Talanta 65 (2005) 9297 Talent 65 (2005) 92-97 Direct determination of toxic trace metals in honey and sugars using

Direct Determination of toxic trace metals in honey and sugars Using inductively coupled plasma atomic emission spectrometry inductively coupled plasma atomic emission spectrometry MD Ioannidou, GA Zachariadis Ioannidou MD, GA Zachariadis , AN Anthemidis, JA Stratis , AN Anthemidis, JA Stratis Laboratory of Analytical Chemistry, Department of Chemistry, Aristotle University of Thessaloniki, GR-541 24 Thessaloniki, Greece Laboratory of Analytical Chemistry, Department of Chemistry, Aristotle University of Thessaloniki, GR-541 24 Thessaloniki, Greece Received 19 January 2004; received in revised form 24 March 2004; accepted 12 May 2004 Received 19 January 2004, received in Revised form 24 March 2004 Accepted 12 May 2004 Available online 14 June 2004 Available online 14 June 2004 Abstract Abstract A rapid method for the determination of Pb, Cd, Cu, Cr, Co, Ni, Mn and Zn in honey and sugars without prior digestion or ashing of the A rapid method for the Determination of Pb, Cd, Cu, Cr, Co, Ni, Mn and Zn in honey and sugars without Prior Digestion or ashing of the sample was developed, using inductively coupled plasma atomic emission spectrometry (ICP-AES). WAS samples developed, using inductively coupled plasma atomic emission spectrometry (ICP-AES). The critical instrumental parameters such The critical instrumental parameters Suche as sample flow rate and radio frequency incident power were thoroughly optimized. I sample flow rate and incident radio frequency power Were Thoroughly optimized. The effect of matrix type and its concentration was also The effect of matrix type and WAS ITS Also Concentration examined for glucose/fructose, sucrose and honey matrices. Examined for glucose / fructo, sucrose and honey Matrices. The sensitivity was investigated using calibration curves obtained in presence of The sensitivity calibration curves Obtained Using WAS investigated in Presence of the above matrices. Above the Matrices. The obtained recoveries for Cd, Cu, Cr, Co, Ni and Mn at the gl The Obtained recoveries for Cd, Cu, Cr, Co, Ni and Mn at the gl 1 -1 level were satisfactory and practically independent of Were Practically satisfactory level and independent of the matrix used for the calibration standards. The matrix used for the calibration standards. The recoveries of Pb and Zn were less sufficient. The recoveries of Pb and Zn Were less sufficient. Various commercial samples of honey, sugar, Various commercial samples of honey, sugar, glucose and fructose were analyzed with respect to their toxic metal content. glucose and analyzed with respect to fructo Were Their toxic metal content. The method can be applied for routine analysis, quality and The method for routine analysis Can Be Applied, quality and environmental pollution control purposes at the gl Environmental Pollution Control Purposes at the gl 1 -1

level of concentration, after suitable dilution of the samples. level of Concentration, Suitable after Dilutional of the samples. 2004 Elsevier BV All rights reserved. 2004 Elsevier BV All rights reserved. Keywords: Toxic metals; Honey; Sugars; Glucose; Inductively coupled plasma; Atomic emission spectrometry Keywords: Toxic metals, Honey, Sugars, Glucose, Inductively Coupled Plasma, Atomic Emission Spectrometry

1. 1. Introduction Introduction Honey is high-viscosity liquid foodstuff containing Honey is high-viscosity liquid containing Foodstuff a range of nutritiously important complementary ele- the range of nutritiously important complementary elements. ments. It contains a mixture of carbohydrates, such as It contains a mixture of carbohydrates, Suche I fructose (2545% m/m), glucose (2537% m/m), maltose Fructo (25-45% m / m), glucose (25-37% m / m), maltose (212% m/m), sucrose (0.53% m/m) with traces of many (2-12% m / m), sucrose (0.5-3% m / m) with traces of Many other sugars depending on the floral source and water Other sugars depending on the floral source and water (1518% m/m) [1] . The mean content of mineral sub- (15-18% m / m) [1]. The mean content of mineral substances in honey has been calculated to be 0.17% m/m, stances in honey HAS Been Calculated To Be 0.17% m / m, although it varies within a wide range. although it varies Within the wide range. Honey is the result Honey is the result of a bio-accumulation process useful for the collection of of the bio-accumulation process for the collection of Useful information related to the environment where the bees live. information related to the environment WHERE live the Bees. Since the forage area of the hive is very large (more than Since the area of the hive Forage is very large (More Than 7km 7km 22 ) and the bees come in contact not only with air but ) And the Bees come in contact only with air But Not also with soil and water, the concentration of heavy metals Also with soil and water, heavy metals of the Concentration in honey reflects their amount in the whole region. Their honey reflects the amount in the Whole Region. There- Therefore, honey has been recognized as a biological indicator forces, honey HAS Been recognized as a Biological Indicator of environmental pollution [2] . But nevertheless, determi- of Environmental Pollution [2]. But nevertheless, deter* Corresponding author. Corresponding author. Fax: +30 2310997719. Fax: +30 2310997719. E-mail address: zacharia@chem.auth.gr (GA Zachariadis). E-mail address: zacharia@chem.auth.gr (GA Zachariadis).

nation of heavy metals in honey is of high interest mainly nation of heavy metals in honey is of high interest mainly for quality control and nutritional aspect. for quality control and nutritional aspect. High levels of High levels of metals are undesirable because of their known or supposed Because of metals is undesirable Their Known or supposed toxicity, so that, for instance a limit of 1 mg kg Toxicity, so That, for instance the limit of 1 mg kg 1 -1 for lead for lead is set in some countries [3] . Some countries is set in [3]. Furthermore, sugars (such as glucose, fructose or sugar) Furthermore, sugars (glucose as Suche, fructo or infant) are classified as generally recognized safe food ingredients. is classified as GeneRally recognized safe food Ingredients. Because of their extended usage, high concern exists about Because of Their extended usage, concern exists about high the contribution of these ingredients to the total dietary in- The Contribution of These Ingredients to the Total Dietary intake of trace metals that may be present as contaminants. take of trace metals present as contaminants That May Be. Specific attention has focused on lead because of increas- Attention Focused on specific HAS Because of increas-lead ing knowledge of adverse health effects from lead at vari- ing knowledge of adverse health effects from lead at various levels of exposure [4] . However, these ingredients also ous levels of exposure . [4] However, tissue Also Ingredients serve as potential sources of exposure to other metals, such serve as potential sources of exposure to Other metals, Suche as cadmium, copper and tin, which occur at various levels as cadmium, copper and tin, Which occur at various levels in the environment [5,6] . in the environment [5,6]. Metals determination in sugar-rich foodstuffs has been Metals Determination in sugar-rich Foodstuffs HAS Been a challenging analytical task due to the interference arising a challenging task due to the Analytical Interference arising from the matrix. from the matrix. Sample pre-treatment is usually required Sample pre-Treatment is usually required to destroy the organic matrix and to extract the metal ions to destroy the organic matrix and to extract the metal ions bound in organic complexes. bound in organic complexes. Obviously, the selection of the Obviously, the selection of the digestion procedure must take into account the analytes, the procedure must take into account Digestion the analytes, the 0039-9140/$ see front matter 2004 Elsevier BV All rights reserved. 0039-9140 / $ - see front matter 2004 Elsevier BV All rights reserved. doi:10.1016/j.talanta.2004.05.018 doi: 10.1016/j.talanta.2004.05.018

