- Email: [email protected]
Determination of nickel, chromium and cobalt in wheat flour using slurry sampling electrothermal atomic absorption spectrometry
Talanta 48 (1999) 1051 – 1060
Determination of nickel, chromium and cobalt in wheat flour using slurry sampling electrothermal atomic absorption spectrometry Mar Gonza´lez, Mercedes Gallego, Miguel Valca´rcel * Department of Analytical Chemistry, Faculty of Sciences, Uni6ersity of Co´rdoba, E-14004 Co´rdoba, Spain Received 3 June 1998; received in revised form 29 September 1998; accepted 2 October 1998
Abstract The slurry technique was applied to the determination of Ni, Cr and Co in wheat flour by electrothermal atomic absorption spectrometry (ETAAS). The influence of the graphite furnace temperature programme was optimized. Optimum sensitivity was obtained by using a mixture of 15% HNO3 – 10% H2O2 as suspended medium for a 3% w/v slurry in the determination of Ni; lower concentrations of HNO3 were necessary for the determination of Co and Cr (viz. 5 and 10%). The precision of direct analyses of the slurries was improved by using mechanical agitation between measurements; thus, the RSD of the measurements was ca. 5% for repeatability. The direct slurry sampling (SS) technique is suitable for the determination of Ni and Cr in wheat flour samples at levels of 150 – 450 and 30 – 72 ng g − 1, respectively, as it provides results similar to those obtained by ashing the sample. However, the typically low level of Co in these samples precluded its determination by the proposed method (the study was made in an SRM spiked wholemeal flour), at least in those samples that were contaminated with elevated concentrations of the metal (viz. more than 90 ng of Co per g of flour). The method provides a relative standard deviation of 6, 8, and 4% for Ni, Cr, and Co, respectively. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Electrothermal atomic absorption spectrometry; Slurry sampling; Wheat flour; Nickel; Chromium; Cobalt
1. Introduction Cereal flours are staple foods in most countries. Thus, wheat flour is consumed every day in bread, cake, sauces, etc. Analyses for trace element in flours are therefore important from both a nutritional and a toxicological point of view. Soil is the * Corresponding author. Tel.: + 34-57-218614; fax: +3457-218606; e-mail: [email protected].
main vehicle by which heavy metals enter plants; in response, a European Union directive (86/278/ CEE) has been issued to control the maximum allowable contents of some metals in agricultural sewage sludge [1]. The average daily intake in food for metals is well documented; for example Ni, Cr, and Co occur at very low concentrations in cereals (viz. 100–400, 20–50 and 10–20 ng g − 1, for Ni, Cr, and Co, respectively) [2]. While Cr, Co, and Ni have been shown to be essential
0039-9140/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 9 1 4 0 ( 9 8 ) 0 0 3 1 7 - 8
1052
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
for humans, the necessary levels of Ni have not been quantified. Trace metals can be determined by using various techniques. In recent times, the ICP-MS [3] technique has gained momentum for this purpose. However, ETAAS has for some time been at most laboratories and used in many determinations of Ni, Cr, and Co. The low concentrations of Co normally found in cereals usually require preconcentration prior to the determination proper. Co in feed grains has been determined in this way following digestion [4], as has in cereals after decomposition of organic matter and extraction into a 2-nitroso-1-naphthol solution in xylene [5], or in heptan-2-one [6]. Ni in various matrices including rice flours has also been determined by ETAAS, following microwave-assisted digestion [7]; Cr in grain and cereal products require prior wet digestion [8]. Cr and Ni, in addition to other metals, were determined in wheat flour by using a graphite boat with direct Zeeman-AAS [9]. Co and Ni were determined in cereals after digestion with the HNO3/HClO4 mixture with recoveries from 87 to 104% [10]. The contents of Ni, Cr, and Co in foods including cereals on the Swedish market between 1983 and 1990 were studied by the Nordic Committee on Food Analysis; the official methodology selected for determination of different elements was dry ashing followed by ETAAS [11]. Slurry sampling (SS) ETAAS methodology, originally developed by Brady et al. [12] is by now well established and widely used in the determination of trace elements in food samples. The most attractive advantages of SS over dry or wet ashing can be summarized as follows: (i) it reduces sample pretreatment and analysis times; (ii) it minimizes contamination/loss risks; (iii) it uses a conventional sample introduction system; (iv) it ensures appropriate calibration with aqueous standards; and (v) it provides acceptable accuracy and precision. However, this technique has some disadvantages that arise essentially from non-homogeneous distribution of the trace elements in the slurry as well as differences in the chemical species under which they
are present [13]. The benefits of SS were demonstrated in an international collaborative study involving 25 laboratories that was intended to assess the state of the art in the technique. Preliminary results [14] suggested that SS is mature enough for routine analyses; at a later stage, the usefulness of ultrasonic SS was evaluated [15]. The results showed that extracting analyte into the liquid phase of the slurry is not a prerequisite for accurate slurry analyses. Slurry sampling for electrothermal atomization is a very active, widely documented area applications of which were reviewed [16]. A comprehensive review of SS for foods in atomic spectrometry has also been published [17]. Reported applications involving cereals are scant relative to other foods. Also, Co and Ni have been determined less frequently than Cr in this way. Vin˜as et al. used their experience in the use of slurry procedures to determine Co and Ni in vegetables and legumes [18], and Cr in vegetables [19], following treatment of dried samples with the ethanol–H2O2 –HNO3 mixture; in all instances, calibration was against aqueous standards. Direct slurry sampling was used to determine Ni and Co in rice by ICP-MS [20], and Cr and Ni in spinach leaves by simultaneous multi-element atomic absorption spectrometer with continuum source (SIMAAC) [21]. In the latter application, the slurry solution (5– 10 mg of sample in 5 ml of 5% HNO3) was homogenized by immersing an ultrasonic probe in the autosampler cup, which avoided settling of particles. In this work, the potential of slurry sampling for the determination of Ni, Cr, and Co in wheat flour by ETAAS was explored. The influence of various parameters such as the drying and pyrolysis time, pyrolysis and atomization temperature, and presence of modifier on the atomic signal was studied. Efforts were aimed at using the advantages of the HNO3 –H2O2 mixture as the medium for sample preparation in order to transfer the analytes to the aqueous phase. The proposed method was applied to various samples with acceptable recoveries for Ni and Cr.
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
1053
Table 1 Instrumental parameters and optimized furnace conditions for the determinations of Ni, Cr and Co in wheat floura
Lamp current/mA Wavelength/nm Bandpass/nm
Step Dry 1 Dry 2 Pyrolysis Atomize Clean a
Ni
Cr
Co
25 232.0 0.2
10 357.9 0.7
30 240.7 0.2
Temperature (°C)
Ramp (s)
Hold (s)
Temperature
Ramp Hold
Temperature
Ramp Hold
100 300 1400 2600 2650
5 10 20 0 1
15 30 30 0 1
100 300 1700 2600 2650
5 10 20 0 1
100 300 1400 2600 2650
5 10 20 0 1
15 30 30 6 3
15 30 30 6 3
A stream of argon at 300 ml min−1 was used (the flow was stopped during the atomization step); injected volume, 20 ml
2. Experimental
2.1. Apparatus Slurry samples were analysed by using a model 1100-B atomic absorption spectrometer from Perkin-Elmer (U8 berlingen, Germany) equipped with a deuterium-lamp background corrector, an HGA-700 graphite furnace, and an AS-70 autosampler, and interfaced to an Epson FX-850 printer. Analyses were carried out by using platforms inserted into pyrolytically coated graphite tubes (Perkin-Elmer) and measurements (integrated absorbance peak areas) were made by using single-element hollow cathode lamps (Perkin-Elmer). Argon was used as sheeting gas for the furnace in all cases. The operational parameters recommended by the manufacturer for Ni, Cr, and Co are listed in Table 1. Slurries were homogenized in an ultrasonic bath (Bandelin, Tk52, Berlin, Germany) or a vortex mixer (Heidolph, Kelheim, Germany).
