Microsoft Word - 292-2047-3-LE-rev ACTA IMEKO  ISSN: 2221‐870X  April 2016, Volume 5, Number 1, 15‐21    ACTA IMEKO | www.imeko.org  April 2016 | Volume 5 | Number 1 | 15  Multi‐elemental composition of Slovenian milk: analytical  approach and geographical origin determination   Doris Potočnik 1,3 , Marijan Nečemer 2 , Darja Mazej 1 , Radojko Jaćimović 1  and Nives Ogrinc 1,3  1  Department of Environmental Science, Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia  2  Department of Low and Medium Energy Physics, Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia  3  Jožef Stefan International Postgraduate School, Jamova 39, 1000 Ljubljana, Slovenia      Section: RESEARCH PAPER   Keywords: milk; EDXRF; k0‐INAA; ICP‐MS; LDA; geographical origin  Citation: Doris Potočnik, Marijan Nečemer, Darja Mazej, Radojko Jaćimović and Nives Ogrinc, Multi‐elemental composition of Slovenian milk: analytical  approach and geographical origin determination, Acta IMEKO, vol. 5, no. 1, article 5, April 2016, identifier: IMEKO‐ACTA‐05 (2016)‐01‐05  Section Editor: Claudia Zoani, Italian National Agency for New Technologies, Energy and Sustainable Economic Development affiliation, Rome, Italy  Received September 6, 2015; In final form December 20, 2015, year; Published April 2016  Copyright: © 2016 IMEKO. This is an open‐access article distributed under the terms of the Creative Commons Attribution 3.0 License, which permits  unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited  Funding: Project funding by IAEA under Contract No. 17897  Corresponding author: Nives Ogrinc, e‐mail: nives.ogrinc@ijs.si    1. INTRODUCTION  Milk is an important everyday nutrition, since it is a source for many key nutrients including proteins, energy, many essential minerals and vitamins, factors that all contributed to an increased consumption in recent years. Milk contains more than 20 different micro and macro elements. Among them are elements such as Cu, Co, Se, Zn, Mn, Mo, I and Fe that are essential for human health in certain concentrations. These elements have a beneficial effect on animals and humans, since they are important for normal functioning of metabolism, growth and development and any concentration discrepancies of these elements may cause disturbance in organisms [1], [2]. Concentration of micro and macro elements in milk is not constant, but is influenced by genetics of individual animal species, stage of lactation, farm management (nutritional regime) and environment factors such as season, locality of the farm, nature of soil, fertilizer application and industrial activities [3]. Industrial and other anthropogenic activities could also influence the distribution and levels of elements, mainly toxic elements such as Cd, Ni, Co, Pb, Cr, Hg in milk, since they are ABSTRACT  The main objective  in multi‐elemental analysis  in food has traditionally been and still  is, to ensure food quality and safety. Three  different methods were investigated in the present study to obtain the elemental content in milk samples: energy dispersive X‐ray  fluorescence spectrometry (EDXRF), k0‐instrumental nuclear activation analysis (k0‐INAA) and the  inductively coupled plasma mass  spectrometry (ICP‐MS). Results indicate that the obtained data with different methods are in good agreement with certified reference  materials. Precision was found to be satisfactory with relative standard deviations always between 1 and 10 %, except for Se, As, Cd  and  Pb.  The  concentrations  of  these  elements  were  close  to  the  detection  limit  and  thus,  precision  higher  than  10  %.  An  intercomparison  exercise  between  EDXRF  and  k0‐INAA  showed  satisfactory  agreement  and  only  two  samples  exceeded  95  %  confidence interval for Rb and Zn with lower data obtained by k0‐INAA. It was found that EDXRF was the cheapest, simplest and most  environmental friendly method for analysis of multi‐elemental composition (P, S, Cl, K, Ca, Zn, Br, Rb, Sr) in milk samples, while for the  determination of Mn, Fe, Cu, Se content and possible identification of pollutants such as As, Cd and Pb ICP‐MS it was a method of  choice due to its excellent sensitivity and accuracy. These two methods were also used to determine the multi‐elemental composition  in Slovenian raw cow milk from different geographical regions: Alpine, Mediterranean, Dinaric and Pannonian  in December  2013.  Linear discriminant analysis (LDA) was used to explore multi‐elemental analysis of milk samples to obtain classification according to  geographical regions. Regional discrimination was most successful taking into account Ca, S, P, K, and Cl with prediction ability of 66.7  %.  