MD Ioannidou et al. Ioannidou MD et al. / Talanta 65 (2005) 9297 / Talant 65 (2005) 92-97 93 93 sample matrix and the time requirements of the analytical sample matrix and the time of the Analytical Requirements technique considered. Considered technique. Several techniques have been proSeveral Techniques have Been proposed for the determination of heavy metals in honey and possess for the Determination of heavy metals in honey and other sweeteners, but in most cases the matrix mineraliza- Other sweeteners, But in most Cases the mineralized matrixtion is required [3,4,7] . However, in common mineraliza- tion is required [3,4,7]. However, in common-mineralized tion procedures there are risks of contamination or analyte tion Procedures There are Risks of Contamination or analytic loss as a result of prolonged sample manipulation and loss as a result of prolonged sample manipulation and heating. heating. Flame atomic absorption spectrometry (FAAS), due to its Flame Atomic Absorption Spectrometry (FAAS), due to ITS relatively low cost and quite good analytical performance relatively low cost and quite good performance Analytical has been widely used for determination a various metals in Been widely used for HAS Determination of various metals in honey [811] . The conventional way to carry such deter- Honey [8-11]. The conventional way to carry fixed-Suche minations involves a mineralization stage to obtain a final minations involves the mineralization stage to Obtain the final solution suitable for introduction into the flame nebulizer Introduction into the Solution Suitable for flame nebulizer [9] . The destruction of the organic matter eliminates both [9]. The destruction of the organic matter eliminates Both spectral interferences and the accumulation of residues in spectral interferences and the accumulation of residues in the burner head and spray chamber. the spray chamber and burner head. However, a drawback is However, the drawback is posed since the linear response range of the AAS is narrow. since the linear response range possess of the ASA is narrow. Difficulties associated to the high organic matter content are Difficulties Associated to the high organic matter content is overcome by dissolving the samples in acidified water and overcome by dissolving the samples in water and acidified then directly introducing the resulting solution in the neb- THEN Directly introducing the resulting solution in the Nebulizer [10] . Electrothermal atomic absorption spectrometry ulizer [10]. Electrothermal Atomic Absorption Spectrometry (ETAAS) has been proposed as a more sensitive analytical (ETAAS) HAS Been proposed as a more sensitive Analytical technique compared to FAAS for the determination of toxic technique compared to FAAS for the Determination of Toxic

metals in honey and sugars [1,12] . metals in honey and sugars [1.12]. Inductively coupled plasma-based techniques (ICP-AES Inductively coupled plasmabased Techniques (ICP-AES and ICP-MS) have been applied as multi-elemental tech- and ICP-MS) have Been Applied as multi-element techniques for the determination of heavy metals in honey niques for the Determination of heavy metals in honey and other sweeteners [7,1316] . ICP-AES is attractive for and Other sweeteners [7.13 to 16]. ICP-AES is attractive for trace analysis, owing to the satisfactory sensitivity cou- trace analysis, owing to the satisfactory sensitivity coupled with the advantage of simultaneous determinations blanket with the Advantage of simultaneous determinations of several metals at several spectral lines. Several of Metals at Several spectral lines. However, the However, the great disadvantage of sample dry or wet digestion still re- great disadvantage of wet or dry sample Digestion still remains. Mains. Hence, the sample is usually digested by wet-acid Hence, the sample is usually digested by wet-acid or dry ashing by heating in a microwave oven or a fur- or dry ashing by heating in a microwave oven or a furnace. NACE. The aim of this study was to investigate the possibility The aim of this study to investigate the Possibility WAS of the simultaneous and direct measurement of heavy met- of the simultaneous and direct measurement of heavy metals in honey and sugars using ICP-AES without any dry or als in honey and sugars Using ICP-AES without Any dry or wet sample dissolution. Dissolution wet sample. The effect of the matrix type and The effect of the matrix type and its concentration was examined using aqueous (AQ) matrix ITS WAS Concentration Examined Using aqueous (AQ) matrix and three different carbohydrate matrices: a 1+1 mixture of Three different carbohydrate and Matrices: a 1 +1 mixture of glucose and fructose mixture (GF), sucrose (SU) and honey glucose and fructo mixture (GF), sucrose (SU) and honey (HO). (HO). The sensitivity of the method with respect to each The sensitivity of the method with respect to Each metal was evaluated using the resulted slope of the calibra- Using the metal resulted WAS evaluated slope of the calibratedtion curves. tion curves. As there are not readily available standard ref- Not readily available I There are standard reference materials [17,18] , the recoveries of the analytes were erence materials [17.18], the recoveries of the analytes Were measured at different concentrations using spiked solutions Measured at different concentrations Solutions Using Spiked and used as estimation of the accuracy of the method. and used as estimation of the Accuracy of the method. The The precision was evaluated by measuring the repeatability of evaluated by measuring the precision WAS Repeatability of

the method for all analytes and at different sugar matrixes. the method for all analytes and at different sugar matrixes. The capability of the method as a routine analysis method The capability of the method as a routine analysis method was estimated through the determination of the detection WAS estimated through the Determination of the detection limits of every heavy metal studied. Every heavy metal limits of studied. The proposed method The proposed method was applied in a number of commercial samples of honey, Applied in the number of WAS commercial samples of honey, sugar, glucose and fructose. sugar, glucose and fructo. Table 1 Table 1 Operating conditions and description of the ICP-AES instrument Operating Conditions and description of the ICP-AES instrument rf generator rf generator 40 MHz, free-running 40 MHz, free-running rf incident power incident rf power Optimized Optimized Torch, injector, id Torch, injector, id Fassel type, Alumina, 2.0 mm Fassel type, alumina, 2.0 mm Argon flow rates Argon flow rates Auxiliary 0.5 l min Auxiliary 0.5 l min 1 -1 ; nebulizer , Nebulizer 0.85 l min 0.85 L min 1 -1 ; plasma 15 l min ; Plasma 15 l min 1 -1 Air flow rate Air flow rate 18 l min 18 l min 1 -1 Spray chamber Spray chamber Scott double-pass Scott double-pass Nebulizer Nebulizer Gem tip cross flow Gem cross-flow type Sample propulsion Sample Propulsion Peristaltic pump, three channel Peristaltic pump, three channel Sample flow rate Sample flow rate Optimized Optimized Polychromator/resolution Polychromator / resolution Echelle/0.006 nm at 200 nm Echelle/0.006 nm at 200 nm Detector Detector Segmented-array charge-coupled (SCD) Segmented-array charge-coupled (SCD)