2.2. Reagents Working metal standards were prepared daily −1 from a 1000 mg l stock metal solution (Panreac, Barcelona, Spain) by diluting appropriate aliquots with 0.2% HNO3. A 10.0 g l − 1 solution of magnesium nitrate (Merck, Darmstadt, Germany) was
used as matrix modifier. Triton X-100 (Serva Feinbiochemica, Heidelberg, Germany) was tested as stabilizing agent.
2.3. Certified reference material SRM wholemeal flour no. 189, with reference contents (non-certified values) for Ni and Cr obtained from the European Commission (Belgium) was dried to constant mass in an oven at 103°C as per the supplier’s recommendations, and used for method validation. Because the Co concentration in this reference material was too low for its determination by the slurry sampling technique, the material was spiked with this metal from an aqueous salt solution. For this purpose, 40 ml of −1 a solution containing 250 ng ml Co was added to 20.0 g of the SRM. The slurry thus obtained was dried at room temperature for 2 weeks in a closed fume hood to avoid contamination and then to constant mass in an oven at 103°C. The wholemeal flour spiked with 500 ng of cobalt per g of sample was employed to optimize the cobalt determination in this matrix.
2.4. Sample preparation Flour samples were prepared at a pilot plant from wheat produce in different Spanish locations. An amount of 0.5–1 kg of wheat contain-
1054
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
ing ca. 15% w/w of water was ground in a metal rolling mill to obtain 60% of white flour and 40% of byproducts. The flour was screened with a 130 mm sieve. All other samples were purchased at a local supermarket.
2.5. Procedures In order to avoid contamination, all polytetrafluoroethylene (PTFE) materials, pipettes, and calibrated flasks were immersed in freshly made 10% HNO3 for 24 h and then rinsed thoroughly with high purity water (Milli-Q Water System, Millipore, Madrid, Spain) before use. For slurry analyses, reference and sample materials were accurately weighed (ca. 150 mg) into PTFE tubes and supplied with 5 ml of 15% HNO3 containing 10% v/v H2O2. Slurries (3% w/v, i.e. 3 g into 100 ml) were homogenized by agitating for 15 min in an ultrasonic immediately before each analysis. Aliquots of 2 ml were placed in the 2 ml polyethylene vials of the autosampler; between measurements, samples were mixed in a vortex shaker for 5 s in order to ensure reproducible results using autosampler cups provided with covers. Five sample replicates were analysed in each case. The sample blanks contained the same concentration of nitric acid and hydrogen peroxide as the slurry samples. Calibration graphs (spanning −1 −1 the range 0– 20 mg l for Cr, 0 – 40 mg l for Co, −1 and 0–50 mg l for Ni) were obtained from variable volumes of standard solutions containing −1 30, 40, and 50 mg l Cr, Co, and Ni, respectively, that were mixed in the graphite tube with appropriate amounts of 0.2% HNO3 to a volume of 20 ml; 20 ml of 0.2% HNO3 was used as blank. For mineralization of the flours, 3.0 g of material was accurately weighed into a platinum crucible and carbonized in a burner at a low temperature for about 3 h; then, the black residue was supplied with several drops of concentrated H2O2 and ashed in a muffle furnace at 600–650 °C for 2 h. The completely ashed sample obtained was dissolved in 10 ml of 0.5% HNO3; three sample replicates and four injections per replicate were analysed. Cr and Co were measured directly, whereas Ni required ten-fold dilution in 0.2% HNO3.
3. Results and discussion The different factors that influence the performance of slurry ETAAS have been comprehensively examined by several authors [14–17,22–25]. Slurry introduction has been found to pose some problems related to concentration and particle size, which influence stability, deposition and atomization efficiency; these in turn may affect the accuracy and precision of analyses. Ensuring accurate results and good reproducibility in this context entails using homogeneous slurries; mechanical mixing devices, gas bubbling, and ultrasonic agitation has proved useful for this purpose [23]. Thus, Miller-Ihli [26] designed an ultrasonic, pneumatically movable slurry sampler and, more recently, Lo´pez-Garcı´a et al. [27] based on previous experiments of other authors [23], developed an efficient slurry sampling device that uses argon bubbles to homogenize slurries in the autosampler cups without the need to alter the instrument operation.
3.1. Furnace temperature programs and chemical modifiers The temperature programs and the effect of using magnesium nitrate as modifier on both standards and slurries were carefully examined. The SRM whole meal flour spiked with 0.5 mg of cobalt per g was the sample used in this study. In all instances, slurries of 3% w/v were prepared in 15% HNO3 containing 10% H2O2. A 15 ml volume of standard/slurry and 5 ml of chemical modifier −1 (10 g l magnesium nitrate) were injected into the pyrolytic graphite tube with platform. Two drying steps were required to ensure mild, totally dry conditions for the sample, and no splattering; the slurry sample required higher temperatures (100 and 300°C) than the standard (100 and 140°C).
3.1.1. Nickel The pyrolysis and atomization temperatures were tested, in the absence of chemical modifier, over the ranges 500–1900°C, and 1900–2650°C, respectively. At a constant atomization temperature of 2600°C, the pyrolysis temperature did not
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060 −1
affect the signal for the standard (10 mg l Ni) up to 1500°C; on the other hand, the optimum pyrolysis temperature for the slurry sample was 1400°C. A similar behavior was observed with magnesium nitrate as modifier. Peak areas increased with increasing atomization temperature up to 2600°C, both for the standard and for the sample. As can be seen in Fig. 1A, the background signal for the slurry sample decreased at high atomization temperatures. On the other hand, the use of magnesium nitrate at the optimum atomization temperature (2600°C) increased of background signal for the slurry sample (see Fig. 1A). Although this modifier is recommended for the determination of nickel in different matrices, its use in this case stabilized a higher proportion to the concomitants. Based on these considerations, the pyrolysis and atomization temperature were fixed at 1400°C and 2600°C, respectively, and no chemical modifier was employed for this element.
3.1.2. Cobalt The pyrolysis temperature had no effect on the determination of the standard (10 mg l − 1 Co) neither on the spiked slurry whole meal over the range 800–1400°C. As no cobalt was lost up to 1400°C, this temperature was fixed while the atomization temperature was changed. The best results were obtained at an atomization temperature of 2600°C. The use of magnesium nitrate as modifier provided no advantages for the whole meal slurry sample (the background signal was ca. 0.050 A s with and without modifier); on the other hand, the modifier increased the background signal for the standard solution two-fold (which was 0.060 and 0.120 A s in its absence and presence, respectively). In order to facilitate reliable deuterium background correction, use of the modifier was discarded. 3.1.3. Chromium The maximum pyrolysis temperature was the same with and without modifier, 1700°C, for both −1 the standard solution (5 mg l Cr) and the slurry (3% whole meal in the HNO3 – H2O2 mixture). Moreover, the addition of 50 mg of magnesium nitrate increased slightly both the atomic absorp-
1055
tion and background signals, so its presence it provided no advantage. As can be seen from Fig. 1B, better separation of atomic and background signals was obtained in the absence of this modifier. On the other hand the addition of 5 ml of palladium nitrate (1.0 g l − 1) to the slurry sample provided results similar to those obtained with magnesium nitrate for the three elements studied. Therefore, the temperature programs listed in Table 1 were employed for standards and slurries (prepared in the HNO3 –H2O2 mixture), with no chemical modifier.
