X-ray Detector Response Mechanism And Applied Modeling Technology

The detection and application of X-ray plays an important role in materialcomposition, structural analysis, and in other studies. One of the most importantbranches of the application is Energy Dispersive X-ray Fluorescence (EDXRF)technology. EDXRF faced a main problem that how to improve on the analysisaccuracy and automation level during the technology developing, and spectrum andcontent analysis are two key breakthrough points. First of all, accuracy spectrumanalysis technique is the important premise to ensure EDXRF calculate accuratecontent. Spectrum analysis can be solved via establishing detector response function(DRF), which can improve the automation level and computational accuracy ofspectrum data processing. The DRF can analyze the mathematical character andphysical character of X-ray fluorescence spectrum, its functional form is mainlyrelated to the type of ray, the energy of ray, the type of detector, so, to establish anoptimized DRF model with strong generality is very essential. Secondly, study ofquantitative analysis technology is the core content regard to EDXRF technologylevel. Traditional quantitative analysis method influenced the analysis speed ofEDXRF technology, because they ether relied on the accuracy of standard sample, thesimilarity between standard sample and sample to be analyzed, or need verycomplicated parameter calculation and abundance mathematical operation. So, whencarrying the study of EDXRF quantitative analysis method, we especially need toconsider its features like portable, flexible and rapid, that is we can adopt simple andfast method to get quantitative or approximate quantitative analysis data during theallowed range.For these two key problems mentioned above, this thesis, based on in-depthstudy on previous works of others, centered on X-ray detector response mechanism and applied modeling technology, established the computing method of X-ray fullenergy peak standard deviation （E）, the detector response function model withstrong generality, and proposed a full spectrum quantitative analysis method for thefirst time. Methods established in this paper are test and verified their validity viavarious practical applications. The main research contents and conclusions are:(1) A kind of discretization statistic computing method is established to calculateσ(E). This method, with simple computational process and practical characteristic,used the distribution law and standard deviation in discretization random process, andthe result of σ(E) can be long-term used even when the measuring conditions changedin a permissible range. The relation curves between σ(E) and E is obtained of silicondrift detector (SDD) and silicon PIN (Si-PIN) detector during the X-ray energy rangeof4.5keV-26keV, fitting functions are also obtained to analysis the inner relation ofto Fano factor F and ionization energy, which are compared with other studyresults. The calculated results of （E） is the most import parameter in detectorresponse model, which has been well used to the DRF model and full energy spectrumfitting procedure.(2) A detector response function model with strong generality has beenestablished for SDD and Si-PIN detectors. Based on in-depth study on foreignprevious research achievements, the author established a set of generality DRF modelfor SDD and Si-PIN detectors via the analysis of X-ray physical detect procedure. Themodel contains four part: a flat continuum shift from zero to full energy, S(E,Ek);Gaussian shaped principal component of full energy peak, G(E,Ek); Gaussian shapedSi escape peak, ES(E,Ek-1.74); exponential tail on the low energy side of the fullenergy peak, T(E,Ek). The model, with simple format, has clear and definite physicalmeanings for each part. Parameters are achieved by the program, developed by theauthor, based on Weighted Least-Squares (WLS) method. Nineteen elements’ X-rayspectra in total were fitted in the study, and the fitting goodness factors2rwere alsoobtained simultaneously. The results are shown that this DRF model can representcharacteristic X-ray spectrum peak shape very well. The DRF model is also used tofundamental parameter analysis method to get accurate peak net area of elements. Sixelements of Fe, Ni, Cu, Zn, Pb, Sn in nine copper alloy samples are analyzed, therelative error of Cu and Zn is in the range of0.04%-7%, while the relative error aremuch larger for Fe, Ni, Sn, Pb which have low content in samples. This result is equalto or better than other analysis methods, and it also can meet the need of industryapplications. So the DRF can be illustrated to be validity and applicability. (3) A kind of full spectrum quantitative analysis method has been established forEDXRF technology. This method is referred to the theory of weighted least-squaresand library least-squares, and used the DRF established in former and full spectruminformation of single element to reach the aim that analyze element content in a fastspeed. This method regards that if a sample contains several elements then thecomposite spectrum recorded in the analyzer due to all the X-rays which strike thedetector is just a linear combination of the response functions corresponding to eachX-ray present in the spectrum. The intensity of a given X-ray is proportional to thetotal number of counts in its mono-energetic spectrum which contribute to thecomposite spectrum. The computational process contains initial content guess forelement, quantitative analysis model, iterative algorithm etc. This method overcomesthe shortages of traditional methods, such as large computational work, need lots ofstandard samples which are difficult to be prepared, so this method is much faster andpractical to carry quantitative analysis. Full spectrum quantitative analysis method isused to analyze Fe, Ni, Cu, Zn, Pb, Sn in nine copper alloy samples, the relative errorof Cu（56%-65%）and Zn（30%-42%）is in the range of0.1%-5%, while the relativeerror are much larger for Fe, Ni, Sn, Pb which have low content in samples. Thisanalysis result is similar to fundamental parameter method, which can demonstratethat this new method is stability and validity preliminarily. The full spectrumquantitative analysis method, as compared to traditional method, used the full energyspectrum information, and it has features of few calculating parameter, fastercomputation speed, and flexibility and practical applicability.Some applications are studied at last, which mainly contains the application ofDRF modeling theory to heavy charged particle spectrum analysis (Alpha particleanalysis), DRF model application to Monte Carlo simulated flux broadening, fullspectrum quantitative analysis method application to Si and Ca light element contentanalysis measured by Super fast SDD detector. The Au-Si surface barrier detectorresponse function for alpha particle is established, and overlapped spectra of alphaparticles from239+240Pu source are resolved. Alpha spectra, measured under differentrelative air pressure, are fitted out, and the variation trend of energy standarddeviation, peak position is consistent with the theoretical prediction. A newbroadening algorithm is established for the simulated flux spectrum obtained fromMonte Carlo simulation, and pulse height spectrum (PHS) can be transformed bydetector response function, which can make the simulated flux near to the realexperiment spectrum. The copper alloy sample (No.RC102) is taken into the simulation and broaden procedure, and we get a relatively good result. Full spectrumquantitative analysis method is used to analysis Si and Ca light element content,measured by super fast SDD detector, and the analysis results demonstrate that thismethod is available analyze element content preliminarily.This thesis has four innovation points:(1) Akind of practical calculation method, which is based on discrete statistics, isestablished to calculate X-ray full energy peak standard deviation σ(E).(2) A kind of generality detector response function is established to SDD andSi-PIN detector employed by EDXRF analysis system.(3) A new broaden algorithm for Monte Carlo simulated flux spectrum isproposed based on DRF model.(4) A simple but practical full spectrum quantitative analysis method isestablished for EDXRF technology.Above all, this thesis, based on amount of experiments and comprehensivesumming-up existed technology, established simple but effective (E) calculationmethod, established a strong generality DRF model, and accordingly, establishedMonte Carlo simulated flux spectrum broaden algorithm, finally established apractical full spectrum quantitative analysis method. These studies make someexploratory words in earlier stage for the automation tenor of EDXRF technology, andalso accumulated some research experience which has reference values to someextent.