2. 2. Experimental Experimental 2.1. 2.1. Instrumentation Instrumentation A Perkin Elmer Optima 3100 XL axial viewing in- A Perkin Elmer Optima 3100 XL axial viewing inductively coupled plasma atomic emission spectrometer ductively coupled plasma atomic emission Spectrometer was used, according to the operating conditions given WAS used, According to the Operating Conditions given in Table 1 . The analytical wavelengths for each analyte in Table 1. The Analytical wavelengths for Each analytic was, Pb: 220.353, 217.000, 261.418 nm; Cd: 214.440, WAS, Pb: 220.353, 217.000, 261.418 nm, CD: 214 440, 226.502, 228.802 nm; Cu: 324.752, 224.700, 327.393 nm; 226.502, 228.802 nm; with: 324.752, 224.700, 327.393 nm; Cr: 283.563, 284.325, 267.716 nm; Co: 228.616, 238.892, Cr: 283.563, 284.325, 267.716 nm, Co 228.616, 238.892, 230.786 nm; Ni: 221.648, 232.003, 341.476 nm; Mn: 230.786 nm, Ni: 221.648, 232.003, 341.476 nm, Mn: 257.610, 259.372, 260.568 nm; Zn: 213.857, 202.548, 257.610, 259.372, 260.568 nm, Zn: 213.857, 202.548, 206.200 nm. 206.200 nm. For the optimization of the instrument's perfor- For the optimization of the instrument's performance different radio frequency (rf) incident power levels Mance different radio frequency (rf) incident power levels and sample flow rates were investigated. and sample flow rates Were investigated. 2.2. 2.2. Reagents and solutions Reagents and solutions All chemicals were of analytical reagent grade and were All reagent grade chemicals Were of Analytical and Were provided by Riedel de Haen. provided by Riedel of Haen. The chemical reagents used for The chemical reagents used for preparation of matrix matched standards (glucose, fructose preparation of matrix matched Standards (glucose, fructo and sucrose) were of analytical grade, provided by Riedel de and sucrose) Were of Analytical grade, provided by Riedel of Haen. Haen. De-ionized water of MilliQ quality was used through- De-Ionized water of MilliQ quality WAS used throughout. out. Mixed working standard solutions of the analytes (Pb, Mixed standard working solutions of the analytes (Pb, Cd, Cu, Cr, Co, Ni, Mg and Zn) were prepared by appropri- Cd, Cu, Cr, Co, Ni, Mg and Zn) Were Prepared by appropriate stepwise dilutions of stock standard solutions containing ETA stepwise dilutions of standard stock solutions containing 1000 mg l 1000 mg L 1 -1 of each analyte in 0.5 mol l Each analytic in 0.5 mol of L 1 -1 HNO HNO 33 (Riedel (Riedel

de Haen), according to the procedures described below. for Haen) According to the Procedures described Below. Eight-point calibration curves were obtained using ma- Eight-point calibration curves Were Obtained Using Itrix matched standards prepared in aqueous (AQ), glu- Prepared Standards matched Trix in aqueous (AQ), glucose/fructose in water (GF), sucrose in water (SU) and Baskets / fructo in water (GF), sucrose in water (SU) and honey in water (HO) matrix. honey in water (HO) matrix. The final mixed working stan- The final mixed working standard solutions had the following concentrations: 0.25, 0.50, Following the solutions Dard HAD concentrations: 0.25, 0.50, 1.00, 2.50, 5.00, 10.0, 25.0 and 50 gl 1.00, 2.50, 5.00, 10.0, 25.0 and 50 gl 1 -1 for each metal. Each metal for. In case of the carbohydrate matrices, a 2% m/v solution In case of the carbohydrate Matrices, 2% w / v solution of the corresponding sugar (glucose/fructose, sucrose, or of the corresponding sugar (glucose / fructo, sucrose, or honey) was used for dilution. honey) WAS used for Dilutional. In order to check the overall In order to check the Overall repeatability of the ICP-AES detector two different calibra- Repeatability of the ICPAES detector calibrate two differenttion procedures were made at two different time periods. tion made at two different Procedures Were Time Periods. The slope of the calibration curves was used to estimate the The slope of the calibration curves used to estimate the WAS sensitivity of the method. sensitivity of the method. 94 94 MD Ioannidou et al. Ioannidou MD et al. / Talanta 65 (2005) 9297 / Talant 65 (2005) 92-97 Fig. Fig. 1. 1. Effect of the sample flow rate on the intensity of 50 gl Effect of the sample flow rate on the intensity of 50 gl 1 -1 Mn at GF matrix, at the following spectral lines and carbohydrate concentrations: ( ) Mn at GF matrix, at the spectral lines and carbohydrate concentrations Following: () 257.610 nm1% m/v; ( ) 257.610 nm10% m/v; ( ) 259.372 nm1% m/v; ( ) 259.372 nm10% m/v; ( ) 260.568 nm1% m/v; () 260.568 nm10% m/v. 257.610 nm to 1% w / v, () 257 610 nm to 10% w / v, () 259 372 nm to 1% w / v, () 259 372 nm to 10% w / v, () 260 568 nm to 1% m / v, () 260 568 nm to 10% w / v. Samples of honey, glucose and fructose are considered as Samples of honey, glucose and fructo is Considered as not homogeneous matrices due to their very high viscosity, Not very homogeneous Their Matrices due to high viscosity, thus it is crucial to obtain representative sub-samples for thus it is crucial to Obtain representative samples for subfurther analysis [7,16] . In case of honey, glucose and fructose Frther analysis [7.16]. In the case of honey, glucose and fructo

samples, the sample container was warmed to ca. samples, the sample container WAS warmed to ca. 50 50 C in a C in the water-bath and sonicated for 5 min. water-bath and sonicated for 5 min. Then an amount of 20 g Then year amount of 20 g of the sample was accurately weighed in a 250-ml beaker accurately weighed sample of the WAS in a 250-ml Beaker and diluted with about 80 ml of doubly de-ionized water. and diluted with about 80 ml of doubly de-Ionized water. Finally, the mixture was transferred to a 1000 ml volumetric Finally, the mixture of 1000 ml WAS Transferred to volumetric flask, diluted to volume with de-ionized water. flask, diluted to volume with de-Ionized water. The obtained The Obtained solution was used as a 2% m/v honey solution. WAS solution used as a 2% m / v honey solution. In case of In case of sugar samples the above procedure was followed without a Above the infant samples WAS FOLLOWED procedure without the warming step. warming step. 3. 3. Results and discussion Results and discussion 3.1. 3.1. Effect of the incident rf power Effect of the incident rf power In order to examine the effect of rf power on the in- In order to examine the effect of rf power on the intensity, of 50 gl tensity, of 50 gl 1 -1 concentration level of each analyte at Concentration Level of Each analytic at two carbohydrate concentration levels (2 and 5% m/v) in Two carbohydrate Concentration levels (2 and 5% w / v) in GF matrix, at three different power levels (1300, 1400 and GF matrix, at different power levels Three (1300, 1400 and 1500 W) were tested. 1500 W) Were Tested. As it was found, by increasing the rf IT WAS I found, by Increasing the rf power the intensity is also increased, for all analytes. Also the power intensity is increased, for all analytes. How- However, the plasma became unstable at 1400 and 1500 W, thus ever, the plasma at 1400 and 1500 Became unstable W, thus the rf power was adjusted to 1300 W for the rest of the the rf power to 1300 W WAS Adjusted for the rest of the experiments. experiments. 3.2. 3.2. Effect of the sample flow rate Effect of the sample flow rate In ICP nebulizers, the sample flow rate defines the amount In ICP nebulizers, the sample flow rate defines the amount of analyte in the torch and affects the sensitivity of the of analytic in the torch and affects the sensitivity of the method. method. The effect of sample flow rate on the intensity was The effect of sample flow rate on the intensity WAS studied in the range 1.02.03.0 ml min studied in the range 1.0-2.0-3.0 ml min 1 -1 and the Ar flow and the flow would