3.2. Optimization of the slurry preparation The factors potentially influencing the accuracy and precision of the analytical results were studied, namely suspension medium, the homogenization and stability of the slurry, and slurry concentration (% w/v) were examined in order to optimize preparation of the slurries. The effect of the slurry particle size was not studied as all flours were finely powdered (less than 130 mm), which is similar to the SRM whole meal flour (less than 125 mm). The slurries of SRM whole meal flour spiked with cobalt, at 3% w/v, were prepared at variable HNO3 concentrations (0–20% v/v) and contained 10% H2O2. Agitation for 5 min in a vortex mixer was needed to homogenize the slurry before it was transferred into the autosampler cup and 20 ml was injected into the graphite furnace. As can be seen in Fig. 2, the nitric acid concentration had a marked effect on the Ni signal, and somewhat lesser effects on the Cr and Co signals. A concentration of nitric acid of 5%, 10%, and 20% for Co, Cr, and Ni, respectively, was needed to favor the extraction of the analytes into solution. The determination of Cr and Co required no addition of H2O2 to the slurry sample when prepared in 10% HNO3; however, that of Ni required 10% H2O2 and high nitric acid concentration (20%). The higher nitric acid concentration required for the determination of Ni can be explained probably because Co and Cr are extracted more efficiently into the liquid phase than is Ni or because these elements are more efficiently determined in the
1056
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
Fig. 1. Atomic absorption profile and background lines for Ni (A) and Cr (B) in SRM wholemeal slurry samples (3% w/v in 15% HNO3 containing 10% H2O2). A, Ni determination (pyrolysis temperature, 1400°C): (a) and (b) in the absence of chemical modifier, at an atomization temperature of 2250°C and 2600°C (background signals, 0.154 and 0.071 A s, respectively); and (c) in the presence of 50 mg of Mg(NO3)2, at a atomization temperature of 2600°C (background signal, 0.122 A s). B, Cr determination with Mg(NO3)2 (a) and without chemical modifier (b) (pyrolysis and atomization temperature, 1700°C and 2600°C, respectively). Solid and dashed lines represent atomic and background signals, respectively. Injected slurry volume, 15 and 20 ml for Ni and Cr, respectively.
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
1057
Fig. 2. Effect of the concentration of the nitric acid as the medium used to prepare the slurry of SRM wholemeal flour spiked with cobalt. All slurries also contained 10% H2O2.
solid phase than is Ni. In order to decrease the high nitric acid concentration (20% v/v) required by the Ni determination, the effectiveness of ultrasonic and mechanical (vortex) agitation for slurry homogenization was studied. For this purpose, various samples of 3% w/v SRM whole meal spiked flour slurry were prepared in a medium containing 15% HNO3 plus 10% H2O2 for all of the elements. The slurries were shaken in a vortex mixer or an ultrasonic bath for 1 – 30 min. Both homogenization devices required at least 15 min to ensure stabilization of integrated absorbance average values (n =3 at each time). However, ultrasonic mixing provided higher precision (RSD 6.0, 3.5, and 12.6% for Ni, Co, and Cr, respectively) than mechanical stirring (RSD 7.5, 6.5, and 19.0%, respectively). As Cr detection was scarcely precise, Triton X-100 was tested as a stabilizing agent. Thus, several samples of 3% w/v whole meal flour slurry (in 15% HNO3 containing 10% H2O2) were spiked with different amounts of surfactant from 0 to 0.5% v/v and homogenized by ultrasonication for 15 min before analysis. In the presence of Triton X-100, the sensitivity for Cr decreased gradually as the concentration of the surfactant
was increased (e.g. the signal decreased by about 35% in the presence of 0.1% Triton X-100). On the other hand, the surfactant made no difference to the Ni and Co determinations. The precision, as repeatability, of Cr measurements (after homogenization in an ultrasonic bath for 15 min) was similar with and without Triton X-100 (RSD 9.8 and 12.8%, respectively). Such low precision can be ascribed to rapid deposition of the slurry in the bottom of the autosampler cup; probably, a portion of the Cr present remained in the solid phase and required introducing the solid particles into the platform and hence homogenizing slurry. The repeatability can be raised to 7.5% with mechanical mixing of the autosampler cup contents between measurements. The slurry concentration is a critical variable in the proposed method owing to the differential concentration levels of these elements in flour samples and their sensitivities (e.g. Ni is present in −1 these matrices at the mg g level, whereas Co and −1 Cr occur at the ng g level). The effect of variable slurry concentrations on sensitivity and precision was investigated by using the spiked SRM sample. Based on reported facts [29], concentrations above 5% may result in inefficient
1058
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
Fig. 3. Effect of slurry concentration (% w/v) for a SRM wholemeal flour spiked with cobalt on the integrated absorbance signal in the determination of Co, Cr, and Ni by slurry sampling. Solid line, working range. Injected slurry volume, 20 ml (10 ml for cobalt).
pipetting of the slurry aliquot; the problem worsens with several matrices as a result of particles settling too early, which affects both the accuracy and the precision. Fig. 3 shows the influence of the slurry concentration in the whole meal flour sample spiked with cobalt on the integrated absorbance for Co, Cr, and Ni. A linear response was obtained for Co throughout the range studied (1–6%); on the other hand, this variable was critical for Ni and Cr. The wide range for Co can be ascribed to the fact that it was spiked to the whole meal flour, so it was more readily extracted from the liquid phase than were Ni and Cr. However, the results obtained for the three elements in slurries of vegetables and SRM materials to which no Co was added were similar [28]. As expected, the precision suffered when using highly diluted slurries because only a small number of particles was sampled. Increased slurry concentrations led to more marked matrix effects and to more pronounced deterioration of sensitivity for Ni and Cr. From the experiments described above it can be concluded that the determination of Ni in this matrix requires a higher HNO3 concentration (as
medium composition) than do Co and Cr, probably because the determination of the former is more favorable in the liquid phase than in the solid phase. The determination of Cr is more markedly affected by the slurry homogeneity than by the HNO3 concentration, probably because Cr can be determined in the solid particle and is easily released with no occlusion of the solid matrix. In order to sequentially determine the three elements in the same sample, the optimum conditions for Ni were selected. It should be noted that the main problem encountered in the atomization of biological slurries is that carbonaceous residues build up inside the platform owing to incomplete ashing of the organic matrix. In this respect, the HNO3 –H2O2 mixture acts as an oxidant modifier [19,28] that allows one to dispense with conventional chemical modifiers (e.g. Mg and Pd). Therefore, sample slurries were prepared at concentrations between 1 and 3% w/v in a 15% HNO3 –10% H2O2 medium, and homogenized by ultrasonication for 15 min; in this medium the slurry samples remained undigested, which was visually observed by the white color of the suspension. Cr require mechanical mixing of the cup
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
contents between measurements. Under these conditions, the precision, as RSD, was ca. 5% for the three elements. The blank signal (15% HNO3 – 10% H2O2 solution) was 0.006 A s.