ACTA IMEKO | www.imeko.org  April 2016 | Volume 5 | Number 1 | 16  accumulated through the food chain [1], [4], [5]. They can easily accumulate in plants grown on polluted soils and thus cause toxic effects in cattle and consequently also in humans who consume “contaminated” milk [4], [6]. For this reason, it is necessary to control the concentration of micro and macro elements/toxic elements in consumed food. Further, multi-element analysis has been successfully used for the determination of the geographical origin of different foodstuff such as wine [7], [8] onion [9]-[11], tea [12] and tomatoes [13]. In relation to milk and dairy products, most of the studies have been performed on cheese for descriptive purposes [14] and for origin authentication purposes [15]. Sacco et al. [16] used multi-elemental analysis in combination with stable isotope analysis for differentiation between Southern Italy and foreign milk. Trace elements in milk and dairy product have been used to investigate and to provide information about correlation between animals fodder, time of year, elemental conditions. It was found that season has more influence on mineral concentration than regions [15]. Several different techniques are available for simultaneous multi-element chemical analysis including: energy dispersion X- ray fluorescence spectrometry (EDXRF), neutron activation (k0-instrumental nuclear activation analysis; k0-INAA), inductively coupled plasma atomic emission spectroscopy (ICP- AES) and inductively coupled plasma mass spectrometry (ICP- MS) [1]. Each of these techniques has the capacity to provide a chemical profile or chemical fingerprinting, which can be used to characterize the material in question and to compare it with other samples or reference materials. Among inductively coupled plasma systems ICP-AES and ICP-MS have been widely used for the analysis of minor and trace elements in food, since they provide satisfactory result and allow performing simultaneous-sequential determination of multiple elements. The advantages of these techniques are short time of analysis and low limit of detection [4], [17]. On the other hand, EDXRF as a highly sensitive, fast and cheap technique offers the possibility of performing direct multi-element analysis of solid samples over a wide dynamic range. This technique is especially useful in food authentication and traceability studies where there is a requirement for a database of genuine samples to which the sample can be compared to establish its authenticity. For effective data processing, the large number of independently measured parameters is needed including multi- elemental composition [18]-[20]. Another nondestructive multi- element technique is k0-INAA. The main disadvantages of this method are long turn-around time, is labour intensive and a nuclear reactor or other source of activating particles is required. Nevertheless, providing a different approach, this method is often an important asset in studies regarding the analysis of standard reference materials [21]-[23] and as an additional technique in evaluation of the accuracy of analytical results [24]. Thus, the main objectives of this paper are: (1) to verify the methods for elemental composition in milk samples including EDXRF, k0-INAA and ICP-MS; and (2) to differentiate the milk according to geographical region in Slovenia based on elemental composition. Sample preparation and the analysis procedure for each of the above mentioned analytical techniques are described and a comparison of analytical and other parameters such as uncertainty, accuracy, limits of detection (LOD), cost of analysis per sample, instrumental cost etc. is critically evaluated. 2. MATERIALS AND METHODS  2.1. Sampling  Samples of the whole milk were provided by four Slovenian diary producers: Ljubljanske mlekarne d.d., Pomurske mlekarne d.d., Mlekarna Planika d.o.o. and Mlekarna Celeia d.o.o. in December 2013 covering different geographical regions (Mediterranean, Pannonia, Dinaric and Alpine) in Slovenia. The samples were stored at -20oC before analysis. All together 40 samples were obtained for multi-elemental analysis. 