rate in the nebulizer was adjusted to 0.800.850.90 l min Adjusted rate in the nebulizer WAS l min to 0.80-0.85-0.90 1 -1 ,, respectively. respectively. Two matrices (GF and SU), two carbohydrate Two Matrices (GF and SU), two carbohydrate concentrations 1.0 and 10% m/v) and 50 gl concentrations 1.0 and 10% w / v) and 50 gl 1 -1 concentra- focustion of each analyte were used through this study. tion of analytic Were Each study used through this. In Fig. In Fig. 1 , 1, the effect of the sample flow rate on the intensity of Mn the effect of the sample flow rate on the intensity of Mn is demonstrated. is Demonstrated. According to the obtained results the in- Obtained According to the results the intensity was increased by increasing the flow rate from 1 to WAS tensity increased by Increasing the flow rate from 1 to 2mlmin 2mlmin 1 -1 and levelled of over 2 ml min and levelled of over 2 ml min 1 -1 for all the exam- for all the examined analytes. INED analytes. Also, another trend was observed: significant Also, Another trend observed WAS: Significant depressing of the intensity, by increasing the carbohydrate depressing of the intensity, by Increasing the carbohydrate concentration in the matrix from 1 to 10% m/v. Concentration in the matrix from 1 to 10% w / v. Similar behaviour was found in presence of the other ex- Similar behavior of the WAS found in Presence Other examined matrix type (SU, HO). amined matrix type (SU, HO). Moreover, the obtained re- Moreover, the re-Obtained sults were statistically analyzed using the paired t -test at a sults analyzed statistically Were Using the paired t-test at a confidence level of 95%, in order to define the optimum confidence level of 95%, in order to define the optimum value of sample flow rate. value of sample flow rate. The results in flow rates of 1 and The results in flow rates of 1 and 2mlmin 2mlmin 1 -1 are statistically different and in all cases the in- is statistically different in all Cases and the intensity at 2 ml min tensity at 2 ml min 1 -1 was higher. Higher WAS. However, there is not a sig- However, There is not a significant statistical difference between the results obtained at Statistical difference between the planners Obtained results at flow rate of 2 and 3 ml min flow rates of 2 and 3 ml min 1 -1

. . Therefore, 2 ml min Therefore, 2 ml min 1 -1 sample sample flow rate was selected for the rest of the experiments. flow rate for the rest of WAS selected the experiments. 3.3. 3.3. Effect of the carbohydrate matrix Effect of the carbohydrate matrix Generally, the matrix type and the carbohydrate concen- GeneRally, the matrix type and the carbohydrate concentration tration affects significantly the viscosity of the sample and tration significantly affects the viscosity of the sample and consequently the performance of ICP-AES methods. consequently the performance of ICP-AES methods. There- Therefore, the effect of the matrix type and the carbohydrate con- forces, the effect of the matrix type and the carbohydrate concentration on the emission intensity was investigated using centration on the emission intensity investigated Using WAS the GF, SU and HO matrices at increasing concentration (1.0, The GF, DM and HO Matrices at Increasing Concentration (1.0, 2.0, 5.0, 10 and 25% m/v) of carbohydrates. 2.0, 5.0, 10 and 25% w / v) of carbohydrates. The optimum The optimum MD Ioannidou et al. Ioannidou MD et al. / Talanta 65 (2005) 9297 / Talant 65 (2005) 92-97 95 95 Fig. Fig. 2. 2. Effect of matrix type and matrix concentration on the recovered Effect of matrix type and matrix Concentration on the recovered analyte concentration of Co230.786 nm. Concentration of Co-analytic 230.786 nm. ( / ) Glucose and fructose; (/) Glucose and fructo; ( / ) sucrose; ( /) honey. (/) Sucrose, (/ ) honey. Unfilled symbols refer to 10 gl Unfilled symbols refer to 10 gl 1 -1 analyte analytic concentration and filled symbols to 50 gl Filled symbols and Concentration to 50 gl 1 -1 . . The horizontal dotted lines The horizontal dotted lines represent the concentration level of 10 and 50 gl Represent the Concentration Level of 10 and 50 gl 1 -1 aqueous standard standard aqueous solutions in absence of carbohydrate concentration. Concentration in the absence of carbohydrate solutions. of these concentrations defines also the dilution of the real Also defines the tissue concentrations of Dilutional of the real samples, which should be used in order to be analyzed. samples, in order Which should Be Used To Be analyzed. All All the studied analytes were measured at the three spectral lines Were the analytes studied Measured spectral lines at the Three and at two analyte concentrations 10 and 50 gl and analytic at two concentrations 10 and 50 gl 1 -1