Table 2 Analysis of wheat and legume flours by direct slurry samplinga Sample Element
3.3. Analytical features The analytical figures of merit of the proposed method were established by using aqueous standards (0.2% HNO3) and the furnace program shown in Table 1. The sensitivities (expressed as average slopes, n =3, of the calibration graphs) were 0.2490.02, 0.51 9 0.03, and 1.359 0.05 A s −1 ng for Ni, Co, and Cr, respectively. In order to compare the slopes of the calibration and standard addition graphs for slurries of SRM whole meal flour spiked with Co, the t-test was applied. No significant differences at a confidence level of 95% between the slopes of the additions and calibration graphs for the three elements were found, which suggests that both calibrations were statistically similar. The absence of significant matrix effects affords direct calibration with aqueous standards when using sample blank. The detection limits (calculated as three times the standard deviation of the signals obtained from 15 sample blanks) were 44, 30, and 23 ng −1 g (for a 3% slurry sample) for Ni, Co, and Cr, respectively. The characteristic masses, based on integrated absorbances (amounts, in picograms, providing a signal of 0.0044 A s), were 21, 9, and 4 for Ni, Co, and Cr, respectively.
3.4. Analysis of slurried flour samples Only the concentrations of Ni and Cr in the SRM whole meal flour were stated by the supplier. In addition, the rather low cobalt concentra−1 tion in this sample (ca. 8 ng g ) entailed preparing highly concentrated slurries (ca. 25% w/v) that gave high background signals and poor analytical results, and made pipetting cumbersome. Consequently, as stated under Section 2, the reference material was spiked with Co at an −1 elevated level (viz. 500 ng g ). The reliability of the proposed slurry method was checked by analyzing five individual SRM whole meal flour samples in triplicate, as well as a wheat flour 1
1059
Whole meal flour (SRM No. 189)
Ni
Direct slurry (n = 5) 355 9 15
(380)b 65 94
Co Coc
62 94 (57–76)b 8 91 —
Ni Cr Co
200 910 70 9 4 209 1
205 915 62 9 5 —
Ni 160 9 10 280 920 230 915 255 920 208 915 210 915 150 910 450 9 30
Cr 30 9 2 46 9 2 36 9 2 43 9 5 35 9 3 40 9 2 34 9 4 72 9 6
Cr
Wheat flour 1
Wheat flour Wheat flour Wheat flour Wheat flour Wheat flour Wheat flour Corn flour Chick-pea flour
Dry ashing (n =3) 347 915
2 3 4 5 6 7
— 500 920
Concentrations expressed in ng g−1 9S. Non-certified value. c SRM No. 189 spiked with 500 ng of Co per g a
b
sample, and comparing the results with those for samples digested by dry ashing. As can be seen in Table 2, the results obtained for Ni and Cr with the slurry procedure and by dry ashing in both samples were quite consistent; also the Co recovery from the spiked SRM sample was near 100%. Therefore, the proposed slurry method is fairly accurate and can be used to determine these elements in similar samples. Table 2 summarizes the results for Ni and Cr in slurried flours (3% w/v); the concentration of Co in all samples was below the detection limit. Aqueous standards for calibration graphs and sample blanks were employed. The metal levels found in all samples were lower than their tolerated limits in foods. It should be noted that the highest Ni and Cr levels were obtained in the chick-pea flour. Flours containing Ni and Cr at concentrations below 200 and 40 ng −1 g , respectively, were also determined, using the
1060
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
standard additions method, which provided concentrations similar to those of the direct method.
References [1] Council Directive of June 12, On the protection of the environment, and in particular of the soil, when sewage sludge is used in agriculture, Official J. EC L 181/6 – 12, Brussels, 1986. [2] A. Kabata-Pendias, H. Pendias, Trace Elements in Soils and Plants, 2nd ed., CRC Press, Boca Raton, FL, 1992. [3] C.J. Park, J.K. Suh, J. Anal. At. Spectrom. 12 (1997) 573. [4] Z. Wu, Z. Wu, Fenxi-Huaxue 16 (1988) 710. [5] O.K. Borggaard, H.E.M. Christensen, S.P. Lund, Analyst 109 (1984) 1179. [6] W.J. Blanchflower, A. Cannavan, D.G. Kennedy, Analyst 115 (1990) 1323. [7] R. Chakraborty, A.K. Das, Anal. Lett. 30 (1997) 283. [8] M. Plessi, A. Monzani, J. Assoc. Off. Anal. Chem. 73 (1990) 798. [9] A. Rosopulo, K.H. Grobecker, U. Kurfuerst, Fresenius J. Anal. Chem. 319 (1984) 540. [10] Y. Zhang, J.H. Pan, Guangpuxue Yu Guangpu Fenxi 17 (1997) 94. [11] L. Jorhem, B. Sundstrom, J. Food Compos. Anal. 6 (1993) 223.
.
[12] D.V. Brady, J.G. Montalvo, J. Jung, R.A. Curran, At. Absorpt. Newsl. 13 (1974) 118. [13] V.I. Slaveykova, M. Hoenig, Analyst 122 (1997) 337. [14] N.J. Miller-Ihli, Spectrochim. Acta 50 (Part B) (1995) 477. [15] N.J. Miller-Ihli, J. Anal. At. Spectrom. 12 (1997) 205. [16] Atomic Spectrometry Update, J. Anal. At. Spectrom. 12 (1997) 119R and 327R. [17] M.A.Z. Arruda, M. Gallego, M. Valca´rcel, Quı´m. Anal. 14 (1995) 17. [18] P. Vin˜as, N. Campillo, I. Lo´pez-Garcı´a, M. Herna´ndezCo´rdoba, At. Spectrosc. 16 (1995) 86. [19] P. Vin˜as, N. Campillo, I. Lo´pez-Garcı´a, M. Herna´ndezCo´rdoba, Talanta 42 (1995) 527. [20] T. Mochizuki, A. Sakashita, H. Iwata, Y. Ishibashi, N. Gunji, Fresenius J. Anal. Chem. 339 (1991) 889. [21] N.J. Miller-Ihli, J. Anal. At. Spectrom. 3 (1988) 73. [22] J.A. Holcombe, V. Majidi, J. Anal. At. Spectrom. 4 (1989) 423. [23] C. Bendicho, M.T.C. de Loos-Vollebregt, J. Anal. At. Spectrom. 6 (1991) 353. [24] N.J. Miller-Ihli, Fresenius J. Anal. Chem. 345 (1993) 482. [25] N.J. Miller-Ihli, J. Anal. At. Spectrom. 9 (1994) 1129. [26] N.J. Miller-Ihli, At. Spectrosc. 13 (1992) 1. [27] I. Lo´pez-Garcı´a, M. Sa´nchez-Merlos, M. Herna´ndez-Co´rdoba, J. Anal. At. Spectrom. 12 (1997) 777. [28] A. Carlosena, M. Gallego, M. Valca´rcel, J. Anal. At. Spectrom. 12 (1997) 479. [29] S. Lynch, D. Littlejohn, Talanta 37 (1990) 825.