2.2. Energy Dispersive X‐ray Fluorescence Spectrometry (EDXRF)  Multielement determination of macro (P, S, Cl, K, Ca) and micro elemental (Zn, Br, Rb, Sr) content was nondestructively performed by Energy Dispersive X-ray Fluorescence Spectrometry (EDXRF). The analysis was performed on freeze- dried samples. The pellets were prepared from 0.5 to 1.0 g of powder sample material by a pellet die and hydraulic press. For excitation, the disc radioisotope excitation source of Fe-55 (25 mCi) and Cd-109 (20 mCi) from Eckert and Ziegler were used. The emitted fluorescence radiation was measured by en energy dispersive X-ray spectrometer constituted of a Si(Li) detector (Canaberra), a spectroscopy amplifier (Canberra M2024), ADC (Canberra M8075) and PC based MCA (S-100 Canberra). The spectrometer was equipped with a vacuum chamber (Fe-55) for the measurement of light elements P-Cl, while the energy resolution of the spectrometer was 175 eV at 5.9 keV. The analysis of complex X-ray spectra was performed by the AXIL spectral analysis program [24], [25]. For the evaluated uncertainty of this procedure, it is required to include the statistical uncertainty of measured intensities and the uncertainty of the mathematical fitting procedure. For this purpose, quantification was then performed utilizing QAES (Quantitative Analysis of Environmental Samples) software, developed in our laboratory [24], [25]. The estimated uncertainty of the analysis was around 5 % to 10 %. Rather high total estimated uncertainty is mainly due to contributions of matrix correction and geometry calibration procedures, which include errors of tabulated fundamental parameters, and contributions of spectrum acquisition and analysis. In the case of EDXRF, the limits of detection calculated from the signal to background ratio [26] on the realistic sample NIST 1549 for P, S, Cl, K, Ca, Zn, Br, Rb and Sr were 400, 200, 100, 20, 15, 4, 2, 1, 1 µg/g of dry milk sample, respectively. 2.3. k0‐INAA measurements  For k0-INAA measurements aliquot of the freeze-dry sample (0.10 to 0.20 g) was weighed into a polyethylene ampoule (SPRONK System, The Netherlands). For the determination of long-lived radionuclides the sample was irradiated together with the standard Al-Au (0.1 %) for 12 hours. Irradiation was carried out in a reactor TRIGA Mark II. After irradiation the sample was kept in a polyethylene ampoule and gamma activity of radionuclides induced in the sample after 4, 9-15, and 24-33 days of cooling was measured. All measurements were performed on HPGe detector (40 and 45 % yield). For the evaluation of gamma spectra we used the program HyperLab 2002, and element content was calculated with the program Kayzero for Windows. Limits of detection for Br, Ca, K, Rb, Zn calculated as three times the standard deviations of the blank sample, are 1, 500, 500, 5, 5 µg/g milk sample, respectively. ACTA IMEKO | www.imeko.org  April 2016 | Volume 5 | Number 1 | 17  The uncertainty level for elemental determination by k0- INAA was 610 %, in the case of lower elemental content such as Se and Mn the uncertainty is higher, around 20 %. 2.4. ICP‐MS measurements  For multi-elemental analysis of milk samples with ICP-MS two procedures of sample preparation were used: a) Microwave digestion: about 1 mL of milk sample (0.15 g of lyophilized sample) was weighed into quartz tubes. 1 mL 65 % HNO3, (Merck, Germany, suprapur) and 1 mL 30 % H2O2, (Merck, Germany, suprapur) were added and samples were subjected to closed vessel microwave digestion (Microwave system ETHOS 1, MILESTONE SN 130471) at max. power of 1500 W: ramp to 130°C 10 min, ramp to 200 °C 10 min, hold 20 min, cooling 20 min. Then the samples were equilibrated to room temperature. The solution was quantitatively transferred into 10 mL polyethylene graduated tubes and filled to the mark with Mili-Q water. Before determination by ICP-MS, samples were diluted five times. External calibration was made with ICP Multi Element Standard solution XVI CertiPUR (Merck). b) Dilution (adapted from Barany [27]): An aliquot of 1 mL of milk sample (0.15 g of lyophilized sample) was diluted five times with alkaline (Merck, suprapur) solution containing Triton X-100 (Sigma Aldrich, sigmaultra) and ethylenediaminetetraacetic acid disodium salt dehydrate (EDTA, Fisher Scientific, analytical reagent grade). For calibration the standard addition procedure was utilised. Measurements of prepared solutions were made by an Octapole Reaction System Inductively Coupled Plasma Mass Spectrometer (7500ce, Agilent) equipped with an ASX-510 Autosampler (Cetac). Instrumental conditions: Babington nebulizer, Scott-type spray chamber, spray chamber temperature 5 °C, plasma gas flow rate 15 L/min, carrier gas flow rate 0.8 L/min, make-up gas flow rate 0.1 L/min, sample solution uptake flow rate 1 mL/min, RF power 1500 W, reaction cell gas helium 4 mL/min, isotopes monitored 55Mn, 56Fe, 63Cu, 66Zn, 75As, 78Se, 111Cd, 114Cd, 206Pb, 207Pb, 