, respec- , Respectivetively. tively. Indicative results on the effect of the matrix type and Indicative results on the effect of the matrix type and concentration on Co at two analyte concentrations appear in Concentration on Co concentrations at two analytic Appear in Fig. Fig. 2 . In all cases the recovery was almost the same for the 2. In all Cases the recovery for the WAS Almost The Same low carbohydrate concentrations (12% m/v) in the solution, low carbohydrate concentrations (1-2% w / v) in the solution, while at higher concentrations the recovery decreased and while at higher concentrations and decreased the recovery the baseline becomes unstable. Becomes the baseline unstable. Therefore, 2% m/v carbohy- Therefore, 2% w / v carbohydrate concentration in the solution or 2% m/v dilution of the drate Concentration in the solution or 2% w / v Dilutional of the samples was adopted for further study, as a compromise be- Adopted WAS Frther samples for study, as a compromise between sufficient sensitivity and low matrix influence. sufficient sensitivity and low matrix tween Influence. How- However, higher concentrations (up to 10% m/v) can be used for ever, higher concentrations (up to 10% w / v) Can Be Used For higher sensitivity but with lower precision. But higher sensitivity with lower precision. Statistical paired t -test was used to evaluate the differences Statistical paired t-test used to evaluated the differences WAS in the values obtained using different matrices. Obtained Using the values in different Matrices. In Table 2 In Table 2 results on the experimentally calculated t -values are listed Calculated results on the t-values experimentally is Listed for several analytes at two analyte concentrations, where Several analytic for analytes at two concentrations, WHERE Table 3 Table 3 Slope standard error and correlation coefficients of the calibration equations for all analytes in various matrices Slope standard error of the calibration coefficients and Correlation equations for all analytes in various Matrices Analyte spectral line Analytic spectral line Matrix Matrix Aqueous Aqueous Glucose/fructose Glucose / fructo Sucrose Sucrose Honey Honey Pb220.353 Pb-220 353 6.1 1.4 (0.9770) 6.1 1.4 (0.9770) 7.3 1.4 (0.9520) 7.3 1.4 (0.9520) 5.3 2.2 (0.8111) 5.3 2.2 (0.8111) 5.0 2.2 (0.7962) 5.0 2.2 (0.7962) Cd228.802 Cd-228 802 61.9 0.3 (0.9999) 61.9 0.3 (0.9999) 57.1 0.2 (0.9999) 57.1 0.2 (0.9999) 57.0 1.2 (0.9994) 57.0 1.2 (0.9994) 58.0 0.2 (0.9999) 58.0 0.2 (0.9999)

Cu324.752 Cu-324 752 469 9 (0.9995) 469 9 (0.9995) 474 15 (0.9985) 474 15 (0.9985) 437 13 (0.9986) 437 13 (0.9986) 457 31 (0.9934) 457 31 (0.9934) Cr283.563 Cr-283 563 309 11 (0.9981) 309 11 (0.9981) 285 7 (0.9990) 285 7 (0.9990) 264 13 (0.9960) 264 13 (0.9960) 266 9 (0.9982) 266 9 (0.9982) Co230.786 Co-230.786 36.4 0.3 (0.9999) 36.4 0.3 (0.9999) 33.9 0.2 (0.9999) 33.9 0.2 (0.9999) 33.9 0.2 (0.9999) 33.9 0.2 (0.9999) 33.0 0.3 (0.9999) 33.0 0.3 (0.9999) Ni341.476 Ni-341 476 40.7 3.2 (0.9907) 40.7 3.2 (0.9907) 43.9 1.4 (0.9984) 43.9 1.4 (0.9984) 37.9 3.5 (0.9874) 37.9 3.5 (0.9874) 41.5 2.4 (0.9952) 41.5 2.4 (0.9952) Mn257.610 Mn-257 610 1135 5 (0.9999) 1135 5 (0.9999) 1068 3 (0.9999) 1068 3 (0.9999) 1052 6 (0.9999) 1052 6 (0.9999) 1041 8 (0.9999) 1041 8 (0.9999) Zn213.857 Zn-213 857 511 136 (0.9999) 511 136 (0.9999) 114 88 (0.8161) 114 88 (0.8161) 15 20 (0.4340) 15 20 (0.4340) 201 65 (0.8701) 201 65 (0.8701) Table 2 Table 2 Test of significance ( t -test) of the difference between carbohydrate matrices Test of Significance (t-test) of the difference between carbohydrate Matrices at two analyte concentrations analytic at two concentrations Analyte Analytic Analyte Analytic concentration Concentration ( gl (Gl 1 -1 )) Comparison between matrices Comparison between Matrices tt crit crit GFSU GF-SU a of GFHO GF-HO a of SUHO SU-HO a of Cd CD

10 10 0.66 0.66 0.19 0.19 0.53 0.53 2.14 2.14 Cd CD 50 50 1.82 1.82 2.04 2.04 1.22 1.22 2.14 2.14 Cu With 10 10 0.08 0.08 1.57 1.57 1.49 1.49 2.14 2.14 Cu With 50 50 0.19 0.19 2.60 2.60 2.37 2.37 2.14 2.14 Co Co. 10 10 0.27 0.27 0.63 0.63 0.94 0.94 2.14 2.14 Co Co. 50 50 0.02 0.02 1.28 1.28 0.72 0.72 2.14 2.14 Mn Mn 10 10 1.92 1.92 1.54 1.54 0.70 0.70 2.14 2.14 Mn Mn 50 50 1.08 1.08 0.86 0.86 0.36 0.36 2.14 2.14 a of (GF) glucose/fructose in water; (SU) sucrose in water; and (HO) (GF) glucose / fructo in water, (SU) sucrose in water, and (HO)

honey in water. honey in water. tt crit crit denotes the critical value taken from statistical tables of denotes the critical value of Taken from Statistical Tables the student t -test distribution for 14 degrees of freedom at The Student t-test for 14 degrees of freedom distribution at a confidence level of 95%. the confidence level of 95%. The statistical analysis proved Statistical analysis proved the that, differences obtained between the three different studied That, differences between the Three different studied Obtained matrices (glucose/fructose, sucrose and honey) were not of Matrices (glucose / fructo, sucrose and honey) Were Not of statistical significance at the 95% confidence level (except Statistical Significance at the 95% confidence level (except for Cu at 50 gl for AT 50 gl 1 -1 concentration). Concentration). 3.4. 3.4. Regression analysis Regression analysis Calibration curves were obtained for all the analytes at Calibration curves for all the analytes Were Obtained at three spectral lines for each one in the following matrices: Three spectral lines in the Following for Each One Matrices: AQ, GF, SU, HO. AQ, GF, SU, HO. The carbohydrate concentration in the GF, The carbohydrate Concentration in the GF, SU and HO was 2% m/v. HO WAS and SU 2% w / v. The sensitivity of each calibration The sensitivity of Each calibration is expressed by the slope of the linear regression equation Is Expressed by the slope of the linear Regression Equation in the range 050 gl in the range 0-50 gl 1 -1 . . The slopes ( S ) together with their The Slopes (S) together with Their corresponding standard errors (SEs) and the linear correla- corresponding standard errors (SES) and the linear-correlated tion coefficient ( r ), for each analyte are listed in Table 3 . tion coefficient (r) is analytic for Each Listed in Table 3. Among the three spectral lines of each analyte, those with Among the spectral lines of Three Each analytic, Thos with the better sensitivity ( S ), reproducibility (SEs) and linearity the better sensitivity (S), reproducibility (SES) and linearity ( r ) were given, and were used for the rest the study. (R) Were given, and used for the rest Were the study. As it is As it is shown, the slope values referring to different matrices were Shown, the slope values Were Referring to different Matrices similar and the values of the standard error of the slopes similar and the values of the standard error of the Slopes were very low. Were very low. This observation confirms that the results This observation confirms the results That are independent of the sugar matrix. is independent of the sugar matrix. The correlation coeffi- The Correlation coeffi-