Determination of nickel, chromium and cobalt in wheat flour using slurry sampling electrothermal atomic absorption spectrometry Mar Gonza´lez, Mercedes Gallego, Miguel Valca´rcel * Department of Analytical Chemistry, Faculty of Sciences, Uni6ersity of Co´rdoba, E-14004 Co´rdoba, Spain Received 3 June 1998; received in revised form 29 September 1998; accepted 2 October 1998
Abstract The slurry technique was applied to the determination of Ni, Cr and Co in wheat flour by electrothermal atomic absorption spectrometry (ETAAS). The influence of the graphite furnace temperature programme was optimized. Optimum sensitivity was obtained by using a mixture of 15% HNO3 – 10% H2O2 as suspended medium for a 3% w/v slurry in the determination of Ni; lower concentrations of HNO3 were necessary for the determination of Co and Cr (viz. 5 and 10%). The precision of direct analyses of the slurries was improved by using mechanical agitation between measurements; thus, the RSD of the measurements was ca. 5% for repeatability. The direct slurry sampling (SS) technique is suitable for the determination of Ni and Cr in wheat flour samples at levels of 150 – 450 and 30 – 72 ng g − 1, respectively, as it provides results similar to those obtained by ashing the sample. However, the typically low level of Co in these samples precluded its determination by the proposed method (the study was made in an SRM spiked wholemeal flour), at least in those samples that were contaminated with elevated concentrations of the metal (viz. more than 90 ng of Co per g of flour). The method provides a relative standard deviation of 6, 8, and 4% for Ni, Cr, and Co, respectively. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Electrothermal atomic absorption spectrometry; Slurry sampling; Wheat flour; Nickel; Chromium; Cobalt
1. Introduction Cereal flours are staple foods in most countries. Thus, wheat flour is consumed every day in bread, cake, sauces, etc. Analyses for trace element in flours are therefore important from both a nutritional and a toxicological point of view. Soil is the * Corresponding author. Tel.: + 34-57-218614; fax: +3457-218606; e-mail: [email protected].
main vehicle by which heavy metals enter plants; in response, a European Union directive (86/278/ CEE) has been issued to control the maximum allowable contents of some metals in agricultural sewage sludge [1]. The average daily intake in food for metals is well documented; for example Ni, Cr, and Co occur at very low concentrations in cereals (viz. 100–400, 20–50 and 10–20 ng g − 1, for Ni, Cr, and Co, respectively) [2]. While Cr, Co, and Ni have been shown to be essential
0039-9140/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 9 1 4 0 ( 9 8 ) 0 0 3 1 7 - 8
1052
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
for humans, the necessary levels of Ni have not been quantified. Trace metals can be determined by using various techniques. In recent times, the ICP-MS [3] technique has gained momentum for this purpose. However, ETAAS has for some time been at most laboratories and used in many determinations of Ni, Cr, and Co. The low concentrations of Co normally found in cereals usually require preconcentration prior to the determination proper. Co in feed grains has been determined in this way following digestion [4], as has in cereals after decomposition of organic matter and extraction into a 2-nitroso-1-naphthol solution in xylene [5], or in heptan-2-one [6]. Ni in various matrices including rice flours has also been determined by ETAAS, following microwave-assisted digestion [7]; Cr in grain and cereal products require prior wet digestion [8]. Cr and Ni, in addition to other metals, were determined in wheat flour by using a graphite boat with direct Zeeman-AAS [9]. Co and Ni were determined in cereals after digestion with the HNO3/HClO4 mixture with recoveries from 87 to 104% [10]. The contents of Ni, Cr, and Co in foods including cereals on the Swedish market between 1983 and 1990 were studied by the Nordic Committee on Food Analysis; the official methodology selected for determination of different elements was dry ashing followed by ETAAS [11]. Slurry sampling (SS) ETAAS methodology, originally developed by Brady et al. [12] is by now well established and widely used in the determination of trace elements in food samples. The most attractive advantages of SS over dry or wet ashing can be summarized as follows: (i) it reduces sample pretreatment and analysis times; (ii) it minimizes contamination/loss risks; (iii) it uses a conventional sample introduction system; (iv) it ensures appropriate calibration with aqueous standards; and (v) it provides acceptable accuracy and precision. However, this technique has some disadvantages that arise essentially from non-homogeneous distribution of the trace elements in the slurry as well as differences in the chemical species under which they
are present [13]. The benefits of SS were demonstrated in an international collaborative study involving 25 laboratories that was intended to assess the state of the art in the technique. Preliminary results [14] suggested that SS is mature enough for routine analyses; at a later stage, the usefulness of ultrasonic SS was evaluated [15]. The results showed that extracting analyte into the liquid phase of the slurry is not a prerequisite for accurate slurry analyses. Slurry sampling for electrothermal atomization is a very active, widely documented area applications of which were reviewed [16]. A comprehensive review of SS for foods in atomic spectrometry has also been published [17]. Reported applications involving cereals are scant relative to other foods. Also, Co and Ni have been determined less frequently than Cr in this way. Vin˜as et al. used their experience in the use of slurry procedures to determine Co and Ni in vegetables and legumes [18], and Cr in vegetables [19], following treatment of dried samples with the ethanol–H2O2 –HNO3 mixture; in all instances, calibration was against aqueous standards. Direct slurry sampling was used to determine Ni and Co in rice by ICP-MS [20], and Cr and Ni in spinach leaves by simultaneous multi-element atomic absorption spectrometer with continuum source (SIMAAC) [21]. In the latter application, the slurry solution (5– 10 mg of sample in 5 ml of 5% HNO3) was homogenized by immersing an ultrasonic probe in the autosampler cup, which avoided settling of particles. In this work, the potential of slurry sampling for the determination of Ni, Cr, and Co in wheat flour by ETAAS was explored. The influence of various parameters such as the drying and pyrolysis time, pyrolysis and atomization temperature, and presence of modifier on the atomic signal was studied. Efforts were aimed at using the advantages of the HNO3 –H2O2 mixture as the medium for sample preparation in order to transfer the analytes to the aqueous phase. The proposed method was applied to various samples with acceptable recoveries for Ni and Cr.
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
1053
Table 1 Instrumental parameters and optimized furnace conditions for the determinations of Ni, Cr and Co in wheat floura
Lamp current/mA Wavelength/nm Bandpass/nm
Step Dry 1 Dry 2 Pyrolysis Atomize Clean a
Ni
Cr
Co
25 232.0 0.2
10 357.9 0.7
30 240.7 0.2
Temperature (°C)
Ramp (s)
Hold (s)
Temperature
Ramp Hold
Temperature
Ramp Hold
100 300 1400 2600 2650
5 10 20 0 1
15 30 30 0 1
100 300 1700 2600 2650
5 10 20 0 1
100 300 1400 2600 2650
5 10 20 0 1
15 30 30 6 3
15 30 30 6 3
A stream of argon at 300 ml min−1 was used (the flow was stopped during the atomization step); injected volume, 20 ml
2. Experimental
2.1. Apparatus Slurry samples were analysed by using a model 1100-B atomic absorption spectrometer from Perkin-Elmer (U8 berlingen, Germany) equipped with a deuterium-lamp background corrector, an HGA-700 graphite furnace, and an AS-70 autosampler, and interfaced to an Epson FX-850 printer. Analyses were carried out by using platforms inserted into pyrolytically coated graphite tubes (Perkin-Elmer) and measurements (integrated absorbance peak areas) were made by using single-element hollow cathode lamps (Perkin-Elmer). Argon was used as sheeting gas for the furnace in all cases. The operational parameters recommended by the manufacturer for Ni, Cr, and Co are listed in Table 1. Slurries were homogenized in an ultrasonic bath (Bandelin, Tk52, Berlin, Germany) or a vortex mixer (Heidolph, Kelheim, Germany).
