Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: A particular subject of X-ray fluorescence analysis is its application in studies of the multielemental sam-
Received 20 September 2014 ple of composition in a wide range of concentrations, samples with different matrices, also inhomoge-
Received in revised form 19 June 2015 neous ones and those characterized with different grain size. Typical examples of these kinds of
Accepted 2 July 2015
samples are soil or geological samples for which XRF elemental analysis may be difficult due to XRF dis-
Available online 13 August 2015
turbing effects. In this paper the WDXRF technique was applied in elemental analysis concerning differ-
ent soil and geological samples (therapeutic mud, floral soil, brown soil, sandy soil, calcium aluminum
Keywords:
cement). The sample morphology was analyzed using X-ray microtomography technique. The paper dis-
Soil and geological samples
Wavelength dispersive X-ray fluorescence
cusses the differences between the composition of samples, the influence of procedures with respect to
analysis the preparation of samples as regards their morphology and, finally, a quantitative analysis. The results of
Matrix effects the studies were statistically tested (one-way ANOVA and correlation coefficients). For lead concentration
X-ray microtomography determination in samples of sandy soil and cement-like matrix, the WDXRF spectrometer calibration was
X-ray powder diffraction performed. The elemental analysis of the samples was complemented with knowledge of chemical com-
position obtained by X-ray powder diffraction.
Ó 2015 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.nimb.2015.07.136
0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
86 A. Kubala-Kukuś et al. / Nuclear Instruments and Methods in Physics Research B 364 (2015) 85–92
ray fluorescence determination of a particular element in a given making up the sample according to their routine analytical
material requires information concerning the chemical composi- procedures.
tion and the expected concentration ranges of all sample con- The last sample was calcium aluminium cement which is a
stituents in all analyzed samples. hydraulic binder used for the concretes and prefabricated ele-
In a typical XRF sample preparation procedure, the analyzed ments, in heating devices operating up to a temperature of
material is crushed, ground (with different milling time) and 1600 °C, in a special industrial building, in building chemistry as
formed as pellet (using different pressure). Sometimes different a component of mortars, in mining and in individual reparation
wax binders are used in the pellet formation process. The appropri- and installation works (stacks, ventilation, fireplaces). Such exten-
ate sample preparation procedure can reduce the disturbing sive use of calcium aluminium cement is possible due to its special
effects. properties, such as: a short bonding time, rapid increase in
The aim of the presented study was the analysis of the soil and mechanical strength and resistance to aggressive sulfate, sea
geological samples with different matrices using the wavelength water, carbon oxide, methane, the ability to use under subzero
dispersive X-ray fluorescence technique (WDXRF) [3]. The samples temperatures ( 10 °C). Typical chemical composition of this mate-
had different elemental composition, grain sizes and different rial is as follows: Al2O3 (minimum 40%), CaO (minimum 36%), SiO2
material packing. The main motivation of the undertaken research (3–6%), Fe2O3 (8–12%) and mineralogical basic phase is calcium
topic was the improvement of the qualitative and quantitative monoaluminate. Sandy soil and calcium aluminum cement sam-
analysis of the materials with a particular kind of matter, namely, ples, for which elemental composition was known, can be used
the soil and geological materials, and study of the influence of the in presented studies as a reference materials.
preparation procedure on sample morphology and next on the
reduction of XRF disturbing effects. The WDXRF elemental analysis 2.2. Sample preparation procedure
was complemented with chemical composition analysis using the
XRPD technique [12]. The morphology of each sample, especially For the wavelength dispersive X-ray fluorescence analysis
sample grain size, was characterized in the micrometers scale (WDXRF) the samples had to be prepared as a pellet. Prior to pellet
using X-ray microtomography technique [13]. The obtained results formation, the samples analyzed in the presented studies (thera-
of the analysis of the soil and geological samples were interpreted peutic mud, floral soil and brown soil samples) were dried and
in the context of influence of the sample matrix on the elemental ground with the compact ball mill (MiniMill2) at the rotation
quantitative analysis. Additionally, some practical aspects of this speed 300 rpm and different grinding time t = 1, 2, 3 and 6 min.