cients for all calibration curves were >0.990, except Pb and cients for all calibration curves Were> 0990, except Pb and Page 5 Page 5 96 96 MD Ioannidou et al. Ioannidou MD et al. / Talanta 65 (2005) 9297 / Talant 65 (2005) 92-97 Zn, illustrating the good linearity in the studied range. Zn, illustrating the good linearity in the range studied. Very Very high sensitivities were obtained in the determination of Mn high sensitivities Were Obtained in the Determination of Mn followed by Cu, Zn, Cr, Cd, Co and Ni. FOLLOWED BY Cu, Zn, Cr, Cd, Co and Ni. The sensitivity in The sensitivity in the determination of Pb was rather low. Determination of Pb Rather the low WAS. 3.5. 3.5. Accuracy and precision of the method Accuracy and precision of the method Usually, in order to assess the accuracy of a method, stan- Usually, in order to Assess the Accuracy of the method, standard reference materials are analyzed, the obtained results Dard reference materials is analyzed, the results Obtained are compared with the certified ones and the absolute or the is compared with the certified ones and the absolute or the relative error is calculated. Calculated relative error is. However, an SRM material with However, with material SRM year certified metals concentration is not available for honey, as certified Metals Concentration is not available for honey, I already mentioned by Caroli et al. [17,18] . These authors as- already mentioned by Caroli et al. [17.18]. tissue authors ascribed the lack of a honey-based certified reference material cribed the Lack of a honey-based certified reference material due to its practical problems of transformation into a sta- due to practical problems of transformation into ITS sitble, homogeneous mass, adequate for a certification project. ble, homogeneous mass, Adequate for the certification project. Therefore, the accuracy of the proposed method was eval- Therefore, the Accuracy of the proposed method eval-WAS uated by determining the recoveries of the analytes at var- uated by determining the recoveries of the analytes at various concentrations (1, 5, 25 and 50 gl ious concentrations (1, 5, 25 and 50 gl 1 -1 ), using calibra- ), Using calibratetion curves at different matrices. tion curves at different Matrices. Results from the recovery Results from the recovery experiments appear in Table 4 . Quantitative recoveries were Appear experiments in Table 4. Quantitative recoveries Were obtained at high metal concentrations (25 and 50 gl Obtained at high metal concentrations (25 and 50 gl 1 -1 ) and ) And they were independent of the matrix used. Were they independent of the matrix used. As it is shown, As it is Shown,

the recoveries obtained for all analyte concentrations were Obtained recoveries for all the analytic concentrations Were ranged between 93 and 110%. ranged between 93 and 110%. The recoveries of Cd and Mn The recoveries of Cd and Mn were very good, while those of Cu, Cr, Co and Ni were ad- Were very good, while Thos of Cu, Cr, Co and Ni Were ad equate at the 25 and 50 gl Equator at the 25 and 50 gl 1 -1 levels. levels. On the other hand Pb On the Other hand Pb and Zn recoveries were not satisfactory ( < 80%). Were Not satisfactory recoveries and Zn (<80%). The precision of the method was evaluated using the rel- The precision of the method Using the rel-evaluated WAS ative standard deviation, s Ativ Standard Deviation, and rr , of repeated determination of , Of Repeated Determination of the analytes at 25 gl the analytes at 25 gl 1 -1 concentration level, in presence of Concentration level in the Presence of 2% m/v carbohydrate in the solution. 2% w / v carbohydrate in the solution. For Cd, Cu, Cr, Co, For Cd, Cu, Cr, Co, Ni, and Mn the s Ni, Mn and the s rr ranged between 0.3 and 17% as listed in ranged between 0.3 and 17% as Listed in Table 5 . The variation of s Table 5. The variation of s rr is similar for all matrices tested. is similar for all Matrices Tested. However, the repeatability of the method was not adequate However, the Repeatability of the Method Was Not Adequate in the case of Pb and Zn, and in some cases it was higher in the case of Pb and Zn, and in Some Cases IT WAS higher than 50%. Than 50%. Table 4 Table 4 Recovery of analytes at aqueous and carbohydrate matrices Recovery of analytes at aqueous and carbohydrate Matrices Matrix Matrix Added Added Recovery (%) Recovery (%) Cd CD Cu With Cr Cr Co Co. Ni Us Mn Mn AQ AQ 50 gl 50 gl 1 -1 94 94 104 104 88 88

93 93 91 91 93 93 GF GF 102 102 103 103 95 95 99 99 86 86 98 98 SU SU 102 102 112 112 103 103 99 99 94 94 100 100 AQ AQ 25 gl 25 gl 1 -1 94 94 115 115 97 97 93 93 102 102 93 93 GF GF 102 102 113 113 106 106 100 100 103 103 98 98 SU SU 102 102 123 123 114 114 100 100 121 121 100 100 AQ AQ 5 gl 5 gl 1 -1 101 101 200 200 123 123 96 96 52 52 96 96 GF GF

109 109 200 200 134 134 103 103 49 49 102 102 SU SU 110 110 217 217 144 144 103 103 54 54 103 103 AQ AQ 1 gl 1 gl 1 -1 99 99 nq NQ nq NQ nq NQ nq NQ 107 107 GF GF 106 106 nq NQ nq NQ nq NQ nq NQ 113 113 SU SU 107 107 nq NQ nq NQ nq NQ nq NQ 115 115 nq: not quantified. NQ: not quantified. Table 5 Table 5 Precision of the method at various matrices (25 gl Precision of the method at various Matrices (25 gl 1 -1 of the analyte, of the analytic, relative standard deviation s relative standard deviation s rr %, n = 10) %, N = 10) Analyte Analytic Matrix Matrix Aqueous Aqueous Glucose/fructose Glucose / fructo Sucrose Sucrose

Honey Honey Pb Pb 6.2 6.2 19 19 25 25 >50 > 50 Cd CD 1.0 1.0 0.7 0.7 1.1 1.1 1.3 1.3 Cu With 2.8 2.8 2.2 2.2 2.5 2.5 1.0 1.0 Cr Cr 7.9 7.9 6.6 6.6 6.9 6.9 1.9 1.9 Co Co. 1.0 1.0 1.7 1.7 3.2 3.2 1.6 1.6 Ni Us 3.7 3.7 17 17 5.6 5.6 15 15 Mn Mn 0.3 0.3 0.6 0.6 0.5 0.5 0.6 0.6 Zn Zn 17 17 >50 > 50 >50 > 50 >50 > 50 Table 6 Table 6 Detection limits ( gl Detection limits (gl 1 -1 ) obtained for all analytes in aqueous and carbo- ) Obtained for all analytes in aqueous and carbohydrate matrices Hydra Matrices Analyte Analytic Aqueous Aqueous Glucose/fructose Glucose / fructo