2.2. Reagents Working metal standards were prepared daily −1 from a 1000 mg l stock metal solution (Panreac, Barcelona, Spain) by diluting appropriate aliquots with 0.2% HNO3. A 10.0 g l − 1 solution of magnesium nitrate (Merck, Darmstadt, Germany) was
used as matrix modifier. Triton X-100 (Serva Feinbiochemica, Heidelberg, Germany) was tested as stabilizing agent.
2.3. Certified reference material SRM wholemeal flour no. 189, with reference contents (non-certified values) for Ni and Cr obtained from the European Commission (Belgium) was dried to constant mass in an oven at 103°C as per the supplier’s recommendations, and used for method validation. Because the Co concentration in this reference material was too low for its determination by the slurry sampling technique, the material was spiked with this metal from an aqueous salt solution. For this purpose, 40 ml of −1 a solution containing 250 ng ml Co was added to 20.0 g of the SRM. The slurry thus obtained was dried at room temperature for 2 weeks in a closed fume hood to avoid contamination and then to constant mass in an oven at 103°C. The wholemeal flour spiked with 500 ng of cobalt per g of sample was employed to optimize the cobalt determination in this matrix.
2.4. Sample preparation Flour samples were prepared at a pilot plant from wheat produce in different Spanish locations. An amount of 0.5–1 kg of wheat contain-
1054
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
ing ca. 15% w/w of water was ground in a metal rolling mill to obtain 60% of white flour and 40% of byproducts. The flour was screened with a 130 mm sieve. All other samples were purchased at a local supermarket.
2.5. Procedures In order to avoid contamination, all polytetrafluoroethylene (PTFE) materials, pipettes, and calibrated flasks were immersed in freshly made 10% HNO3 for 24 h and then rinsed thoroughly with high purity water (Milli-Q Water System, Millipore, Madrid, Spain) before use. For slurry analyses, reference and sample materials were accurately weighed (ca. 150 mg) into PTFE tubes and supplied with 5 ml of 15% HNO3 containing 10% v/v H2O2. Slurries (3% w/v, i.e. 3 g into 100 ml) were homogenized by agitating for 15 min in an ultrasonic immediately before each analysis. Aliquots of 2 ml were placed in the 2 ml polyethylene vials of the autosampler; between measurements, samples were mixed in a vortex shaker for 5 s in order to ensure reproducible results using autosampler cups provided with covers. Five sample replicates were analysed in each case. The sample blanks contained the same concentration of nitric acid and hydrogen peroxide as the slurry samples. Calibration graphs (spanning −1 −1 the range 0– 20 mg l for Cr, 0 – 40 mg l for Co, −1 and 0–50 mg l for Ni) were obtained from variable volumes of standard solutions containing −1 30, 40, and 50 mg l Cr, Co, and Ni, respectively, that were mixed in the graphite tube with appropriate amounts of 0.2% HNO3 to a volume of 20 ml; 20 ml of 0.2% HNO3 was used as blank. For mineralization of the flours, 3.0 g of material was accurately weighed into a platinum crucible and carbonized in a burner at a low temperature for about 3 h; then, the black residue was supplied with several drops of concentrated H2O2 and ashed in a muffle furnace at 600–650 °C for 2 h. The completely ashed sample obtained was dissolved in 10 ml of 0.5% HNO3; three sample replicates and four injections per replicate were analysed. Cr and Co were measured directly, whereas Ni required ten-fold dilution in 0.2% HNO3.
3. Results and discussion The different factors that influence the performance of slurry ETAAS have been comprehensively examined by several authors [14–17,22–25]. Slurry introduction has been found to pose some problems related to concentration and particle size, which influence stability, deposition and atomization efficiency; these in turn may affect the accuracy and precision of analyses. Ensuring accurate results and good reproducibility in this context entails using homogeneous slurries; mechanical mixing devices, gas bubbling, and ultrasonic agitation has proved useful for this purpose [23]. Thus, Miller-Ihli [26] designed an ultrasonic, pneumatically movable slurry sampler and, more recently, Lo´pez-Garcı´a et al. [27] based on previous experiments of other authors [23], developed an efficient slurry sampling device that uses argon bubbles to homogenize slurries in the autosampler cups without the need to alter the instrument operation.
3.1. Furnace temperature programs and chemical modifiers The temperature programs and the effect of using magnesium nitrate as modifier on both standards and slurries were carefully examined. The SRM whole meal flour spiked with 0.5 mg of cobalt per g was the sample used in this study. In all instances, slurries of 3% w/v were prepared in 15% HNO3 containing 10% H2O2. A 15 ml volume of standard/slurry and 5 ml of chemical modifier −1 (10 g l magnesium nitrate) were injected into the pyrolytic graphite tube with platform. Two drying steps were required to ensure mild, totally dry conditions for the sample, and no splattering; the slurry sample required higher temperatures (100 and 300°C) than the standard (100 and 140°C).
3.1.1. Nickel The pyrolysis and atomization temperatures were tested, in the absence of chemical modifier, over the ranges 500–1900°C, and 1900–2650°C, respectively. At a constant atomization temperature of 2600°C, the pyrolysis temperature did not
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060 −1
affect the signal for the standard (10 mg l Ni) up to 1500°C; on the other hand, the optimum pyrolysis temperature for the slurry sample was 1400°C. A similar behavior was observed with magnesium nitrate as modifier. Peak areas increased with increasing atomization temperature up to 2600°C, both for the standard and for the sample. As can be seen in Fig. 1A, the background signal for the slurry sample decreased at high atomization temperatures. On the other hand, the use of magnesium nitrate at the optimum atomization temperature (2600°C) increased of background signal for the slurry sample (see Fig. 1A). Although this modifier is recommended for the determination of nickel in different matrices, its use in this case stabilized a higher proportion to the concomitants. Based on these considerations, the pyrolysis and atomization temperature were fixed at 1400°C and 2600°C, respectively, and no chemical modifier was employed for this element.
3.1.2. Cobalt The pyrolysis temperature had no effect on the determination of the standard (10 mg l − 1 Co) neither on the spiked slurry whole meal over the range 800–1400°C. As no cobalt was lost up to 1400°C, this temperature was fixed while the atomization temperature was changed. The best results were obtained at an atomization temperature of 2600°C. The use of magnesium nitrate as modifier provided no advantages for the whole meal slurry sample (the background signal was ca. 0.050 A s with and without modifier); on the other hand, the modifier increased the background signal for the standard solution two-fold (which was 0.060 and 0.120 A s in its absence and presence, respectively). In order to facilitate reliable deuterium background correction, use of the modifier was discarded. 3.1.3. Chromium The maximum pyrolysis temperature was the same with and without modifier, 1700°C, for both −1 the standard solution (5 mg l Cr) and the slurry (3% whole meal in the HNO3 – H2O2 mixture). Moreover, the addition of 50 mg of magnesium nitrate increased slightly both the atomic absorp-
1055
tion and background signals, so its presence it provided no advantage. As can be seen from Fig. 1B, better separation of atomic and background signals was obtained in the absence of this modifier. On the other hand the addition of 5 ml of palladium nitrate (1.0 g l − 1) to the slurry sample provided results similar to those obtained with magnesium nitrate for the three elements studied. Therefore, the temperature programs listed in Table 1 were employed for standards and slurries (prepared in the HNO3 –H2O2 mixture), with no chemical modifier.