analysis were discussed. The measurements were done with the The sandy soil and calcium aluminium cement samples did not
WDXRF spectrometer – AXIOS (Panalytical) [14], X-ray diffrac- require grinding. Next, the ground samples of 5–10 g were formed
tometer X’Pert (Panalytical) [15] and X-ray microtomograph to the pellet. Usually in pellet formation the pelletizing pressure
SkyScan 1172 [16]. The first part of the paper describes the sam- was 100 kN but for brown soil sample milled for 6 min additional
ples and the preparation procedure. Next, the experimental devices pressure of 150 kN and 200 kN was applied. In general, it was
and measurement conditions are discussed. A further part of the expected that proposed sample preparation will result with the
work concentrates on the obtained results and discussion. The sec- reduction of grain-size effect appearing for soil and geological sam-
tion conclusion summarizes WDXRF, XRPD and X-ray microtomog- ples. Calcium aluminium cement samples were also prepared by
raphy studies of soil and geological samples. The results obtained adding to the sample a wax binder (C18H36O2N2) whose mass
can be used to improve the accuracy and reliability of the XRF was from 1% to 7% of the sample mass. The aim of the above test
measurements. was to check influence of the amount od added binder on the on
the quantitative analysis of the studied geological samples. Next,
the elemental composition of prepared pellets was analyzed using
2. Sample description and preparation procedure the WDXRF method.
Soil and geological samples which were analyzed in presented The WDXRF analysis was performed using the AXIOS spectrom-
studies have been chosen taking into account their different matri- eter (Panalytical) equipped with an Rh-anode X-ray tube with
ces, it was namely: peloid (therapeutic mud), floral soil, brown soil, maximum power of 2.4 kW. The wavelength dispersive system of
sandy soil and calcium aluminium cement samples. the spectrometer used five crystals (LiF (2 0 0), Ge (1 1 1), PE
Peloid in form of mud or clay is used in balneotherapy and ther- (0 0 2), PX1 and LiF (2 2 0)), which were automatically selected dur-
apeutic bathing. Is used in the treatment of rheumatic diseases, ing the measurements. The characteristic X-rays induced in the
gynecological diseases and certain diseases of the internal organs. sample were diffracted on one of the crystals and measured by
Laboratory processed mud is a raw material for the production of flow proportional counter for optimal detection of elements up
various kinds of medicines and cosmetics. The samples of peloid to Fe or a scintillation detector for heavier elements. In order to
analyzed under the presented studies came from a Polish health cover the X-ray energy (wavelength) range of the interest it was
resort Solec Zdrój located in the south-central part of Poland. necessary to perform 12 scans with different diffraction crystal-
Floral soil samples were commercially available different soil detector configurations. Energy resolution of the setup (10–
samples standard used in the cultivation of potted plants, while 50 eV) allows for unambiguous identification of element intensity
samples of brown soil (main type of soil in Poland) came from a even for very rich in elemental composition samples. The measure-
leafy forest from southern Poland. ments were performed in vacuum. The quantitative analysis of the
Sandy soil was a test sample of Worldwide Open Proficiency spectra was performed with the PANalytical analytical program
Test for X-ray Fluorescence Laboratories PTXRFIAEA10 [17] orga- Omnian [14]. The Omnian package is available for the standardless
nized by the IAEA Nuclear Science and Instrumentation analysis of all types of samples. Omnian includes advanced algo-
Laboratory in Seibersdorf, Austria. Sandy soil material, prepared rithms designed to profile known limitations inherent to XRF and
and tested by an external independent laboratory, was as a pow- includes spectral interference. The darkmatrix correction provides
dered, homogenized, and dried material distributed to laboratories better accuracy in cases where light elements such as C, H and O
in order to determine the mass fractions of chemical elements contribute to significant absorbance. In generally, corrections
A. Kubala-Kukuś et al. / Nuclear Instruments and Methods in Physics Research B 364 (2015) 85–92 87
Table 1
Concentration (%) of Na, Mg, Al, Si, P, S, K, Ca and Fe elements in soil and geological samples. Mud, floral soil and brown soil samples were grinding 3 min before pellets formation
in which 100 kN pressure was applied. In the table also experimental uncertainties and sum of the elements concentration are given.
