Beam Interactions with Materials 8 Atoms
ELSEVIER
Nuclear
Instruments
and Methods
in Physics
Research
Quality assurance of environmental
B 136138
(1998) 981-985
PIXE analysis in Prague
Z. Nejedl$ ‘,*, J. Kr&l a, J. Voltr a, R. Krejei b, E. Swietlicki ‘, J. cernf J. Svejda d
d, P. Kubik d,
a Ion Beam Laboratory, Department of Physical Electronics. Faculty oj’ Nuclear Science and Physical Engineering, Czech Technical University, V HoleSovi?krich 2, 180 00 Prague 8. C:ech Republic b Depurtmrnt oj’Environmenta1 Geochemistry. Czech Geological Survey. Kkiroc 31131, 118 21 Prague 1, Czech Republic ’ Division of Nuclear Physics, Physics Department, Lund Institute oj’ Technology. Box 118. S-221 00 Lund, Sweden ’ Nuclear Centre. Faculty oj’ Mathematics and Physics. Charles Lfniversity. V HoleSovi?kLich 2. 180 00 Prague 8. Czech Republic
Abstract The paper describes calibration, laboratory Quality Assurance/Quality Control (QA/QC) program, and environmental applications of the Prague PIXE system. To detect, correct, and prevent problems in the measurement process, an intercalibration between the Ion Beam Laboratory of the Czech Technical University and the Aerosol Group at the Lund Institute of Technology has been carried out. Seventy-eight rural fine fraction air particulate samples collected on Nuclepore filters were analyzed first in Lund and then in Prague. In case of the most abundant trace elements like S, K, Ca. Fe, Zn, the comparison of elemental concentrations measured by the two laboratories revealed small random error and bias, which stayed below lo”/ of magnitude. The limits of detection obtained in Prague and in Lund were comparable and differed only slightly at extremes of low and high Z. Further applications of PIXE in Prague in environmental science are mentioned including characterizations of trace elements in lichens by an external beam. 0 1998 Elsevier Science B.V.
PACS:
82.8O.Ej; 92.6O.S~; 41.75.Ak PIXE; Aerosol; Lichen; Air quality; Air particulate; Ion beam analysis
Keywords:
1. Introduction
The first ion beam analysis on the present systern in the Laboratory of Ion Beams at the Czech Technical University (CTU) in Prague was performed in 1989 [l]. Since then, proton induced X-ray emission (PIXE) [2] analysis of various sam-
*Correspondingauthor.
Fax:
+42
02 688 4818;
e-mail:
[email protected]. 0168-583X/98/$19.00 0 1998 Elsevier Science B.V. All rights reserved. PIISO168-583X(97)00746-5
ples [3-51 including, e.g., optical crystals, animal tissue samples, human bones, or liquids has been carried out. The PIXE system [3] is built around the 2.5 MeV van de Graaff accelerator of the Nuclear Centre of the Charles University. It has been designed for analysis both in vacuum and with an external beam [4] at low or ambient pressure, e.g., in a helium atmosphere. This report focuses on the technical aspects of PIXE analysis of environmental samples and only illustrative examples of some applications are listed here.
2. Calibration and stability of the PIXE system Characterization of the Prague PIXE system is carried out in two stages. First, the system is calibrated using a set of twenty single- and double-elemental standard targets by MicroMatter [6] certified at a level of +5%. The instrumental calibration parameter H [7] is then derived from this set. After the calibration curve is confirmed, only a limited number of standards are regularly analyzed during the analytical run. In case of thin targets, we have selected the MicroMatter CuS, and KCl. which span the X-ray energy region from 1 to 8 keV. The stability of the calibration measured on these two targets is generally within &2’%. Randomly selected aerosol samples are repeatedly analyzed as a further step on Level 1 Quality Assurance (intra-laboratory quality assurance (QA) procedures), as defined by Cahill in [2]. In case of the inter-comparison discussed below, the reproducibility of the trace element concentrations was between 0.5% and 2% for those trace elements, that were present in relatively high concentrations, e.g., S. K. or Fe. The reproducibility stayed in the l-5% range in case of Ca, Zn, and Cl, which carried higher statistical/fit error of about l-3%, and was within l-8% for Si, Ti, Cr. and Mn (statistical error between 5% and 8%). The total error associated with the elemental concentrations is a combination of the statistical/fit error, calibration error, uncertainties in the database used to deconvolute the spectra, etc. It is about 7% for those elements with low statistical error but it worsens significantly for elements present near their limit of detection. The Prague PIXE spectra are processed by the GUPIX software package [7]. Various software utilities have been created to simplify the spectra evaluation process. As an illustration, a program called PIXIN, which helps the user to maintain a database of the standards with initial parameters required by GUPIX, and to speed up processing of their spectra.
