification being the true output of the whole system. Due to the strong dependence of this output on the free hardware parameters, an adaptive network is designed to act on these parameters in such a way that the system converges automatically to the best identification.