mainly by counting statistics, and for major elements (>1% of con- not presented for all samples but for floral soil (three kinds), brown
centration) is 2–5% and for minor elements (about 0.1% of concen- soil (two kinds) and calcium aluminate cement (two samples). In
tration) it is 10–15%. For elements with concentration at about the figure Rh–Ka and Rh–Kb lines are marked together with the
0.01% the uncertainty is 10–20%, while for the lowest measured corresponding Compton scattering peaks. Having considered each
concentrations (about 0.001%) it is about 50%. Taking the latest spectrum ratio of the fluorescence line to its Compton scattering
observation into account, element concentration equal to 0.001% peak it can be observed that spectra are of three types (correspond-
(10 lg/g) is a detection limit of the applied WDXRF technique. ing with three groups of samples): (1) the high intensity of the
On the other hand for the concentration on the level of tens of per- Compton scattering peak (floral soils) – demonstrating strong scat-
cent the experimental uncertainty is below 1%. tering of Rh K-lines in a light matrix, (2) the low intensity of the
As it can be interpreted from the table therapeutic mud and flo- Compton scattering peak (cement material) – demonstrating weak
ral soil are materials with low Z (atomic number) matrix. The total Compton scattering and strong attenuation of Rh K-lines in a heavy
concentrations of the determined elements are relatively low, i.e. matrix, (3) an intermediate type (brown soils). In the quantitative
only 1.64% and 4.19%, respectively. The matrix of these samples analysis performed with Omnian standardless calibration the peak
are build mainly from light organic elements (H, C, N and O) most intensity of Compton scattering of primary X-ray can be taken into
of which are not detected in spectrometer. It is worth noting that account in the analysis of unmeasured matrix compounds. For this
relatively high concentration of sulfur in these samples is observed purpose so-called Compton validation factor (CVF) was determined
comparing with the rest of samples, namely brown soil, sandy soil by comparing the measured Compton scattering peak with this one
and cement. The matrices of the latter materials are systematic calculated on the basis of the obtained concentration. The value of
‘‘heavier”, for brown soil specified by high concentration of Si, Al, the CVF being about one, which was the case for the of analyzed
Fe, Ca and K, for sandy soil by Si and Al, and for cement by Ca, soil and geological samples, is an indication that, in quantitative
Al, Fe and Si. Also higher (from 44.5% to 58.5%), comparing to pre- analysis, the matrix of the analyzed sample was reconstructed
vious two materials, total sum of the discussed elements is properly. We have observed that for some special kind of samples,
observed. however, this standardless Omnian calibration introduce system-
Differences in sample matrix can be observed in the spectra of atic error that can be concluded on the basis of CVF significantly
the primary X-ray beam scattered in the sample. The results of different from one. This issues is addressed and discussed later in
such measurements are presented in the Fig. 3. The spectra were presented paper.
Assuming that the percentage elemental composition is equiv-
alent to percentage concentration of the oxides typical for soil
and geological samples, namely: Na2O, MgO, Al2O3, SiO2, P2O5,
SO3, K2O, CaO and Fe2O3, the oxides concentration were calculated.
Table 2 presents concentration values of these oxides in brown soil
ground 1, 2, 3 or 6 min under 100 kN of the pelletizing pressure.
Possible dependence of the oxide concentration on the time of
grinding was checked statistically. One-way ANOVA analysis [20]
Table 2
Concentration (%) of Na2O, MgO, Al2O3, SiO2, P2O5, SO3, K2O, CaO and Fe2O3 oxides in
brown soil ground 1, 2, 3 or 6 min (100 kN pelletizing pressure). In the table also
experimental uncertainties are given.
Fig. 4. Dependence of the SiO2 oxide concentration (%) on the time of material
grinding for brown soil sample. In the figure, the fitted line equation and correlation Fig. 6. The mean value of grain size estimated from X-ray microtomography images
factor R are presented. The confirmation of statistically significant correlation is for brown and floral soil samples versus grinding time. Marked uncertainties are
given by p value. standard deviations.
90 A. Kubala-Kukuś et al. / Nuclear Instruments and Methods in Physics Research B 364 (2015) 85–92
brown soil. Additionally, decreasing, as a function of grinding time, soil or calcium aluminum cement. Lead determination is one of
value of relative standard deviation results from better homogene- the main topics in elemental analysis of the soil, geological, or
ity of the sample. more general, environmental samples but applied procedure can
Possible dependence of oxides concentration in brown soil sam- be adopted to any element. Calibration samples were prepared
ples (see Table 2) on the pressure used in pellets formation by the addition to sandy soil and to cement material determined
(100 kN, 150 kN and 200 kN, t = 6 min) was checked statistically amount of the lead dioxide (PbO2) in order to get a few samples
for each oxides. In general, pressure used in pellets formation influ- with a different Pb concentration. For sandy soil material it was
ences on the characteristic X-ray intensity emitted from the sam- 5 samples with lead concentration of 0.5, 1, 3, 5 and 10 (%) and
ple, wherein the magnitude of this correlation depends of the for cement it was 0.06, 0.1, 0.2, 0.4 and 0.9 (%). The pellets were
grain size. In case of presented studies, however, only for K2O oxide formed under 100 kN pressure. In preparation of calibration sam-
concentration the correlation was statistically confirmed, with ples, the natural amount of lead in the samples was taken into
negative value of the correlation coefficient (R = 0.998). In our account (Pb = 26.3 lg/g in sandy soil, Pb = 77.8 lg/g in calcium alu-
opinion it is due to relatively small number of experimental points minium cement).