208Pb, isotopes monitored as internal standard added to all solutions: 45Sc, 69Ga, 89Y, 157Gd. Tuning of the instrument was made daily using a solution containing Li, Mg, Y, Ce, Tl and Co. Concentrations of Cd, Pb and As in milk samples are usually very low <0.5 ng/g and thus, insufficient detection limits were obtained with microwave digestion. Therefore preparation with simple dilution of milk samples was introduced. Limits of detection for Cd, Pb, As, Se, Mn, Cu, Zn and Fe, calculated as three times the standard deviations of the blank sample, were 0.1, 0.2, 0.05, 2, 1, 6, 35 and 30 ng/g milk sample, respectively. Estimated combined uncertainty of results was 5 % for Cu, Zn, Fe, 10 % for Se, Mn and 15 % for As, Cd, Pb. 2.5. Quality control  The accuracy of results was checked as follows: a) analysis of the certified reference materials: Whole milk powder NIST 8435, Non-fat Milk Powder NIST 1549 (both National Institute of Standard and Technology) and Skim Milk Powder BCR 150 (EC-JRC-IRMM), ERM- BD150 and ERM-BD151 (EC-JRC-IRMM); b) comparison with independent methods: EDXRF and k0- INAA; c) for As there are no reference materials, therefore, comparison was made by an independent method radiochemical neutron activation analysis [28], [29]: result for internal quality control milk sample by k0-RNAA 0.28 ± 0.02 ng/g and by ICP-MS 0.36 ± 0.05 ng/g; d) participation in interlaboratory comparison scheme FAPAS (Food and Environmental Research Agency, Sand Hutton,York, VB). 2.6. Statistical evaluation   Statistical calculations and multivariate analysis were carried out using XLSTAT software package (Addinsoft, New York, USA) Basic statistics included mean values (median and arithmetic mean), standard deviation (S.D.), minimum and maximum. Multivariate analysis involved linear discriminant analysis (LDA). 3. RESULTS AND DISCUSSION  3.1. Results on certified reference materials   Results of the validation of the accuracy of multi-elemental analysis performed by ICP-MS with certified reference materials (NIST 1549, NIST 843) are collected in Table 1, while results of the validation of the accuracy with certified reference material ERM-BD 150 and ERM-BD 151 with all three methods are collected in Table 2. Results indicate that the obtained data with different methods are in good agreement with certified reference materials. Precision was also found to be satisfactory with relative standard deviations (RSDs) always between 1 and 10 %. Only in four cases (of Se, As, Cd and Pb), for which the Table 1. Validation of the accuracy of ICP‐MS multi‐elemental procedures proposed (NIST 1549, NIST 8435 and BCR 150).      NIST 1549   NIST 8435   BCR 150  Element  Certified   value  Found   Value  Certified   value  Found   value  Certified  value  Found  value    dry weight mg/kg  Mn  0.26 ± 0.06  0.25 ± 0.03  0.17 ± 0.05  0.19 ± 0.02  ‐  ‐  Fe  1.78 ± 0.10  1.71 ± 0.09  1.80 ± 1.10  2.0 ± 0.1  11.8 ± 0.6  9.79 ± 0.48  Cu  0.70 ± 0.10  0.67 ± 0.03  0.46 ± 0.10  0.40 ± 0.02  2.23 ± 0.08  1.95 ± 0.1  Zn  46.1 ± 2.2  39.3 ± 2.0  28.0 ± 3.1  23.8 ± 1.2  ‐  ‐  Se  0.11 ± 0.1  0.11 ± 0.1  0.131 ± 0.14  0.128 ± 0.01  ‐  ‐  Cd  0.0005 ± 0.0002  0.0005 ± 00001  ‐  ‐  0.0218 ± 0.0014  0.0203 ± 0.0031  Pb  0.019 ± 0.003  0.023 ± 0.004  ‐  ‐  1.00 ± 0.04  0.84 ± 0.13  ACTA IMEKO | www.imeko.org  April 2016 | Volume 5 | Number 1 | 18  measured concentrations were very close to the detection limit, precision was worse and higher than 10 %. 3.2. Comparison of elemental composition between EDXRF and  k0‐INAA methods  Comparison of results of milk samples obtained by EDXRF and k0-INAA is presented in Table 3. It was found that only two samples exceeded 95 % confidence interval for Rb and Zn with lower data obtained by k0-INAA (bold values in Table 3). The possible reason for this discrepancy could be in the inhomogeneity of distribution of these two elements in the two aliquotes of the same sample analysed by EDXRF and k0- INAA. 3.3. Results of the interlaboratory comparison scheme FAPAS  We participated in the interlaboratory comparison scheme FAPAS in 2012, 2013 and 2014 for milk samples in powdered form with all analytes present at low natural levels. Results are collected in Table 4. The results indicated good agreement between measured and assigned values. With these independent assessment competence of our laboratory to analyse low levels of As, Cd and Pb was demonstrated. Common characteristic of all three analytical techniques applied EDXRF, k0-INAA and ICP-MS is their multi-element capability. Preparation of samples was simple and nondestructive in the