Marksaneringscentrum Norr

1
Improvement of XRF analysis precision by sample
preparation
B. Gustavsson, C. Maurice Div. of Waste Science
and Technology, Luleå University of Technology,
971 87 Luleå bjorn.gustavsson_at_sb.luth.se
46-0920-49 3004
Introduction To measure metal content in
contaminated soil, field X-Ray Fluorescence
Spectroscopy (XRF) is useful, for field
measurements as well as measurements in
laboratory. Several steps of sample preparation
is recommended drying, sieving and eventually
grinding. Further are special sample cup
recommended. The purpose of this study is to
investigate how different sample preparation
steps affect on analysis precision.
Method and material Each experimental setup was
measured 12 times with a XRF, NITON 700. Each
measurement was done during 120 nominally
seconds. Drying of samples was done at 105 C
during 2 hours. Each sample was measured through
sampling bags, freezing bags for food or Mylar 6
µm X-ray film. For statistic calculation
Statgrapchics plus 5.0 was used.
A Heap on ground Not sieved Undried sample
B Sampling bags Not sieved Undried sample
C Sampling bags Not sieved Dried sample
D Sampling bags lt2,00mm Dried sample
E Sampling bags lt0,25 mm Dried sample
F Sampling bags 0,25-2,00 mm Dried sample
G plastic bag for food freezing lt0,25 mm Dried sample
H Sampling bags Measured through Mylar X-ray film lt0,25 mm Dried sample
I Sample cup with Mylar X-ray window sample mixed between measurements lt0,25 mm Dried sample
J Sample cup with Mylar X-ray window cup rotated 30? between measurements lt0,25 mm Dried sample
K Sample cup with Mylar X-ray window sample cup fixed between measurements lt0,25 mm Dried sample
L ICP analysis of soil from the sample cup lt0,25 mm Dried sample
M Standard reference sample in sample cup NIST 2710 cup rotated 30 between measurements
N Standard reference sample in sample cup NIST 2710 Cup fixed between measurements

Figure a. Measured levels.
Figure b. Assuming all contaminant is in the
accepted fraction, the original sample would have
this contaminant levels.
Figure c. Assuming the contaminants in the
rejected fraction has half the levels as in the
measured fraction, the original sample would have
this contaminant levels.
Further work Copper and zinc will be evaluated
according to the same procedure, preliminary they
show similar tendencies. Samples from three more
sites are under measurement while writing. It
would be interesting to investigate whether the
increased measurement precision is due to sample
preparation itself, or due to homogenization
occurring during sample handling or is due to
both. The results of this investigation will be
used for evaluation of the NORSIC model.
Conclusions Increased sample handling tends to
increase precision of measurements. Also the
measured contaminant increases with increased
sample handling.
Acknowledgments This work was
financially supported by European Union
Structural Funds and New Objective 1, North
Sweden Soil Remediation Center, contract no
113-12534-00