3.2. Optimization of the slurry preparation The factors potentially influencing the accuracy and precision of the analytical results were studied, namely suspension medium, the homogenization and stability of the slurry, and slurry concentration (% w/v) were examined in order to optimize preparation of the slurries. The effect of the slurry particle size was not studied as all flours were finely powdered (less than 130 mm), which is similar to the SRM whole meal flour (less than 125 mm). The slurries of SRM whole meal flour spiked with cobalt, at 3% w/v, were prepared at variable HNO3 concentrations (0–20% v/v) and contained 10% H2O2. Agitation for 5 min in a vortex mixer was needed to homogenize the slurry before it was transferred into the autosampler cup and 20 ml was injected into the graphite furnace. As can be seen in Fig. 2, the nitric acid concentration had a marked effect on the Ni signal, and somewhat lesser effects on the Cr and Co signals. A concentration of nitric acid of 5%, 10%, and 20% for Co, Cr, and Ni, respectively, was needed to favor the extraction of the analytes into solution. The determination of Cr and Co required no addition of H2O2 to the slurry sample when prepared in 10% HNO3; however, that of Ni required 10% H2O2 and high nitric acid concentration (20%). The higher nitric acid concentration required for the determination of Ni can be explained probably because Co and Cr are extracted more efficiently into the liquid phase than is Ni or because these elements are more efficiently determined in the
1056
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
Fig. 1. Atomic absorption profile and background lines for Ni (A) and Cr (B) in SRM wholemeal slurry samples (3% w/v in 15% HNO3 containing 10% H2O2). A, Ni determination (pyrolysis temperature, 1400°C): (a) and (b) in the absence of chemical modifier, at an atomization temperature of 2250°C and 2600°C (background signals, 0.154 and 0.071 A s, respectively); and (c) in the presence of 50 mg of Mg(NO3)2, at a atomization temperature of 2600°C (background signal, 0.122 A s). B, Cr determination with Mg(NO3)2 (a) and without chemical modifier (b) (pyrolysis and atomization temperature, 1700°C and 2600°C, respectively). Solid and dashed lines represent atomic and background signals, respectively. Injected slurry volume, 15 and 20 ml for Ni and Cr, respectively.
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
1057
Fig. 2. Effect of the concentration of the nitric acid as the medium used to prepare the slurry of SRM wholemeal flour spiked with cobalt. All slurries also contained 10% H2O2.
solid phase than is Ni. In order to decrease the high nitric acid concentration (20% v/v) required by the Ni determination, the effectiveness of ultrasonic and mechanical (vortex) agitation for slurry homogenization was studied. For this purpose, various samples of 3% w/v SRM whole meal spiked flour slurry were prepared in a medium containing 15% HNO3 plus 10% H2O2 for all of the elements. The slurries were shaken in a vortex mixer or an ultrasonic bath for 1 – 30 min. Both homogenization devices required at least 15 min to ensure stabilization of integrated absorbance average values (n =3 at each time). However, ultrasonic mixing provided higher precision (RSD 6.0, 3.5, and 12.6% for Ni, Co, and Cr, respectively) than mechanical stirring (RSD 7.5, 6.5, and 19.0%, respectively). As Cr detection was scarcely precise, Triton X-100 was tested as a stabilizing agent. Thus, several samples of 3% w/v whole meal flour slurry (in 15% HNO3 containing 10% H2O2) were spiked with different amounts of surfactant from 0 to 0.5% v/v and homogenized by ultrasonication for 15 min before analysis. In the presence of Triton X-100, the sensitivity for Cr decreased gradually as the concentration of the surfactant
was increased (e.g. the signal decreased by about 35% in the presence of 0.1% Triton X-100). On the other hand, the surfactant made no difference to the Ni and Co determinations. The precision, as repeatability, of Cr measurements (after homogenization in an ultrasonic bath for 15 min) was similar with and without Triton X-100 (RSD 9.8 and 12.8%, respectively). Such low precision can be ascribed to rapid deposition of the slurry in the bottom of the autosampler cup; probably, a portion of the Cr present remained in the solid phase and required introducing the solid particles into the platform and hence homogenizing slurry. The repeatability can be raised to 7.5% with mechanical mixing of the autosampler cup contents between measurements. The slurry concentration is a critical variable in the proposed method owing to the differential concentration levels of these elements in flour samples and their sensitivities (e.g. Ni is present in −1 these matrices at the mg g level, whereas Co and −1 Cr occur at the ng g level). The effect of variable slurry concentrations on sensitivity and precision was investigated by using the spiked SRM sample. Based on reported facts [29], concentrations above 5% may result in inefficient
1058
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
Fig. 3. Effect of slurry concentration (% w/v) for a SRM wholemeal flour spiked with cobalt on the integrated absorbance signal in the determination of Co, Cr, and Ni by slurry sampling. Solid line, working range. Injected slurry volume, 20 ml (10 ml for cobalt).
pipetting of the slurry aliquot; the problem worsens with several matrices as a result of particles settling too early, which affects both the accuracy and the precision. Fig. 3 shows the influence of the slurry concentration in the whole meal flour sample spiked with cobalt on the integrated absorbance for Co, Cr, and Ni. A linear response was obtained for Co throughout the range studied (1–6%); on the other hand, this variable was critical for Ni and Cr. The wide range for Co can be ascribed to the fact that it was spiked to the whole meal flour, so it was more readily extracted from the liquid phase than were Ni and Cr. However, the results obtained for the three elements in slurries of vegetables and SRM materials to which no Co was added were similar [28]. As expected, the precision suffered when using highly diluted slurries because only a small number of particles was sampled. Increased slurry concentrations led to more marked matrix effects and to more pronounced deterioration of sensitivity for Ni and Cr. From the experiments described above it can be concluded that the determination of Ni in this matrix requires a higher HNO3 concentration (as
medium composition) than do Co and Cr, probably because the determination of the former is more favorable in the liquid phase than in the solid phase. The determination of Cr is more markedly affected by the slurry homogeneity than by the HNO3 concentration, probably because Cr can be determined in the solid particle and is easily released with no occlusion of the solid matrix. In order to sequentially determine the three elements in the same sample, the optimum conditions for Ni were selected. It should be noted that the main problem encountered in the atomization of biological slurries is that carbonaceous residues build up inside the platform owing to incomplete ashing of the organic matrix. In this respect, the HNO3 –H2O2 mixture acts as an oxidant modifier [19,28] that allows one to dispense with conventional chemical modifiers (e.g. Mg and Pd). Therefore, sample slurries were prepared at concentrations between 1 and 3% w/v in a 15% HNO3 –10% H2O2 medium, and homogenized by ultrasonication for 15 min; in this medium the slurry samples remained undigested, which was visually observed by the white color of the suspension. Cr require mechanical mixing of the cup
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
contents between measurements. Under these conditions, the precision, as RSD, was ca. 5% for the three elements. The blank signal (15% HNO3 – 10% H2O2 solution) was 0.006 A s.