Sucrose Sucrose Honey Honey Pb Pb 6.7 6.7 4.5 4.5 9.7 9.7 8.6 8.6 Cd CD 0.7 0.7 1.2 1.2 0.9 0.9 1.4 1.4 Cu With 1.5 1.5 1.0 1.0 4.8 4.8 1.0 1.0 Cr Cr 17 17 6.8 6.8 15 15 5.7 5.7 Co Co. 0.4 0.4 0.8 0.8 2.0 2.0 1.8 1.8 Ni Us 7.3 7.3 4.3 4.3 16 16 3.1 3.1 Mn Mn 0.2 0.2 0.2 0.2 1.0 1.0 0.2 0.2 Zn Zn 48 48 139 139 309 309 29 29 The detection limit ( c The detection limit (c LL ) was calculated according to the ) WAS Calculated According to the 3 s criterion (three times the standard deviation of 10 blank And Criterion 3 (Three times the standard deviation of 10 blank measurements). Measurements). The results are listed in Table 6 . The lower The results is Listed in Table 6. The lower

detection limits were obtained for Mn, Cd, Co and Cu, mod- Were Obtained detection limits for Mn, Cd, Co and Cu, aerate for Cr, Ni and Pb and higher for Zn. corrections for Cr, Ni and Pb and higher for Zn. Consequently, the Consequently, the detectability of the proposed method is considered satisfac- detectability of the proposed method is Considered meettory for routine analysis and quality control. tory for routine analysis and quality control. Similar detec- Similar detectiontion limits for Mn, Cd, Co and Cu have been also obtained tion limits for Mn, Cd, Co and The Also have Been Obtained using ICP-AES after a preliminary mineralization stage [15] . Using ICP-AES after a preliminary stage mineralization [15]. The ion chromatographic and the voltammetric determina- The ion chromatographic and voltammetric caused the tion of the elements Pb, Cd, Cu, Co and Ni gave much tion of the elements Pb, Cd, Cu, Co and Ni Gave much higher detection limits (40, 15, 3, 2 and 4 gl higher detection limits (40, 15, 3, 2 and 4 gl 1 -1 , respec- , Respectivetively) [3] a fter laborious mineralization stage. tively) [3] fter laborious mineralization stage. On the other On The Other hand, lower detection limits for Pb, Cd, Cu and Zn were hand, lower detection limits for Pb, Cd, Cu and Zn Were reported only in case of using the ETAAS technique [1] . Reported only in case of ETAAS Using the technique [1]. 3.6. 3.6. Analysis of commercial samples of honey and sugars Analysis of commercial samples of honey and sugars Samples of five commercial products of honey, two of Samples of five commercial products of honey, two of sugar, one of glucose and one of fructose were analyzed infant, one of glucose and one of fructo Were analyzed with respect to their content in heavy metals. with respect to heavy metals in Their content. The final The final concentrations of the analytes (in mg kg concentrations of the analytes (in mg kg 1 -1 ) were calcu- ) Were calculating lated relative to the initial sample mass and corrected for lated relative to the initial sample mass and corrected for any dilutions. Any dilutions. Honey, glucose, fructose and sugar samples Honey, glucose, sugar and fructo samples were pre-treated as described in Section 2 . The results of Were Treated as described in the pre- Section 2. The results of mean concentrations and their standard deviation are listed Their mean and standard deviation concentrations is Listed in Table 7 . Zn and Pb were not quantified due to their poor in Table 7. Zn and Pb Were Not quantified due to poor Their repeatability, as it is given in Table 5 . Chromium was not Repeatability, as it is given in Table 5. Chromium Was Not

MD Ioannidou et al. Ioannidou MD et al. / Talanta 65 (2005) 9297 / Talant 65 (2005) 92-97 97 97 Table 7 Table 7 Analytical results of commercial sample analysis (concentrations in mg kg Analytical results of commercial sample analysis (concentrations in mg kg 1 -1 , mean standard deviation, n = 7) , Mean standard deviation, n = 7) Sample Sample Cd CD Cu With Co Co. Mn Mn Ni Us Sugar A Sugar A 0.087 0.013 0087 0013 0.16 0.01 0.16 0.01 0.131 0.012 0131 0012 nd when 0.56 0.06 0.56 0.06 Sugar B Sugar B 0.084 0.016 0084 0016 0.19 0.03 0.19 0.03 0.039 0.007 0039 0007 nd when 0.24 0.03 0.24 0.03 Glucose Glucose 0.084 0.009 0084 0009 0.14 0.02 0.14 0.02 0.036 0.006 0036 0006 nd when 0.09 0.02 0.09 0.02 Fructose Fructo 0.088 0.014 0088 0014 0.14 0.03 0.14 0.03 0.080 0.016 0080 0016 nd when 0.35 0.05 0.35 0.05 Honey 1 Honey 1 0.089 0.010 0089 0010 0.48 0.03 0.48 0.03 0.071 0.010 0071 0010 7.22 0.2 7.22 0.2 nd when Honey 2 Honey 2 0.110 0.024 0110 0024 nd when 0.042 0.006 0042 0006 0.20 0.03 0.20 0.03 nd when

Honey 3 Honey 3 0.078 0.013 0078 0013 nd when 0.010 0.003 0010 0003 0.18 0.03 0.18 0.03 nd when Honey 4 Honey 4 0.097 0.017 0097 0017 0.28 0.05 0.28 0.05 0.078 0.012 0078 0012 0.11 0.02 0.11 0.02 nd when Honey 5 Honey 5 0.222 0.031 0222 0031 0.45 0.04 0.45 0.04 0.087 0.015 0087 0015 0.12 0.03 0.12 0.03 nd when nd: not detected. nd: not detected. detected in all samples. detected in all samples. The concentration of Cd measured The Concentration of Cd Measured was low (0.0780.222 mg kg Low WAS (0078-0222 mg kg 1 -1 ), as well as copper levels ), As well as copper levels (0.140.48 mg kg (0.14-0.48 mg kg 1 -1 ) and similar to those found in honey ) And similar to Thos found in honey samples from various origins [1315] . The Co concentra- samples from various Origins [13-15]. The Co concentrationtion was very low (0.0100.131 mg kg tion WAS very low (0010-0131 mg kg 1 -1 ) and lower than ) And Lower Than the reported values. The Reported values. Ni was not detected in all honey sam- We Was Not detected in all honey Satples and its values measured in sugars were in the range Plese Measured values in sugars and STIs in the range Were 0.090.56 mg kg 0.09-0.56 mg kg 1 -1 . . Finally, Mn was not detected in any of Finally, Mn Was Not Detected in Any of the sugars, in contrast to the honey samples where values The sugars, in contrast to the honey samples WHERE values ranged between 0.11 and 0.20 mg kg ranged between 0.11 and 0.20 mg kg 1 -1 were measured in all Were Measured in all but the first sample, for which, the concentration of Mn was But the first sample, for Which, the Concentration of Mn WAS found to be very high (7.2 mg kg found very high to Be (7.2 mg kg 1 -1 ). ). High values have been High values have Been

