| Zugriffsnummer | 28596 |
| Dokumenttyp | Dissertation |
| Peer Review | unbekannt |
| Sprache | Englisch |
| Titel | Time-resolved and position-resolved X-ray spectrometry with a pixelated detector |
| Autor(in); Institution |
Sievers, Peter; 6.3, Strahlenschutzdosimetrie, PTB-Braunschweig
|
| Quelle/Jahr | (2012), 103 S. |
| Dissertationsvermerk | Dissertation, Universität Erlangen-Nürnberg, 2012 |
| Persistent Identifier | |
| URL | |
| Klassifikationscode | PACS: 07.85.Fv ; PACS: 07.85.Nc ; PACS: 87.64.Aa |
| Freie Schlagworte | spectrometry ; detector ; monte-carlo simulation ; time-resolved ; position-resolved |
| Zusammenfassung | The aim of the work presented here was to measure X-ray spectra with a pixelated detector. Due to effects in the sensor the spectrum cannot be measured directly and has to be calculated by a deconvolution of the measured data. In the scope of this work the deconvolution of the measured spectra could be enhanced considerably by - amongst other things - the introduction of the Bayesian deconvolution method. Those improvements opened the possibilities for further measurements. For the measurements the detectors of the Medipix family have been used. They are nowadays used for a wide range of applications and scientific research. Their main advantage is the very high position resolution gained by a pixel pitch of 55µm and a high number of 65536 pixels. The Timepix detector has, in particular, two special possibilities of measurement: the ToA mode and the ToT mode. In ToA mode the arrival time of an impinging photon is measured and in ToT mode the amount of deposited charge is measured. The most common method of operation is counting the number of impinging photons that release a charge higher than a preset threshold in each pixel. As this released charge is proportional to the energy deposition of the impinging photon, one can perform energy-sensitive measurements. To perform the deconvolution of the measured energy distribution there is a need of an energy response matrix describing the detector response on radiation. For some detectors it is possible to obtain an analytic model of the response functions. Due to the high discrepancy between the impinging spectrum and the measured spectrum in case of detectors of the Medipix family, there is so far no analytic model. Thus, the detector response has to be simulated. As I could improve the precision of the measurement quite extensively, I also intended to tune the simulation with more accurate and appropriate models to gain the same level of accuracy. The results of measurement and simulation have then been compared and show a good agreement. Up to now the measurements of impinging spectra with a Timepix detector have been performed in radiation fields with a relatively high fluence. To cope with the requirement of measuring in radiation fields with a low fluence, there had to be changes in the method of analysis compared to those performed formerly. An important improvement in this context was the employment of the Bayesian deconvolution method. The spectra reconstructed with this method were then compared to the results of two different and established detection systems. Firstly, the shape of the deconvolved spectrum was compared to the one measured with a hpGe detector. Secondly, the calculated value of the kerma rate was compared to the one measured with an ionization chamber. This gave an estimate on the correctness of the absolute number of photons. Both comparisons have shown a good agreement and thus I was able to validate that the method delivers precise results. Compared to the formerly used spectrum-stripping method the Bayesian deconvolution turned out to be very stable and reliable. This robustness of the deconvolution method and the development of a pixel-by-pixel energy calibration were the keys towards position-resolved spectrometry. With such a precise energy calibration the energy resolution was enhanced by up to 45%. This improved accuracy in the measurement has been very demanding on the improvements of the simulation of the response matrix needed for deconvolution. Both this enhanced simulation and a pixel-by-pixel calibrated detector opened the possibility of measuring the anode heel effect. Not only the relative angular dependency of the spectrum emitted but also the change in the absolute photon fluence were measured. Furthermore, it is possible to even use small ROIs down to 4x4 pixels to evaluate a spectrum. This was then applied for the spectrometry of small focal spots of a miniature X-ray source used in therapeutics. Furthermore, the robustness and the stability of the applied Bayesian deconvolution method enabled the possibility of performing time-resolved spectrometric measurements. By measuring in ToA mode and in parallel performing a THL scan, it is possible to gain information on both energy and time. This method was then tested for a conventional X-ray tube for measuring the time dependence of the spectrum emitted during the switching-on process of the radiation. As expected, the results showed a relatively long time-dependent change of the spectrum. This method was then applied for proving that a newly developed X-ray source shows a spectral change of the spectrum emitted on a very low time scale only. As this time dependence is much shorter compared to the total pulse duration of the radiation, it is negligible. This result guarantees that both pulse duration and energy can be adjusted independently. This enables further investigations with this new X-ray tube in the field of pulsed radiation and its use for e.g. type tests. |