Table 2 Analysis of wheat and legume flours by direct slurry samplinga Sample Element
3.3. Analytical features The analytical figures of merit of the proposed method were established by using aqueous standards (0.2% HNO3) and the furnace program shown in Table 1. The sensitivities (expressed as average slopes, n =3, of the calibration graphs) were 0.2490.02, 0.51 9 0.03, and 1.359 0.05 A s −1 ng for Ni, Co, and Cr, respectively. In order to compare the slopes of the calibration and standard addition graphs for slurries of SRM whole meal flour spiked with Co, the t-test was applied. No significant differences at a confidence level of 95% between the slopes of the additions and calibration graphs for the three elements were found, which suggests that both calibrations were statistically similar. The absence of significant matrix effects affords direct calibration with aqueous standards when using sample blank. The detection limits (calculated as three times the standard deviation of the signals obtained from 15 sample blanks) were 44, 30, and 23 ng −1 g (for a 3% slurry sample) for Ni, Co, and Cr, respectively. The characteristic masses, based on integrated absorbances (amounts, in picograms, providing a signal of 0.0044 A s), were 21, 9, and 4 for Ni, Co, and Cr, respectively.
3.4. Analysis of slurried flour samples Only the concentrations of Ni and Cr in the SRM whole meal flour were stated by the supplier. In addition, the rather low cobalt concentra−1 tion in this sample (ca. 8 ng g ) entailed preparing highly concentrated slurries (ca. 25% w/v) that gave high background signals and poor analytical results, and made pipetting cumbersome. Consequently, as stated under Section 2, the reference material was spiked with Co at an −1 elevated level (viz. 500 ng g ). The reliability of the proposed slurry method was checked by analyzing five individual SRM whole meal flour samples in triplicate, as well as a wheat flour 1
1059
Whole meal flour (SRM No. 189)
Ni
Direct slurry (n = 5) 355 9 15
(380)b 65 94
Co Coc
62 94 (57–76)b 8 91 —
Ni Cr Co
200 910 70 9 4 209 1
205 915 62 9 5 —
Ni 160 9 10 280 920 230 915 255 920 208 915 210 915 150 910 450 9 30
Cr 30 9 2 46 9 2 36 9 2 43 9 5 35 9 3 40 9 2 34 9 4 72 9 6
Cr
Wheat flour 1
Wheat flour Wheat flour Wheat flour Wheat flour Wheat flour Wheat flour Corn flour Chick-pea flour
Dry ashing (n =3) 347 915
2 3 4 5 6 7
— 500 920
Concentrations expressed in ng g−1 9S. Non-certified value. c SRM No. 189 spiked with 500 ng of Co per g a
b
sample, and comparing the results with those for samples digested by dry ashing. As can be seen in Table 2, the results obtained for Ni and Cr with the slurry procedure and by dry ashing in both samples were quite consistent; also the Co recovery from the spiked SRM sample was near 100%. Therefore, the proposed slurry method is fairly accurate and can be used to determine these elements in similar samples. Table 2 summarizes the results for Ni and Cr in slurried flours (3% w/v); the concentration of Co in all samples was below the detection limit. Aqueous standards for calibration graphs and sample blanks were employed. The metal levels found in all samples were lower than their tolerated limits in foods. It should be noted that the highest Ni and Cr levels were obtained in the chick-pea flour. Flours containing Ni and Cr at concentrations below 200 and 40 ng −1 g , respectively, were also determined, using the
1060
M. Gonza´lez et al. / Talanta 48 (1999) 1051–1060
standard additions method, which provided concentrations similar to those of the direct method.
References [1] Council Directive of June 12, On the protection of the environment, and in particular of the soil, when sewage sludge is used in agriculture, Official J. EC L 181/6 – 12, Brussels, 1986. [2] A. Kabata-Pendias, H. Pendias, Trace Elements in Soils and Plants, 2nd ed., CRC Press, Boca Raton, FL, 1992. [3] C.J. Park, J.K. Suh, J. Anal. At. Spectrom. 12 (1997) 573. [4] Z. Wu, Z. Wu, Fenxi-Huaxue 16 (1988) 710. [5] O.K. Borggaard, H.E.M. Christensen, S.P. Lund, Analyst 109 (1984) 1179. [6] W.J. Blanchflower, A. Cannavan, D.G. Kennedy, Analyst 115 (1990) 1323. [7] R. Chakraborty, A.K. Das, Anal. Lett. 30 (1997) 283. [8] M. Plessi, A. Monzani, J. Assoc. Off. Anal. Chem. 73 (1990) 798. [9] A. Rosopulo, K.H. Grobecker, U. Kurfuerst, Fresenius J. Anal. Chem. 319 (1984) 540. [10] Y. Zhang, J.H. Pan, Guangpuxue Yu Guangpu Fenxi 17 (1997) 94. [11] L. Jorhem, B. Sundstrom, J. Food Compos. Anal. 6 (1993) 223.
.
[12] D.V. Brady, J.G. Montalvo, J. Jung, R.A. Curran, At. Absorpt. Newsl. 13 (1974) 118. [13] V.I. Slaveykova, M. Hoenig, Analyst 122 (1997) 337. [14] N.J. Miller-Ihli, Spectrochim. Acta 50 (Part B) (1995) 477. [15] N.J. Miller-Ihli, J. Anal. At. Spectrom. 12 (1997) 205. [16] Atomic Spectrometry Update, J. Anal. At. Spectrom. 12 (1997) 119R and 327R. [17] M.A.Z. Arruda, M. Gallego, M. Valca´rcel, Quı´m. Anal. 14 (1995) 17. [18] P. Vin˜as, N. Campillo, I. Lo´pez-Garcı´a, M. Herna´ndezCo´rdoba, At. Spectrosc. 16 (1995) 86. [19] P. Vin˜as, N. Campillo, I. Lo´pez-Garcı´a, M. Herna´ndezCo´rdoba, Talanta 42 (1995) 527. [20] T. Mochizuki, A. Sakashita, H. Iwata, Y. Ishibashi, N. Gunji, Fresenius J. Anal. Chem. 339 (1991) 889. [21] N.J. Miller-Ihli, J. Anal. At. Spectrom. 3 (1988) 73. [22] J.A. Holcombe, V. Majidi, J. Anal. At. Spectrom. 4 (1989) 423. [23] C. Bendicho, M.T.C. de Loos-Vollebregt, J. Anal. At. Spectrom. 6 (1991) 353. [24] N.J. Miller-Ihli, Fresenius J. Anal. Chem. 345 (1993) 482. [25] N.J. Miller-Ihli, J. Anal. At. Spectrom. 9 (1994) 1129. [26] N.J. Miller-Ihli, At. Spectrosc. 13 (1992) 1. [27] I. Lo´pez-Garcı´a, M. Sa´nchez-Merlos, M. Herna´ndez-Co´rdoba, J. Anal. At. Spectrom. 12 (1997) 777. [28] A. Carlosena, M. Gallego, M. Valca´rcel, J. Anal. At. Spectrom. 12 (1997) 479. [29] S. Lynch, D. Littlejohn, Talanta 37 (1990) 825.