reported in the literature [13,15] , which can be attributed ei- Reported in the literature [13.15], they attributed Which Can Bether to the production stages of honey or to the region from There to the production stages of honey from or to the region where this honey has been taken. WHERE HAS Been Taken this honey. 4. 4. Conclusions Conclusions The developed method was proved efficient and re- The developed method proved efficient and re-WAS liable for the simultaneous and direct determination of liable for the simultaneous direct and Determination of several heavy metals in routine analysis of honey and Several heavy metals in routine analysis of honey and sugars using ICP-AES in the optimal instrument's param- Using sugars in the ICPAES instrument's optimal parameters (2 ml min eters (2 ml min 1 -1 sample flow rate and 1300 W rf power). sample flow rate and rf power 1300 W). As it was shown, the type of the carbohydrate matrices IT WAS As Shown, the type of the carbohydrate Matrices (glucose/fructose, sucrose, honey) did not have significant (Glucose / fructo, sucrose, honey) did have Not Significant effect on the performance of the proposed method. effect on the performance of the proposed method. Also it Also it was shown that a 2% m/v concentration of the carbohydrate Shown That WAS a 2% w / v Concentration of the carbohydrate is a compromise for sufficient sensitivity and low matrix is of sufficient sensitivity and compromised for low matrix effect. effect. Therefore, the samples were analyzed after such Therefore, the samples Were analyzed after Suche dilution. Dilutional. Furthermore, calibration curves were drawn at Furthermore, calibration curves Were drawn at the aforementioned matrices, and it was proved that mixed the aforementioned Matrices, and it WAS proved That mixed aqueous standards can be readily employed in preparing aqueous Standards Can Be readily employed in Preparing the calibration curves. The calibration curves. The accuracy of the method was The Accuracy of the method WAS investigated by measuring the recoveries at several metal investigated by measuring the metal recoveries at Several concentrations in spiked solutions, and the precision by the Spiked concentrations in solutions, and the precision by the relative standard deviation. relative standard deviation. Using the proposed method, the Using the proposed method, the obtained recoveries were very good and the detection limits Were Obtained very good recoveries and detection limits the of Mn, Cd, Co, Cu and Ni were sufficient for routine anal- of Mn, Cd, Co, Cu and Ni Were enough for routine analysis of commercial samples of honey, sugar, glucose and ysis of commercial samples of honey, sugar, glucose and fructose. Fructo.

References References [1] P. Vinas, I. Lopez-Garcia, M. Lanzon, M. Hernandez-Cordoba, J. [1] P. Vinas, I. Lopez-Garcia, M. Lanzon, M. Hernandez-Cordoba, J. Agric. Agric. Food. Food. Chem. Chem. 45 (1997) 3952. 45 (1997) 3952. [2] P. Przybylowski, A. Wilczynska, Food Chem. [2] P. Przybylowski, A. Wilczynska, Food Chem. 74 (2001) 289. 74 (2001) 289. [3] PL Buldini, S. Cavalli, A. Mevoli, J. Lai Sharma, Food Chem. [3] PL Buldini, S. Cavalli, A. Mevoli, J. Lai Sharma, Food Chem. 73 73 (2001) 487. (2001) 487. [4] LB Allen, PH Siitonen, HC Thompson, J. Agric. [4] LB Allen, PH Siitonen, HC Thompson, J. Agric. Food Chem. Food Chem. 45 45 (1997) 162. (1997) 162. [5] J. Kubota, RM Welch, DR Van Campen, Environ. [5] J. Kubota, RM Welch, DR Van Campeni, Environ. Geochem. Geochem. Health Health 14 (1992) 91. 14 (1992) 91. [6] G. Sanna, MI Pilo, PC Piu, A. Tapparo, R. Seeber, Anal. [6] G. Sanna, MI pilots, PC Piu, A. Tapparo, R. Seeber, Anal. Chim. Chem. Acta 415 (2000) 165. Acta 415 (2000) 165. [7] S. Caroli, G. Forte, AL Iamiceli, B. Galoppi, Talanta 50 (1999) 327. [7] S. Caroli, G. Forte, AL Iamiceli, B. Galopp, Talant 50 (1999) 327. [8] JS Rowarth, J. Apicult. [8] JS Rowarth, J. beekeepers. Res. Res. 29 (1990) 177. 29 (1990) 177. [9] JL Rodriguez-Otero, P. Paseiro, J. Simal, L. Terradillos, A. Cepeda, [9] JL Rodriguez-Otero, P. Paseiro, J. Simal, L. Terradillos, A. Cepeda, J. Apicult. J. beekeepers. Res. Res. 31 (1992) 65. 31 (1992) 65. [10] E. Sauri-Duch, J. del C., C. Hernandez-Chavez, Rev. [10] E. sacks-Duch, J. del C., C. Hernandez-Chavez, rev. Esp. Esp. Cienc. Ciencia. Technol. Technol. Aliment 34 (1994) 450. Aliment 34 (1994) 450. [11] I. Lopez-Garcia, P. Vinas, C. Blanco, M. Hernandez-Cordoba, Talanta [11] I. Lopez-Garcia, P. Vinas, C. Blanco, M. Hernandez-Cordoba, Talant 49 (1999) 597. 49 (1999) 597. [12] P. Vinas, N. Campillo, I. Lopez-Garcia, M. Hernandez-Cordoba, [12] P. Vinas, N. Campillo, I. Lopez-Garcia, M. Hernandez-Cordoba, Talanta 42 (1995) 527. Talent 42 (1995) 527. [13] RA Morse, DJ Lisk, Am. [13] RA Morse, DJ Liska, Am. Bee J. (1980) 522. Bee J. (1980) 522. [14] R. Petkov, I. Getchev, S. Kuntcheva, Apiacta XXXIII (1998) 107. [14] R. Petkov, I. Getchev, S. Kuntcheva, Apiacta XXXIII (1998) 107. [15] J. Devillers, JC Dore, M. Marenco, F. Poirier-Duchene, N. Galand, [15] J. DEVILLERS, JC Dore, M. Marenco, F. Poirier-Duchene, N. Galand, C. Viel, J. Agric. C. Viel, J. Agric. Food Chem. Food Chem. 50 (2002) 5998. 50 (2002) 5998. [16] P. Fodor, E. Molnar, Mikrochim. [16] P. Fodor, E. Molnar, Mikrochim. Acta 112 (1993) 113. Acta 112 (1993) 113. [17] S. Caroli, G. Forte, M. Alessandrelli, R. Cresti, M. Spagnoli, [17] S. Caroli, G. Forte, M. Alessandrelli, R. Crest, M. Spagnoli, S. D'Ilio, J. Pauwels, GN Kramer, Microchem. S. D'Ilio, J. Pauwels, GN Kramer, Microchem. J. 67 (2000) J. 67 (2000) 227. 227.

[18] G. Forte, S. D'Ilio, S. Caroli, J. AOAC Int. [18] G. Forte, S. D'Ilio, S. Caroli, J. AOAC Int. 84 (2001) 1972. 84 (2001) 1972.