DETERMINATION OF FLUORESCENCE YIELDS USING MONOCHROMATIZED UNDULATOR RADIATION OF HIGH SPECTRAL PURITY AND WELL-KNOWN FLUX

Transcription

1 Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol ISSN DETERMINATION OF FLUORESCENCE YIELDS USING MONOCHROMATIZED UNDULATOR RADIATION OF HIGH SPECTRAL PURITY AND WELL-KNOWN FLUX B. Beckhoff and G. Ulm Physikalisch-Technische Bundesanstalt, AbbestraBe 2-12, D-l 0587 Berlin, Germany ABSTRACT Along with the increasing importance of material and surface analysis by means of X-ray fluorescence analysis (XRF) in the soft X-ray range, the need for standard-free quantitication modes has grown. Here, improving the knowledge of accurate values of the fluorescence yields for photon energies below 2 kev becomes a major task. For this purpose, an irradiation chamber was built by the Physikalisch-Technische Bundesanstalt (PTB), Germany s national metrology institute. This UHV chamber allows an accurate positioning of up to 6 specimens with respect to a set of high purity Al diaphragms defining exactly the excitation beam size and the solid angles of radiation detection. The scattered and fluorescence radiation was registered by a calibrated semiconductor detector. The monitoring of the monochromatized excitation radiation by means of calibrated photodiodes was ensured. For the specimen excitation, the plane grating monochromator beamline of the PTB for undulator radiation at the electron storage ring BESSY II, involving a high spectral purity, was used. In the present investigation, the fluorescence yields associated with the K-edge of boron and of carbon were determined for different mass depositions of the respective elements. The results of the initial experiments, including the achieved accuracy, are given. INTRODUCTION AND METHODICAL CONCEPT The more sophisticated applications of energy-dispersive X-ray fluorescence analysis (EDXRF) in the soft X-ray range, such as material and surface analysis, require improved accuracy of fundamental atomic data for standard-free quantification. With respect to the latter, improving the accuracy of both the absorption and fluorescence cross section values in the energy range below 2 kev becomes a major task. The present work focuses on the K-shell fluorescence yields of boron and carbon. Nowadays, synchrotron radiation is used to investigate fundamental atomic processes. Synchrotron radiation of high spectral purity is an efficient tool to determine not only fundamental atomic data with a high accuracy, but also the efficiency and response function of X-ray detectors. Well-characterized radiation sources and detection systems are crucial for the measurement of fluorescence cross sections. The plane grating monochromator beamline [l] for undulator radiation in the PTB radiometry laboratory at BESSY II provides monochromatic radiation of high spectral purity. By accurately measuring the radiant power of monochromatized radiation with an electrical substitution radiometer and controlling the stored electron heam current, photodiodes and energy-dispersive detectors [2] can be calibrated.

2 ISSN 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 Centre Centrefor fordiffraction DiffractionData Data2001,Advances 2001,AdvancesininX-ray X-rayAnalysis,Vol.44 Analysis,Vol.44 ISSN monochromatic radiation I,,(! *1 : * * I 11i, sprcimcrl,-, It& :) ftr( i :o) L transmission measurements: p(s.lj t = -In I Itr (:,)/lo(:; radi.&)n,) ] p(eo) t = - In [ Itr (I+:( FiFurc I: Determmauon of the absorption correcr~on factor Af~~i(E ) by means of expermlemar measurements at the photo11 energies; of both the exciting and fluorescence radiation of interest transmlsslon Our approach for the measurement of lluorescence yields is straightforward in determining almost all relevant parameters from the specimen measurement instead of using any tabulated values. Apart from the respective measurements only the fact that at low photon energies coherent and incoherent cro4s sections are considerably smaller than the photoelectric absorption cross section is used as an external information. The relation for the detected count rate.v,, of the fluorescence radiation of the element I having the photon energy EY, is 10 as the photon flux of the incident excitation radiation having the energy E,, ( / as the concentration of the element I in the specimen qie \,1 = sill-,/sirl-,,/!!/i;;f,,,,(~,,) as the geometry factor including the detector efficiency ccl,,i E) and -, vir. 7L being the angles of incidence and of observation l/,,(e) = (1 - q,(1~/(/r. j,,/sirl- I - I,,[:, /sin-2)))/(//,,/,, -t p5 l:,,sin-,/~iu-~) the absorption correction factor (cf. fig. 1) involving the total absorption cross section I/~. i oi the \peclmen A1\, a\ the fluorescence yield of the shell (subshell) S/ to (J\i I a\ the probability of eml\\ion of the line 1 T\, /, a\ the photoelectric cro\\ section for.\: I be determmed as

