LONG TERM STATISTICS OF X-RAY SPECTROMETERS

Size: px

Start display at page:

Download "LONG TERM STATISTICS OF X-RAY SPECTROMETERS"

Nathan Bishop
5 years ago
Views:

403 LONG TERM STATISTICS OF X-RAY SPECTROMETERS J. F. Dlouhy*, D. Mathieu Department of the Environment, Environmental Technology Center, River Road, Ottawa, Ontario, Canada Kl A OH3 K. N.

1 403 LONG TERM STATISTICS OF X-RAY SPECTROMETERS J. F. Dlouhy*, D. Mathieu Department of the Environment, Environmental Technology Center, River Road, Ottawa, Ontario, Canada Kl A OH3 K. N. Stoev Bulgarian Academy of Sciences, Institute for Nuclear Research and Nuclear Energy, Blvd. Tzarigradsko Shousse, Sofia, Bulgaria ABSTRACT Quality assurance and quality control programs (QA/QC) are the essential components of any analytical laboratory. As a part of the QA/QC program in this laboratory (Environmental Technology Center), an extensive set of data was collected over a long period of time (about ten years), which was used to study the long term stability of two x-rays spectrometers. A Rh anode x-ray tube was used for the excitation of the x-ray fluorescence of two thin film x-ray standards. A Si(Li) detector, cooled to liquid nitrogen temperature, was used for recording the x-ray fluorescence spectra. All measurements were done in vacuum. Three measurement conditions were used in order to cover all elements of interest. The energy resolution (full width at half maximum), the peak position, the whole spectrum intensity, the intensity of several x-ray lines, and the concentration of several elements, were monitored and evaluated. The collected data were used to study the stability of the spectrometers, to account for the changes of the spectrometer response (either by using monitor ratios or by performing new calibration), and to determine the source of the problems with the fluctuations in the spectrometer response. The sample of collected data is presented in two tables and illustrated in graphic form. The results are discussed and some conclusions are made about the correctness of the different ways for the elimination of fluctuations of the x-ray spectrometer response. *) The corresponding author. Fax: (613) , Dlouhy.Joe@etc.ec.gc.ca

$X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD).$ This document is provided by ICDD in cooperation with the authors and presenters of

This document is provided by ICDD in cooperation with the authors and presenters of

All copyrights for the document are retained by ICDD.

2 This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website ICDD Website -

3 404 Two multichannel pulse height analyzers were or are used as X-ray spectrometers [2] in this laboratory (Kevex, model 770/8000 and model 771/8000). Measurement of the accuracy, precision and stability of the spectrometers is a complex problem. The first task is the check of the differential and integral linearity, which is measured using the precision pulser. That usually only checks the amplifier and the analog-to-digital converter. The measurements are nowadays performed in a different fashion. The primary energy calibration is done using a radioactive source (frequently Fe 55). It has the disadvantage that the X-ray tube and the power supply are not included in the test. The secondary calibration is done by using X-ray lines at the high and low position in the spectrum. The manufacturer recommended the use of an aluminium alloy (NBS 2708) containing cooper. The lines measured are CuKa and AlKa. The measurements were done on the energy scale 10 kev. The desired energy accuracy was achieved using an automatic iterative program. The best energy accuracy was 1 ev, which represents one tenth of a channel under specified conditions. The program resets the amplifier gain (slope of the calibration curve) and the zero offset of the controlling helical potentiometers. The parameters were transferred automatically to other energy scales (20 and 40 kev). The energy and energy resolution calibration procedures were extended in this laboratory by evaluating the spectrum of the NIST thin film X-ray standards NBS1832 and NBS1833 (see Table 1) [3], which are measured with each set of specimens, Table 1 Al Si K Ca Ti V Mn Fe co cu Zn Pb NBS1832 b&w NBS1833 [u&m

4 405 Each specimen spectrum was recalculated to achieve full energy scale agreement for all three measurement conditions. The detection efficiency is the second task. It can be controlled by measuring the integral count rate of the proper standard over a given energy interval. That interval could be the whole measured spectrum or an interval which contains the characteristic excitation interval (without analytical lines) or it can be the interval which is used for the analysis of measured specimens for the given condition. The integral count rate is then used to correct the spectra measured under identical conditions (the same day). There is an easy check of the correctness of the procedure by using thin film X-ray standards (NBS 1832 and NBS1 833, which were already measured to be used for the energy calibration). The corrected spectra are processed and the certified concentrations are calculated and the accuracy is established. The initial setting (Window1 in Table 2) of the first multichannel pulse height analyzer (Kevex, model 770/8000) was selected by late Dave Seilstad (Kevex, Head of the Application Laboratory). Table 2 Window1 Window2 X-ray low high low high tube [kev] [kev] [kevl [kevl WI Condition Condition Condition The X-ray tube current was set as high as possible (<=3.3mA, which is the highest available setting) so as not to exceed 50 % dead time for all standards and specimens. The spectrometer was used for nine years. During that period eight X-ray tubes were used. Full calibration had to be done after each change of the X-ray tube. The tubes did have different parameters including the thickness of the beryllium window.