3. PIXE intercomparison
on aerosol samples
The Level 2 QA [2] included an intercomparison between the CTU and the Aerosol Group at
the Department of Physics, Lund Institute of Technology [S]. The project involved PIXE analysis of 78 samples of fine air particulates. The total aerosol load determined by gravimetric analysis was mostly between 10 and 100 ug/cm’. The air particulates were collected on Nuclepore membrane filters, 25 mm in diameter, with a 0.4 urn pore size, catalog No. 110630. First. the samples were analyzed in Lund with a 2.55 MeV proton beam, 8 mm in diameter, 0.2-1.0 nA/mm’ current density, and 5-20 uC charge. Then. they were transported to Prague and selected samples were re-analyzed in the Ion Beam Laboratory at CTU. The first set of 50 filters was measured at 2 MeV proton energy with a 6 mm diameter proton beam, 4415 nA current and about 0.2-0.7 nA/mm’ current density and 4415 uC charge. identification of outliers was the first step in the intercomparison. A set of 16 samples measured consecutively revealed significant differences in the pair-wise comparison: elemental concentrations of some metals including Fe, Zn, Cu measured in Prague were in some cases by 50-100% higher than those measured in Lund as shown in Fig. 1. On the other hand. concentrations of S, Ca, and K in these samples agreed very well ~ within a few percent (Fig. 1). Also, a regular calibration audit conducted within this batch of filters agreed well with the expected values. Therefore, neither the charge measurement nor an inadequate characterization of the PIXE system was the cause of the problem. Detailed examination of both Prague and Lund spectra did not reveal any problems associated with data processing. In an attempt to resolve this issue. selected samples were reanalyzed first in Prague and then also in the Guelph Scanning Proton Microprobe laboratory [9]. Both experiments confirmed the data obtained in Prague. The causes of this discrepancy including eventual contamination of the samples will be further investigated. Comparison of the first set of samples revealed also a bias in the titanium data: concentrations measured in Prague were on average 50% higher than the data from Lund. The Prague PIXE system used a 50 urn Kapton X-ray filter at that time and. therefore, the sulphur peak dominated the spectrum. The disagreement was caused by the sulphur pile-up, which overlapped with the character-
Z.
2”O””
NejedJti
et ul. I Nucl.
Instr.
umi Meth.
31”
----
in Phys. Rrs. B 136-138
1 .
Zinc
Sulphur 400
**
l-_---l :’
-i &O? ,.
200
0
ET &
H
100
0
Lund [ngkm’]
:’
:’
300
2 s
.
.’
.:
.$
,&\’
\
%
983
(1998) 981-985
a0
0
.’
1’
**
0..
100
200
4””
300
500
Lund [ngkm2]
Lund [ngkm*]
Fig. 1.Linear regression analysis of areal concentrations of S. Zn, and Pb measured at CTU Prague and at the Lund Institute of Technology on a set of 78 Nuclepore aerosol filters. The 16 outliers discussed in the text are circled. Linear regression constants and slopes apply to the remaining
62 samples
only.
Ti peak. To decrease the intense sulphur Xrays and the low-energy background, the Kapton absorber was substituted by a “funny” filter [8]. Our funny absorber consists of 46 ym beryllium foil and 260 pm Mylar with a small aperture (about 3% of the detector active area). Although the new limits of detection (LOD) of S, K, or Ca were higher than before, the LOD of Al or Si as well as the LOD of heavy elements like Fe or Zn were improved. Fig. 2 compares LODs of both Prague and Lund PIXE system. The second set of 28 filters was analyzed with this new funny filter at a proton energy of 2.3 MeV, approximately 1 nA/mm’ current density, and IO-20 PC proton charge. Fourteen trace elements (P, S, Cl, K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, Br, Pb) were detected above the limit of quantification in more than 50% of the samples and, therefore, we focused only on these elements. Comparison of the Prague and Lund data is summarized in Table 1. Linear regression results for selected elements are shown istic
=
1
-___-
5
13 15 17 19 21 23 25 27 29 31 33 35 82
-1
Element atomic number Fig. 2. Comparison of limits of detection applicable to air particulate samples for both the CTU Prague and the Lund PIXE systems. Experimental parameters are discussed in Section 3.
in Fig. 1. The set of the 16 outliers mentioned above was not included in Table 1. The trace elements present at concentration levels significantly above the quantification limit (e.g. S, K, Fe, Ca, and Zn) agreed very well. The comparison showed a very small bias of only a few percent and a small random error - the elemental concentrations were highly correlated. The scatter of concentrations of some other elements is greater partially due to increasing statistical error.