4 Centre Centrefor fordiffraction DiffractionData Data2001,Advances 2001,AdvancesininX-ray X-rayAnalysis,Vol.44 Analysis,Vol.44 ISSN *hi-..:; #lr\... HPGe-detector Si ( 10rnrn2) f--) f calibrated SQhotodiodes for the inten@& and tranrniftted did- A & measurement of direct excitation radiation gfi=2mm phragm CCD Figure 2: Arrangement for the experimrntrd determmatlon of fluorescence yields coefhcient\ In the soft X-m> rang offered by the PTB XRF irradiation chamber and of linear I~;IW ab\orptlon According to Hubbell et al. [ 31, the values of the fluorescence yields of I<- and L-shells associated with absorption edges above about I.S kev are well established with relative uncertainties that range from about IO % for % = 1I to about I 5%for % = 80 and above. The main contributions to these uncertainties are due to uncertainties of the various cross sections used and the detection efficiency, regardless which specific method is applied. At photon energies lower than I.S kev viz. lower atomic numbers. relative uncertainties of the fluorescence yields even larger than IO 5%are quoted in the literature. Hence, all relevant cross sections, i.e. the mass absorption coefficients, were to be determined during the present investigation. Furthermore, the photon flux of the exciting. transmitted. and fluorescence radiation had to be determined by calibrated photodiodes and by energy-dispersive Si(Li) or HPGe detectors. In addition, thin specimens were to he cxcited by monochromatized radiation of high spectral purity and a very well-known beam geometry had to be defined. For the latter purpose, an irradiation chamber was built in the PTB laboratory at BESSY I

5 Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol ISSN This UHV chamber allows an accurate positioning of up to 6 specimens with respect to a set of high purity Al diaphragms defining exactly the excitation beam size and the solid angles of radiation detection. Furthermore, for the sake of measurement accuracy, the scattered and fluorescence radiation can be registered simultaneously by up to 2 energy-dispersive detectors, which were calibrated at photon energies of interest. A schematic sketch of the new irradiation chamber is shown in fig. 2. Using an entrance diaphragm of 1 mm inner diameter, a beam profile of about 1.1 mm diameter is defined at the specimen site. At a distance of 67 mm and under an angle of 30 with respect to the exciting beam, a diaphragm made of high purity aluminum is positioned. The area of the high purity Al diaphragms [4] was determined by a comparison with a reference Al diaphragm using calculable radiation of a BESSY I bending magnet. As the entire diaphragm holder system itself was made out of only one piece, by means of CNC technique, a very accurate positioning of the entrance and detection diaphragms is ensured. Hence, an accurate beam geometry in the irradiation chamber is defined. The contribution of each parameter in the relation for Nxi, as given in equation (1), to the total relative uncertainty Ati.\.i/ti.xi of the fluorescence yield wsi may be quoted as follows. The incident photon flux is well known with a relative uncertainty 41a/10 = lo-, whereas the fitting of XRF spectra with experimental detector response functions contribute with &\rsi/~\r~i equals p2. Transmission measurements at various K-edges can result in ACi/Ci = 11.O lo-. With respect to y(esi) the so.lid angle of detection has a relative uncertainty AR/62 = 7.0 lo-, and the detector efficiency uncertainty &det/edet ranges from 1.5 lo- (above 300 ev) up to M2 viz. 0.0 lo- for the present photon energies below 300 ev. The absorption correction factor has a relative uncertainty nalxi/mxi = e2 including the replacement of ~~~~ by rsi. Hence, the total relative uncertainty ~W,Y~/LJX~ of the fluorescence yield wsi ranges from 5.2 lo- to 9.9 lo- in the present work. EXPERIMENTS AND RESULTS Different kinds of specimens, such as polyimide (PI) foils or boron layers deposited on PI, were used in the present work. The main ob.jective was to provide one-element or compound specimens whose transmittance is high enough to ensure a good statistic of the transmitted beam. In general, this involves a transmittance between unity and lo- in the energetic range of interest. According to the measuring strategy depicted in fig. 1 the absorption correcting factors were determined. The transmission measurements were conducted at 183 ev and 267 ev for boron, and respectively at 277 ev and 474 ev for carbon. The thin specimens of interest are excited by monochromatized undulator radiation of high spectral purity at 267 ev for boron excitation and at 474 ev for carbon excitation. The fluorescence radiation originating from the specimens is registered by the calibrated Si(Li) detector. A dead-time correction was performed. Not only the energy dependence of the counting efficiency of the Si(Li) detector is known but also some of its experimental response functions at selected photon energies of interest. Instead of fitting fluorescence lines in an XRF spectrum by means of any mathematical function these experimental detector response functions are used in the present investigation.

