Font Size: a A A

The Research Of Spectral Processing Method In Energy Dispersive X-ray Fluorescence Spectrometer

Posted on:2011-12-26Degree:MasterType:Thesis
Country:ChinaCandidate:W XuFull Text:PDF
GTID:2178360305454899Subject:Computational Mathematics
Abstract/Summary:PDF Full Text Request
First,this paper introduces the present state and prospects of EDXRF,then introduces the basic theory of EDXRF.(1)X-fluorescence Substances are all constituted of atoms. Each nucleus are surrounded by a number of electronic. Layers of every elements'the extranuclear electron is different from others'.So is the number of electrons on each floor, the flight track shape, and the radius. All these differences formed a variety of electronic energy. When using X-ray irradiation of material, it will hit the inner rail material electrons on a trip into the hole. You need to capture an electron from the outer than to fill the hole, which is signed not steady-state until the atoms return to steady-state.But, the outer electrons with the energy is higher than the inner electrons.and when the transition occurs, the extra energy will be released in the form of X-rays.Such X-ray is called X-ray fluorescence.As the electronic energy levels of atoms of each element is inherent, so X-fluorescence produced by it is inherent. Characteristics of the substance are known as X-rays.(2)X fluorescence excitation source, need to be stimulated by a higher photon excitation energy. In general, we chose to use X-rays to stimulate it. Of course, high-speed electronics and ions can also be used.(3)X-ray fluorescence spectroscopy Substance is made up of one single element or mul-tiple elements. When the X-ray irradiate to the sample, all element in the material will emerge X fluorescence. They mixture with each other.There are variety of methods of separating the various materials'X-fluorescence of others', two of which are commonly used,named wavelength X-ray fluorescence spectrometer, which is through the analyzer crystal diffraction of X-fluorescence of different wavelengths to achieve spectraling of pur-pose,and energy X-ray fluorescence spectrometer,which is through the probe light signal into electrical pulses, after amplification,the use of multi-channel pulse height analyzer (MPHA) for signal processing, which are contained in the elements of X-fluorescence spectrum.(4) Energy X-ray fluorescence spectrometer need to use the detectors to detect X fluo-rescence, and translate them into electrical pulses. There are NaI crystal scintillation detector detectors, inflatable proportional counters, semiconductor SiPIN detectors, high-purity sil-icon detector,high-purity germanium crystal detector and so on. The main quality of the detector performance are its own detection limit of fluorescence detection, resolution and detection energy range decisions, etc.(5)the qualitative and quantitative analysis of X fluorescence spectrometry The qualita-tive analysisof the spectra collected is to judge the elements contained in through the spectral peaks. Quantitative analysis is to use of a method to find such a relationship means of each element concentration of the sample and X-fluorescence spectrum in the peak count of in-tensity correlation. This is precisely the issues to be discussed in this paper.The work done of this article:(1) de-noising of spectrumusing wavelet analysis theory, the signal is decomposed into high and low frequency components, part of high frequency is set to zero,so as to achieve the de-noising effect on the spectrum The following are algorithm steps of smoothing de-noising.a.On the original signal f0 do j (j= 3) layer of wavelet decomposition,we obtain low-frequency part of largest scale Cj(n) and high frequency components Di(n)(i= 0,1,2);b. To remove high-frequency component Di(n)(Di(n)= 0, i= 0,1,2);c. Reconstructing signal.(2) Background processingClayton et al (1987) proposed a method,comparison of yi and The average of two near address mi. if yi is bigger than mi,yi=mi.On the contrary, the same.Beacuse of the times of iterative is too more,so speed of this method is too slow,we make a little change on it.Comparison of yi and The average of forty near address mi. Beacuse of this change,speed is improved. But the background of this time is not smooth,so comparison of yi and The average of two near address of the background mi,if yi is bigger than mi,yi=mi.On the contrary, the same.And the background of this time is what we need.(3) Find peaksGood peak searching method should meet the following basic requirements:a. have a high resolving power of the overlap peaks, can determine peak position of near peaks.b. to identify weak peaks, especially the weak peaks in high background.The peak is divided into two categories:a. obvious peaksb. inflection point, which is made of overlap peaks which are very close.First find the spectrum of first-order differential d (i), for one i, if d (i)>0, d (i+1)<0, then i requested for our first peak.Solely in accordance with the definition of turning point to find out second peak, there will be many false peaks, the results are poor. After studying the first derivative signal. We found that the valley position of positive part of first derivative and the peak position of negative part of first derivative are the inflection point which we need to find.(4)spectral fitting and Spectral decomposition (The LM algorithm has been applied to the X fluorescent detector spectral analysis,in this paper we suggest a new method of select-ing initial value)The main algorithm steps are as follows:a. section of entire signal.b. the i signals using nonlinear least squares fitting, using Levenberg-Marquardt method.If the i signal has k peaks, the form of function of fitting is: For each Gaussian function corresponds to three parameters, then there should be 3k a initial value, three initial value of each peak of each section signal we selected as:bj is finded by method of finding peaks,ajis signal value ofbj,that is aj= y(bj),cj is decised of , Hj is half-height width of bj corresponding gaussian function.c. join together each fitted signal into the whole signal(5)Data processing and calculate result...
Keywords/Search Tags:energy dispersive, XRF, background, Gaussian fitting, wavelet analysis
PDF Full Text Request
Related items