5 406 The calibration constants (ng/cmycps) for each element from Na to Bi were measured using thin film X-ray standards (MicroMatter) with element mass from about 5 ug/cm2 to about 50 ug/cm2. The support film is 6.5 urn Mylar. Over one hundred standards were used. That was complemented by the thin film X-ray standards NBS1832 and NBS 1833, NIST2036 and corresponding blanks (Teflon, Mylar and Nuclepore films).the measurement, calculations and evaluation takes about one month. The procedures set at the beginning were used until Different mathematical analysis [l] partially in parallel to the original one was used since That is the reason that the Figure #l is limited to the period. The monitor ratios for the thin film standard NBS1832 are plotted versus time. There is a new x-ray tube at the beginning of each calibration as illustrated in Figure #l. The intervals with missing points indicate the waiting periods for the replacement x-ray tube or the periods when measurements under different conditions were pursued (eg secondary targets excitation, radiation background etc.). In Calibration 5 and 7 are visible additional sudden changes in the monitor ratios, They are caused by the visit of the service personnel and inadvertent change in the close geometry of the analyzer (X-ray tube, filter bar, sample changer, detector collimator and detector). From Calibration 4 a different pair of the thin film X-ray standards NBS 1832 and NBS 1833 was used. The windows for monitor ratios were changed at the same time (Window2 in Table 1) to cover the interval of the spectrum used for the analysis of the measured specimens. FIGURE #l MONITOR RATIOS NBS1832 CONDITION 2 SPECTROMETER 77OBOOO 1166 DATA POINTS p d E z e I 0.5 l/l/87 l/l188. I89 DATE Figure #l contains monitor ratios for Condition 2 only. It illustrates the changes in each calibration and between different calibrations. The difference in monitor ratios between calibrations is caused by the slight difference in the individual x-ray tubes (the i

6 407 geometry of the tube, the thickness of the Be window, the geometry within the spectrometer, the contamination of the Rh anode etc.) and by the need to adjust the x-ray tube current for each tube to fulfill the condition that the dead time for all standards and blanks (NBS, NIST and MicroMatter) will be for all three conditions under 50%. The monitor ratio is expressed as the ratio of the measured value versus value from the mean spectrum of the same standard measured at the time of the same calibration. The mean monitor ratio for Calibration 6 and Condition 1 is 1.O 15 (relative standard deviation is 2.25%) for NBS1832 standard and (relative standard deviation is 2.19%) for the NBS 1833 standard from 387 measurements. The individual ratios between NBS 1832 and NBS1833 monitor ratios give the mean of with 2.36% relative standard deviation which is statistically identical to the direct values of monitor ratios. Some outliers in the Figure were not removed to illustrate possible differences. Most often the outlier has a higher value than other values in the set on account of the failure of the spectrometer to advance to the correct position for the NBS standard. That leads to the inclusion of the massive Al holder line in the spectrum and corresponding increase in the monitor ratio. Same data for the second spectrometer (Kevex, model 771/8000) are presented in Table 3. Table 3 Standard Deviation Monitor Ratio (NBS1832/NBS1833) 0.33 % Condition 3 FWHMRatio (NBS1832/NBS1833) 1.55 % Condition 2 Si Ratio (NBS1832/NBS1833) 0.46 % Condition 3 Cu NBS1832 Ratio 2.07 % Condition l/condition 2 Pb NBS1833 Ratio 0.76 % Condition l/condition % Condition 2 NBS1833 (Fe/Pb) 0.40 % Condition 2 The relative standard deviation of the monitor ratio for NBS1832 standard and condition 3 is 2.04 %. The ratio of the monitor ratios of both standards is 0.33 %. That represents almost an order of magnitude improvement. The Table contains the ratios of different parameters after normalization and proves the stability of the spectrometers together with the stability of the thin film X-ray standards NBS1832 and NBS1833 across different conditions and different parameters. It is not possible to separate the contribution of any individual standard or spectrometer on the basis of available data. The values of the relative standard deviation indicate acceptable quality of data.

7 408 CONCLUSIONS The applicability of two thin film X-ray standards for the work in this laboratory was demonstrated. It reduces the need for frequent full recalibration of the spectrometer. It takes care of short and long term stability of the spectrometer response by using only the linear corrections. It spans the change of the integration intervals for the monitor ratios. The change of the X-ray tube leads usually to the non-linear changes in the spectrometer response and is not presented here. It proves the long term stability of the thin film X-ray standards NBS1832 and NBS1833. LITERATURE [l] K.N.Stoev, J.F.Dlouhy, X-RAY SPECTROMETRY: 23, 112-l 19,1994 [2] K. Siegbahn, Alpha-, Beta- and Gamma-Ray Spectroscopy, Vol. 1, 1969, North Holland Publishing Co. Amsterdam. [3] NIST is National Institute of Standards and Technology, Gaitersburg, MD, USA. Both very important standards are out of stock. More general aspects of x-ray spectrometry are included in E.P.Bertin, Principles and Practice of X-Ray Spectrometric Analysis, 2nd Edition, 1979, Plenum Press, New York. where there is a part on page 323 and ff. on multichannel pulse height analyzers or in R.E.VanGrieken, A. A.Markowicz, Handbook of X-Ray Spectrometry, 1993, Marcel Dekker, New York. Some aspects of QA/QC are covered in F.M. Garfield, Quality Assurance Principles for Analytical Laboratories, 1988, AOAC, Arlington.

WIDE ANGLE GEOMETRY EDXRF SPECTROMETERS WITH SECONDARY TARGET AND DIRECT EXCITATION MODES

WIDE ANGLE GEOMETRY EDXRF SPECTROMETERS WITH SECONDARY TARGET AND DIRECT EXCITATION MODES Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 11 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42