case of EDXRF and k0-INAA, meanwhile ICP-MS required decomposition of samples. It is obvious that the most sensitive method applied in this study was ICP-MS with LODs in the range of few tens of ng/g. The sensitivity of EDXRF and k0-INAA is comparable to each other, according to obtained estimated uncertainty (5-10 %) and LODs for the analysed elements in the range from hundred to few µg/g. This means that LODs of ICP-MS were approximately two orders of magnitude lower compared to EDXRF and k0-INAA. On the other hand, the determination of elements such as P, S, Cl by ICP-MS was impossible due to the fact that the ions formed by the ICP discharge are typically positive ions, M+ or M+², therefore, elements that prefer to form negative ions (such as Cl, I, F, etc.) are very difficult or impossible to determine by ICP-MS. On the other hand, EDXRF enables analysis of very important macro elements (P, S, and Cl) in milk samples, meanwhile k0-INAA as an alternative analytical tool allows the determination of only Cl. Determination of P was impossible due to the production of pure negatron emitter 32P after activation. In the case of S, poor activation of 37S (short half life t1/2 = 5.05 min) resulted in high LODs (3000-5000 µg/g). Since the S content found in milk was in the concentration range of 2000-3000 µg/g, it was evident that determination of S by k0- INAA was omitted due to its insufficient sensitivity. The advantage of k0-INAA in relation to EDXRF is the determination of other important macronutrients Mg and Na. The determination of Mg and Na by EDXRF was impossible due to insufficient instrumental sensitivity in concentration range from one to few thousand µg/g found in real milk samples. Another disadvantage of the k0-INAA and EDXRF techniques was their inability to perform the analysis of toxic elements such as As, Pb and Cd in the concentration range found in milk (Table 2). Further, considering the cost per sample, its multi-elemental capability, simple non-destructive sample preparation, the minimum number of steps in the measurement and quantification procedure, EDXRF was undoubtedly the cheapest, simplest and most environmental friendly method among the applied analytical techniques and the most suitable for analysis of multi-elemental composition (P, S, Cl, K, Ca, Zn, Br, Rb, Sr) in milk samples. However, for the analysis of elemental content of Mn, Fe, Cu, Se and possible identification of pollutants such as As, Cd and Pb the ICP-MS method was found as a method of choice due to its excellent sensitivity and accuracy. These two methods were also used for the determination of multi-elemental content in authentic Slovenian raw cow milk. 3.4. Multi‐elemental composition of Slovenian milk  The elemental content of Slovenian raw cow milk from different geographical regions collected in December 2013 is presented in Table 5. The elemental composition is as follows K > Ca > P, Cl > S > Zn > Rb > Br > Fe > Sr > Cu> Mn > Table 2. Determination of element concentrations in mg/kg in reference materials by ICP‐MS, EDXRF and k0‐INAA (ERM‐BD 150 and ERM‐BD 151).     ERM‐BD 150  ERM‐BD 151   Element  Certified value  Found value  by EDXRF  Found value  by k0‐INAA  Found value  by ICP‐MS  Certified  value  Found value  by EDXRF  Found value  by k0‐INAA  Found value  by ICP‐MS  Ca  13900 ± 800  13000 ± 1000  13487 ± 1013 ‐ 13900 ± 700  13200 ± 1100  13309 ± 1018  ‐ Mg  1260 ± 100  ‐  1266 ± 117 ‐ 1260 ± 70  ‐  1268 ± 112  ‐ Mn  0.289 ± 0.018  ‐  0.313 ± 0.101 ‐ 0.29 ± 0.03  ‐  0.306 ± 0.072  ‐ P  11000 ± 600  9100 ± 900  ‐  ‐ 11000 ± 600  9600 ± 900  ‐  ‐ K  17000 ± 700  16500 ± 1300  17665 ± 1263 ‐ 17000 ± 800  16200 ± 1300  17447 ± 1246  ‐ Se  0.188 ± 0.014  ‐  0.201 ± 0.053 0.227 ± 0.022 0.19 ± 0.04  ‐  0.188 ± 0.046  0.224± 0.022 Na  4180 ± 190  ‐  4397 ± 313 ‐ 4190 ± 230  ‐  4348 ± 308  ‐ Sr  ‐  3.71 ± 0.3  ‐  ‐ ‐  4.10 ± 0.3  ‐  ‐ Zn  44.8 ± 2.0  44.3 ± 3.5  46.7 ± 3.7  42.2 ± 2.1  44.9 ± 2.3  43.0 ± 3.4  46.1 ± 4.0  44.3 ± 3.2  Cu  1.08 ± 0.06  ‐  ‐  1.05 ± 0.05 5.00 ± 0.23  ‐  ‐  5.24 ±0.33 Cl  9700 ± 2000  9600 ± 900  10143 ± 743 ‐ 9800 ± 1200  9000 ± 900  10077 ± 737  ‐ Cd  0.0114 ± 0.0029  ‐  ‐  0.0122 ± 0.002 0.106 ± 0.013  ‐  ‐  0.106 ±0.008 Pb  0.019 ± 0.004  ‐  ‐  0.019 ± 0.003* 0.207 ± 0.014  ‐  ‐  0.206 ± 0.02 S  ‐  3300 ± 300  ‐  ‐ ‐  3300 ± 300  ‐  ‐ Br  ‐  13.2 ± 1.0  ‐  ‐ ‐  13.6 ± 1.0  ‐  ‐ Rb  ‐  17.7 ± 1.4  ‐  ‐ ‐  17.0 ± 1.4  ‐  ‐ ACTA IMEKO | www.imeko.org  April 2016 | Volume 5 | Number 1 | 19  Se > Pb > As > Cd and is consistent with literature data. The content of the elements, such as Cd, Cu, Fe, Mn, Ni, Pb, Sr