Table 1 Comparison between the Lund Institute of Technology and the CTU Prague. Stat is the average fit/statistical error of the Prague data. A is the averaged difference (Prague - Lund) in pairwise comparison. r is the correlation coefficient for the 62 samples s
K
Fc
Ca
Zn
Cl
Mn
Ti ,’
Ni
cu
Pb
Stat A
1‘%I 0%
l’l/(1 9%
1% 9%
2% 2%
3% 4’%1
6% 15’%1
8’Y0 131%
10% 14%)
0.999
0.999
0.991
0.998
0.99
0.386
0.987
0.998
15% 13% 0.954
16%
r
15% 9% 0.977
J Titanium
data apply only to the second
set of 28 samples
I 0% 0.978
4. Environmental
applications
the process of low-Z element losses described by Reis [l 11.
The laboratory presently runs two air particulate monitoring projects where PIXE is a primary analytical tool. The first program examines dry deposition of pollutants into the spruce stand canopy and the second one investigates the impact of air conditioning on indoor air quality. The objective of a study involving external beam [5] PIXE analysis of ancient samples of lichens was to estimate the development of air pollution over an interval of several tens of years. The PIXE system was calibrated with MicroMatter reference targets and a set of thick metal targets and the calibration then validated with the Reference Material IAEA336 “Trace Elements in Lichen” [lo]. The IAEA336 homogenized powder was pressed into Teflon cups and dried at about 70°C. The reference samples were repeatedly analyzed at different current intensities to investigate the reproducibility of measurement. Fig. 3 compares elemental concentrations obtained during the repeated analysis of three identical samples of IAEA-336 at different current densities. When the current densities were kept around 1 nAlmm2 and the proton dose was low (about 1 PC), the analysis yielded stable results within ?3%. But when the current was raised and the acquisition prolonged, the measured concentrations started to increase by approximately lo20% due to the radiation damage. This agrees with
65o 7-
I
I 4 ,
600 ?O
38 .
1
20 L9
%zll & 500
I
3
We would like to thank to the research staff of the Lund Institute of Technology, meteorological station CHMI Kosetice, and the GSPM laboratory of the University of Guelph for their help and support of this project. The financial support was provided by CTU grants No. 48209, 3096381, and 3097489; grant of the Ministry of Environment GA/l 824193; GACR 202-94-0868. and Swedish SNV grants No. 15210 and 15310. Program PIXIN is available at http://www-troja.fjfi.cvut.cz/-nejedlylpixe.
References
111Z.
450 400
Acknowledgements
-
I I .
I
The PIXE setup in Prague has been successfully used for various types of environmental samples ~ both thin and thick including air particulate and lichen samples. The intercomparison projects confirmed the reliability of the PIXE analysis of air particulate samples in the Laboratory of Ion Beams at the CTU in Prague. The Nuclepore membranes proved as a very suitable medium for the intercomparison. Although some aerosol filters were subjected to repeated PIXE analysis in three laboratories and to the mechanical stress during shipping and mounting, concentrations of the majority of elements of interest were reproducible.
26
*
E 2 550
5. Conclusions
I+--
, .I
-
IO
123456184
II
Repeated measurement
Fig. 3. Concentrations of Fe measured in three samples of Reference Material IAEA-336 at different current densities and proton charges. The labels signify the proton currents for each measurement in nA/mm?. Proton charge was about 1 nC for the first sample and within 3-7 uC for the other two samples. The dotted line shows the 95% confidence interval of the mean reference value [lo].
Hulek, 2. Cespiro, R. Salomonovic. M. Set&k. J. Voltr. Vacuum 41 (1990) 1853. 121S.A. Johansson. J.L. Campbell, K.G. Malmquist, Particle induced X-Ray Emission Spectrometry (PIXE). Wiley. New York. 1995. (CTU Prague) 35 [31 J. Kriil. J. Voltr. Acta Polytechnica (1995) 4. [41 J. Kral. J. Voltr, Nucl. Instr. and Meth. B 85 (1994) 760. PI J. Kril, J. Voltr. 2. Nejedly. Nucl. Instr. and Meth. B 1091 110 (1996) 167. X-ray Fluorescence Calibration Standards, Fl MicroMatter MicroMatter Corp.. Deer Harbor, WA.
Z. NejedJij et ul. I Nucl. Instr. and Merh. in Phys. Rex B 136-138 (1998) 983-985 (71 J.A. Maxwell, W.J. Teesdale. J.L. Campbell. Nucl Instr.
and Meth. B 95 (1995) 407. [8] E. Swietlicki. H. Hansson, N.E.G. Lovestam. Nucl. Instr. and Meth. B 49 (1990) 3951101. [9] Z. Nejedly, J.L. Campbell, W.J. Teesdale, C. Gielen, Nucl. Instr. and Meth. B 132 (1997) 489.
985
[IO] Reference Material IAEA-336. Trace Elements in Lichens, International Atomic Energy Agency, Vienna. Austria. G4.12/1994-10-19. [ll] M.A. Reis. L.C. Alves. A.P. Jesus. Nucl. Instr. and Meth. B 109/110 (1996) 134.