6 Centre Centrefor fordiffraction DiffractionData Data2001,Advances 2001,AdvancesininX-ray X-rayAnalysis,Vol.44 Analysis,Vol.44 ISSN nm B on 115 nm PI foil *i s<( ---- B-Kn detector scattered response exciting and C-Kn., N-Ka radiation + O-KU -I i energy I ev Ftgure 3 X-ray Huore~cencc q~ectrum ot ;t thtn boron layer depostted on it thtn poly~rn~tlt: (Pi) lot1 th tt w,t\ excttecl with monochrom ttrc nt~luttton ev photon energy. The XRF spectrum w;t\ httecl hv III~UI\ of,iii expertmentally cleternunc~l cletector req~on~e tunctton at I X3 ev. Substractiny the adapted detector re\pon\c eyunalent to D - K,, from tlic nic,tstirzd spectrum. ~11 PI component\ appear. 115 nm PI foil (C22H,oNz0,) C-Ko detector scattered and energy Figure 4: X-ray fluorescence of 474 ev photon energy. response function at 277 cv. spectrum. the PI component N-Ko response exciting radiation + 0-Ka I ev spectrum ot i~ thin polyimicle (PI) foil that was excited with monochrom;ttic ntcliation The XRF spectrum wits fitted by means of an experimentally tleterminecl detector Suhstntctin~ the irdapted detector response equivalent to ( - Ii,, from the measurecf N ;tncl ;I xmall ( -I\; pile-up peak appear

7 Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol ISSN The energy of the respective exciting photons is selected so that it is considerably higher than the range of any possible XANES or EXAFS regime above the corresponding absorption edge associated with the shell of interest. Before and after the fluorescence measurements, the absolute photon flux of the excitation radiation is registered by means of calibrated photodiodes with a relative uncertainty of about While recording the XRF spectra, as depicted in fig. 3 and fig. 4, the photon flux of the excitation beam transmitted through the specimen is also monitored by means of a calibrated photodiode. As the characterization work of the plane grating monochromator beamline had not been completed, the effectiveness of higher-order suppression was different at the two selected excitation energies. Hence, a small 0 - I<, peak can be still seen for the 267 ev excitation whereas it vanishes for the 474 ev excitation. The fluorescence yields obtained in preliminary XRF experiments (cf. fig. 3 and fig. 4) are for boron wfi = i and for carbon wic = f Comparing these results with the corresponding reference values [3] for boron wl( = and for carbon (A)/< = , one sees that the values are in agreement within three times the uncertainty bars quoted experimentally and that no systematic deviation has to be stated. Due to a rather limited set of initial experiments the present investigation shall be repeated in order to exclude any possible deviation associated with the selected beam geometry, the incident beam quality, the specimens or the calibrated radiation detectors. Furthermore, the respective relative uncertainties achieved in the present work are to be reduced, too. Taking precautions such as an additional control of the photon energy scale of the Si(Li) detector using the one of the monochromator beamline, it should be possible to reduce the relative uncertainties of the detector efficiency ~F~,~/F~~+ down to the typical 1.5 to e2 range which is achievable at higher photon energies. Hence, the relative uncertainties of the fluorescence yields could be decreased below p2, allowing e.g. for standard-free XRF quantification of minute amounts of low Z elements deposited on heavier substrates. Here, the relative uncertainties of the determined low Z mass depositions are expected to be below lo-, which is of particular interest when appropriate reference specimens are not easily available. ACKNOWLEDGEMENT The authors would like to acknowledge the excellent fabrication and coating procedure of the thin polyimide films by J. Herbst, FE Rbntgenphysik, University of Gottingen. One of the authors (B.B.) would like to acknowledge all the inspiring and fruitful discussions about measuring fluorescence yields with B. KanngieBer, IAAP, Technical University of Berlin. REFERENCES [l] ESenf, EEggenstein, U.Flechsig, W.Gudat, R.Klein, H.Rabus, and G.Ulm, J.Synchr.Rad. 5, ( 1998) [2] B.Beckhoff, R.Klein, M.Krumrey, EScholze, R.Thornagel, and G.Ulm, Nucl. Instr. Meth. A 444, (2000) [3] J.H.Hubbell, P.N.Trehan, Nirmal Singh, B.Chand, D.Mehta, M.L.Grag, R.R.Grag, Surinder Singh, and S.Puri, J.Phys.Chem.Ref.Data 23 (2), (1994) [4] H.Rabus, B.Beckhoff, and R.Thornagel, BESSY Annual Report, (1999)