and Zn is influenced by feed and environmental conditions [30]. The concentrations of toxic elements such as As, Cd and Pb, are low and do not represent a threat to human health. Multi-elemental composition of milk was used for possible discrimination among four Slovenian geographical regions. Results of statistical evaluation by LDA are presented in Figure 1. Partial separation between groups was obtained, where Alpine and Pannonian groups were well separated, while Dinaric and Mediterranean groups were overlapping. The later two groups are close to each other and thus insufficiently separation was obtained; however their discrimination tendency is promising. The most important factors affecting the geographical origin were Ca, S, P, K, and Cl. In a cross validation test 66.7 % of the samples are classified correctly. The highest rate of classification was for the Alpine and Dinaric samples (78.6 % and 71.4 %, respectively). It is expected that more precise and efficient separation between four geographical regions could be obtained by the involvement of more analysed parameters such as stable isotope values of milk samples. Table  3.  Comparison  between  k0‐INAA  and  EDXRF  methods  for  selected  macro‐ and  micro‐elements  in  milk  samples.  Samples  exceeding  95 %  Confidence interval of comparison are marked bold.     k0‐INAA  EDXRF  Sample  no.  Br  Ca  K  Rb  Zn  Br  Ca  K  Rb  Zn  dry weight mg/kg  1  14.0  7970  10969  12.1  28.8  13.9  8710  11200  13.2  30.4  2  11.4  8210  10595  22.1  18.9  11.7  9530  10600  23.3  21.5  3  9.04  7520  8706  6.85  23.3  8.87  7430  8210  7.98  25.2  4  7.71  8680  11800  12.6  31.7  7.85  9210  11600  14.0  33.3  5  7.77  8200  11110  10.3  29.2  8.38  8720  11600  8.85  30.9  6  8.32  8350  11300  12.9  28.5  6.97  8760  11200  14.2  29.9  7  8.10  7670  10390  11.9  27.3  8.38  8740  10600  12.5  28.7  8  7.59  8610  10880  11.5  26.7  8.27  10000  12100  12.9  29.8  9  8.34  8590  11140  16.6  26.6  7.74  9000  11300  19.3  29.6  10  11.4  8120  10980  9.33  29.2  11.4  8810  11300  8.3  30.6  11  8.22  8550  11332  14.0  29.0  9.74  8950  11900  15.0  31.8  12  14.6  8410  11170  12.5  28.6  13.2  9680  12600  11.9  29.6  13  9.12  8420  11230  20.3  27.7  7.93  9660  12400  20.3  29.8  14  16.0  6040  8595  11.1  23.0  15.9  7150  8610  11.4  23.0  15  10.9  7020  9898  22.3  23.6  14.3  6150  8070  28.1  30.9  16  14.1  9470  12251  14.1  29.8  14.0  9930  12600  13.3  30.1  17  16.6  9000  12635  22.8  31.2  17.4  9630  12900  23.6  29.6  18  9.44  6070  8113  9.72  20.7  12.5  6360  8320  15.4  26.0  19  13.7  7340  10140  15.7  25.6  13.8  8700  10700  16.5  26.1  20  27.5  7660  10332  14.8  23.2  28.2  8510  10900  14.9  26.1  Table 4. Results of the metallic contaminants in milk powder in interlaboratory schemes FAPAS in 2012, 2013 and 2014.      2012: FAPAS 07172  2013: FAPAS 07190  2014: FAPAS  Element  Assigned value  Found   value  Assigned  value  Found  value  Assigned  value  Found  value    µg/kg  As  56.4 ± 12.4  45.2 ± 6.1  49.1 ± 10.8 53.5 ± 6.0 58.1 ± 12.8  53.8 ± 1.6 Cd  18.59 ± 4.09  19.9 ± 5.8  16.2 ± 3.56 15.8 ± 1.0 12.2 ± 2.69  12.5 ± 0.8 Pb  66.17 ± 14.6  51.8 ± 3.3  50.8 ± 11.2 47.8 ± 3.4 47.5 ± 10.5  49.3 ± 3.0   Figure  1.  Discrimination  between  geographical  regions  using  significant  elemental parameters. Function 1 represents 72.43 % of variability, while  function 2 represents 19.91 % of variability.   Label: A‐Alpine, D‐Dinaric, M‐ Mediterranean, P – Pannonian.  ACTA IMEKO | www.imeko.org  April 2016 | Volume 5 | Number 1 | 20  4. CONCLUSIONS  There is a need for reliable milk element concentration data to provide information about nutritional uptake and at the same time to provide information about the excess amounts of trace and toxic elements. In addition, multi-elemental composition provides important information on the authenticity and geographical origin of food including milk and dairy products. This paper provides some interesting comparisons between three different techniques (EDXRF, k0-INAA and ICP-MS) in determination of multi-elemental composition in milk samples. Quality assurance proved entirely satisfactory for all involved measurements and an intercomparison exercise between EDXRF and k0-INAA showed satisfactory agreement. The simple, fast, and inexpensive EDXRF method in combination with ICP-MS, which was found the most appropriate technique for the analysis of elemental content of Mn, Fe, Cu, Se and toxic elements such as As, Cd and Pb, were further used for multi-elemental analysis of Slovenian raw cow milk