(only three values of pressure: 100 kN, 150 kN, 200 kN). For this Fig. 7 presents WDXRF spectra of sandy soil samples mixed with
reason also that topic is not discussed in detail in this paper. lead dioxide (PbO2) which amount was from 0 to 10% of sample
The results of the elemental analysis of calcium aluminium mass. The presented spectra (scans) were measured in an energy
cement samples prepared by mixing with a wax binder range from about 12.2 to 15.5 keV covering the energy of Pb–Lb,
(C18H36O2N2) whose mass was up to 7% of the sample mass are Pb–Lc, Rb–Ka and Sr–Ka fluorescence lines. Lead fluorescence lines
included in the Table 3. In quantitative analysis, performed using are very well separated up to natural lead level in the soil sandy
Omnian calibration, the concentration results were corrected for material which is closed to the detection level of the WDXRF tech-
the amount of binder. In the Table also the experimental uncertain- nique. It is also worth noting how increasing lead concentration in
ties are presented. As can be noticed, the sum of oxides concentra-
tion slowly becomes lower with an increase in the mass of the
binder. It is a result of decreasing the concentration of most of
the oxides (Na2O, MgO, Al2O3, SiO2, P2O5, SO3, CaO). Possible
dependence of the oxide concentration on the binder mass was
checked statistically. The correlation coefficients were calculated
for each oxide and negative correlation was statistically confirmed
for Al2O3 (R = 0.879), P2O5 (R = 0.886) and CaO (R = 0.981).
Main reason of the observed correlation is enhancement of scatter-
ing both of primary and secondary radiation because of binder
light element composition. Dilution of sample results in increased
levels of background and difficulty in detection of trace elements.
One-way ANOVA analysis [20] was applied to study the change
of total concentration of oxides under different binder mass. At
the significant level 0.05 total concentration of oxides is not statis-
tically different. Generally, although there are changes in the con-
centration, it can be concluded that the addition of wax binder did
not affect the quantitative analysis of the main components up to
7% of binder mass, the Omnian calibration can be applied without
necessity of preparation special quantitative calibration.
In quantitative analysis of soil and geological samples it was
observed that for matrix which component is element much heav-
ier than the other the standardless calibration introduces system-
atic error that can be concluded from observed values of
Compton validation factor being significantly different from one.
In these cases better is to performed classical calibration of spec- Fig. 7. WDXRF spectra of sandy soil samples mixed with a different amount of lead
oxide (PbO2) (from 0% to 10% of sample mass). The spectra were excited by the
trometer. In presented studies the classical calibrations for
primary X-rays generated in an Rh-anode X-ray tube operated with voltage
WDXRF spectrometer were performed for measurements of the U = 60 kV and current I = 40 mA. The presented spectra (scans) were measured in an
Pb concentration in samples whose matrices are of type of sandy energy range from about 12.2–15.5 keV and using LiF (2 2 0) crystal
(2d = 0.2848 nm).
Table 3
Concentration (%) of Na2O, MgO, Al2O3, SiO2, P2O5, SO3, K2O, CaO and Fe2O3 oxides in calcium aluminum cement samples mixed before the pellets formation with a different
amount of the binder. In Table also experimental uncertainties are included.
Calcium aluminum cement Measured compound concentration (%) for different amount of binder (%)
Oxide 3% 4% 5% 6% 7%
Na2O 0.048 ± 0.007 0.038 ± 0.006 0.035 ± 0.006 0.044 ± 0.006 0.036 ± 0.006
MgO 0.201 ± 0.013 0.200 ± 0.013 0.200 ± 0.013 0.147 ± 0.012 0.169 ± 0.012
Al2O3 42.6 ± 0.196 41.4 ± 0.193 41.9 ± 0.194 41.3 ± 0.193 40.6 ± 0.191
SiO2 2.60 ± 0.048 2.41 ± 0.047 2.47 ± 0.047 2.43 ± 0.047 2.35 ± 0.046
P2O5 0.049 ± 0.007 0.040 ± 0.006 0.039 ± 0.006 0.039 ± 0.006 0.035 ± 0.006
SO3 0.214 ± 0.014 0.189 ± 0.013 0.185 ± 0.013 0.177 ± 0.013 0.179 ± 0.013
K2O 0.044 ± 0.006 0.043 ± 0.006 0.043 ± 0.006 0.044 ± 0.006 0.044 ± 0.006
CaO 34.8 ± 0.177 34.3 ± 0.176 34.1 ± 0.175 33.9 ± 0.175 33.5 ± 0.174
Fe2O3 13.2 ± 0.109 13.0 ± 0.108 13.1 ± 0.109 13.0 ± 0.108 12.8 ± 0.108
Sum 93.8 91.6 92.1 91.1 89.7
A. Kubala-Kukuś et al. / Nuclear Instruments and Methods in Physics Research B 364 (2015) 85–92 91