samples. Multi-elemental content in milk samples was combined further in a linear discriminant model to discriminate milk according to geographical origin. Only partial separation was possible using elemental content with overall predicting ability of 66.7 %. It is expected that if elemental data are used in conjunction with other characteristic chemical indices such as isotope analysis, a more holistic and accurate picture in relation to geographical region separation could be created. These data represent a basis for a database of authentic samples of milk in Slovenia, which could be incorporated into a traceability system. ACKNOWLEDGEMENT  The research represents part of the ERA Chair ISO-FOOD project and was partially supported by the IAEA Project under Contract No. 17897 entitled “The use of stable isotopes and elemental composition for determination of authenticity and geographical origin of milk and dairy products” as part of CRP D5.20.38 “Accessible technologies for the verification of origin of dairy products as an example control system to enhance global trade and food safety”. REFERENCES  [1] N. Khan, I. S. Jeong, I. M. Hwang, J. S. Kim, S. H. Choi, E. Y. Nho, J. Y. Choi, K. S. Park, K. S. Kim, Analysis of minor and trace elements in milk and yogurts by inductively coupled plasma-mass spectrometry (ICP-MS), Food Chem. 147 (2014) pp. 220 – 224. [2] B. H. Schwendel, T. J. Wester, P. C. H. Morel, M. H. Tavendale, C. Deadman, N. M. Shadbolt, D. E. Otter, Invited review: Organic and conventionally produced milk – An evaluation of factors influencing, milk composition, J. Dairy Sci. 98 (2015) pp. 721 – 746. [3] C. Sola-Larrañaga, I. Navarro-Blasco, Chemometric analysis of minerals and trace elements in raw cow milk from the community of Navarra, Spain, Food Chem. 112 (2009) pp. 189 – 196. [4] S. Birghila, S. Dobrinas, G. Stanciu, A. Soceanu, Determination of major and minor elements in milk through ICP-AES, Environ. Eng. Manag. J. 7 (2008) pp. 805 – 808. [5] N. Bilandžić, M. Đokić, M. Sedak, B. Solomun, I. Varenina, Z. Knežević, M. Benić, Trace element levels in raw milk from Table 5. The summary of elemental data in Slovenian raw cow milk collected in December 2013. Elements  Mediterranean  Pannonia  Alpine  Dinaric  Ca (mg/100 g)  110  ±  14  113  ±  11  113  ±  15  111  ±  14  K (mg/100 g)  128  ±  20  145  ±  14  146  ±  20  144  ±  18  Cl (mg/100 g)  74  ±  17  84  ±  9  89  ±  10  91  ±  13  S (mg/100 g)  24  ±  3  27  ±  4  26  ±  5  28  ±  5  P (mg/100 g)  62  ±  12  77  ±  10  78  ±  13  76  ±  12  Na (mg/100 g)  33  ±  4  35  ±  2  35  ±  3  33  ±  7  Zn (mg/100 g)  329  ±  57  392  ±  28  374  ±  34  368  ±  34  Br (mg/100 g)  147  ±  32  106  ±  14  139  ±  40  197  ±  71  Rb (mg/100 g)  190  ±  100  196  ±  67  210  ±  63  215  ±  40  Sr (mg/100 g)  27  ±  6  29  ±  5  30  ±  15  20  ±  7  Fe (mg/100 g)  33  ±  3  25  ±  6  28  ±  5  29  ±  5  Ni (mg/100 g)  4.5  ±  0.7  4.6  ±  1.2  4.5  ±  1.0  5.6  ±  1.1  Mo (mg/100 g)  7.4  ±  1.2  7.5  ±  2.5  8.7  ±  1.8  8.8  ±  2.2  Mn (mg/100 g)  3.9  ±  0.9  2.8  ±  1.1  2.4  ±  0.8  2.9  ±  0.8  Cu (mg/100 g)  7.0  ±  3.4  4.6  ±  1.2  4.4  ±  1.4  5.3  ±  1.5  Se (mg/100 g)  2.0  ±  0.2  1.6  ±  0.4  1.6  ±  0.5  2.0  ±  0.8  As (mg/100 g)  0.043  ±  0.004  0.042  ±  0.017  0.043  ±  0.015  0.057  ±  23  Cd (mg/100 g)  0.006  ±  0.004  0.005  ±  0.003  0.004  ±  0.002  0.007  ±  0.004  Pb mg/100 g)  0.076  ±  0.051 0.055 ± 0.035 0.076 ± 0.057  0.064  ± 0.027 ACTA IMEKO | www.imeko.org  April 2016 | Volume 5 | Number 1 | 21  northern and southern regions of Croatia, Food Chem. 127 (2011) pp. 63 – 66. [6] R. Pilarezyk, J. Wojick, P. Czerniak, P. Sablik, B. Pilarezyk, A. Tomza-Mirciniak, Concentrations of toxic heavy metals and trace elements in raw milk of Simmental and Holstein-Friesian cows from organic farm. Environ. Monit. Assess. 185 (2013) pp. 8383 – 8392. [7] V. F. Taylor, H. P. Longerich, J. D. Greenough, Multielement analysis of Canadian wines by inductively coupled plasma mass spectrometry (ICP-MS) and multivariate statistics. J. Agr. Food Chem. 51 (2003) pp. 856 – 860. [8] P. P. Coetzee, F. E. Steffens, R. J. Eiselen, O. P. Augustyn, L. Balcaen, F. Vanhaecke, Multi-element analysis of South African wines by ICP-MS and their classification according to geographical origin, J. Agr. Food Chem. 53 (2005) pp. 5060 – 5066. [9] K. Ariyama, Y. Aoyama, A. Mochizuki, Y. Homura, M. Kadokura, A. Yasui, Determination of the geographic origin of onions between three main production areas in Japan and other countries by mineral composition, J. Agr. Food Chem. 55 (2007) pp. 347 – 354. [10] G. A. Chope, L. A. Terry, Use of canonical variate analysis to differentiate onion cultivars by mineral content as measured by ICPAES, Food Chem. 115 (2009) pp. 1108 – 1113. [11] E. Furia, A. Naccarato, G. Sindona, G. Stabile, A. Tagarelli, Multielement fingerprinting as a tool in origin authentication of PGI food products: Tropea red onion. J. Agr. Food Chem 59 (2011) pp. 8450–8457. [12] A. Moreda-Piñeiro, A. Fisher, S. J. Hill, The classification of tea according to region of origin using pattern recognition techniques and trace metal data. J. Food Compos. Anal. 16 (2003) pp. 195 – 211. [13] G. Lo Feudo, A. Naccarato, G. Sindona, A. Tagarelli, Investigating the origin of tomatoes and triple concentrated tomato pastes through multielement determination by inductively coupled plasma mass spectrometry and statistical analysis. J. Agr. Food Chem. 58 (2010) pp. 3801 – 3807. [14] L. Pillonel, R. Badertscher, P. Froidevaux, G. Haberhauer, S. Hölzl, P. Horn, A. Jakob, E. Pfammatter, U. Piantini, A. Rossmann, R. Tabacchi, J. O. Bosset, Stable isotope ratios, trace and radioactive elements in emmental cheese of different origin. Food Sci. Technol. 26 (2003) pp. 615 – 623. [15] M. T. Osorio, A. Koidis, P. Papademas, Major and trace elements in milk and Halloumi cheese as markers for authentication of goat feeding regimes and geographical origin, Int. J. Dairy Technol. 67 (2015) pp. 1 – 10. [16] D. Sacco, M. A. Brescia, A. Sgaramella, G. Casiell, A. Buccolieri, N. Ogrinc, A. Sacco, Discrimination between Southern Italy and foreign milk samples using spectroscopic and analytical data, Food Chem. 114 (2009) pp. 1559 – 1563. [17] J. Nobrega, Y.Gelinas, A. Krushevska, R. M. Barnes, Direct Cetermination of Major and Trace Elements in Milk by Inductively Coupled Plasma Atomic Emission and Mass Spectrometry, J. Anal. Atom. Spectrom. 12 (1997) pp. 1243 – 1246. [18] S. Kelly, K. Heaton, J. Hoogewerff, Tracing the geographical origin of food: The application of multi-element and multi- isotope analysis, Trends Food Sci. Tech. 16 (2005) pp. 555 – 567. [19] U. Kropf, M. Korošec, J. Bertoncelj, N. Ogrinc, M. Nečemer, P. Kump, T. Golob, Determination of the geographical origin of Slovenian black locust, lime and chestnut honey. Food Chem. 121 (2010) pp 839 – 846. [20] S. A. Drivelos, C. A. Georgiou, Multi-element and multi-isotope- ratio analysi to determine the geographical origin of foods in the European Union. Trac-Trend. Anal. Chem 40 (21012) pp. 38 – 51. [21] R. Jaćimović, P. Makreski, V. Stibilj, T. Stafilov, Determination of major and trace elements in iron reference materials using k0- NAA, J. Radioanal Nucl. Ch 278 (2008) pp. 795 – 799. [22] M. A. Menezes, R. Jaćimović, k0-INAA quality assessment by analysis of soil reference material GBW07401 using the comparator and neutron flux monitor approaches, Appl. Radiat. Isotopes. 69 (2011) pp. 1057 – 1063. [23] R. Jaćimović, Comparison of relative INAA and k0-INAA using soil and sediment reference materials. J. Radiosnsl. Nucl. Ch. 300 (2014) pp. 663 – 672. [24] M. Nečemer, P. Kump, J. Ščančar, R. Jaćimović, J. Simčič, P. Pelicon, M. Budnar, Z. Jeran, P. Pongrac, M. Regvar, K. Vogel- Mikuš, Application of X-ray fluorescence analytical techniques in phytoremediation and plant biology studies, Spectrochim. Acta B, 63 (2008) pp 1240 – 1244. [25] M. Nečemer, P. Kump, K. Vogel-Mikuš, “Use of X-ray fluorescence-based analytical techniques in phytoremediation”, in: Handbook of phytoremediation, (Environmental science, engineering and technology), I.A. Golubev, New York, 2011, ISBN 978-1-61728-753-4, 1-61728-753-9, pp. 331 – 358. [26] P. Kump, Some considerations on the definition of the limit of detection in X-ray fluorescence spectrometry, Spectrochim. Acta Part B 52 (1997) 405 – 408. [27] E. Barany, I. A. Bergdal., A. Schutz, S. Skerfving, A. Oskarsson, Inductively Coupled Plasma Mass Spectrometry for Direct Multi- element Analysis of Diluted Human Blood and Serum, J. Anal. Atom. Spectrom. 12 (1997) pp. 1005 – 1009. [28] A. R. Byrne, Low level simultaneous determination of As and Sb in standard reference materials using radiochemical neutron activation analysis with isotopic 77As and 125Sb tracers. Fresenius Z. Anal. Chem. 326 (1987) pp. 733 – 735. [29] A.R. Byrne, A. Vakselj, Rapid neutron activation analysis of arsenic in a wide range of samples by solvent extraction of the iodide, Croat. Chem. Acta 46 (1974) pp. 225 – 235. [30] Y. J. Tu, X. Y. Han, Z. R. Xu, Y.Z. Wang, W. F. Li, Effect of cadmium in feed on organs and meat colour of growing pigs, Vet. Res. Commun. 31 (2007) pp. 621 – 630.