Determination of Iodine in Food-related Certified Reference Materials Using Wet Ashing and Detection by Inductively Coupled Plasma Mass Spectrometry ERIK H. LARSEN* AND MERETE B. LUDWIGSEN National Food Agency of Denmark Institute of Food Chemistry and Nutrition 19 Mørkhøj Bygade DK-2860 Søborg Denmark Iodine was accurately determined in food-related certified reference materials (CRMs) by wet ashing in closed steel bombs using a mixture of nitric acid and perchloric acid followed by measurement of iodine by ICP-MS at m/z 127. The iodine concentrations determined were 0.15–4.59 mg g-1 (dry mass). Potentially volatile iodine species such as hydrogen iodide were converted during the ashing procedure to nonvolatile species. Hereby memory problems in the ICP-MS due to volatile iodine were prevented.The concentric nebulizer which was used in combination with a low dead-volume cyclonic spray chamber improved the iodine signal intensity and reduced the wash-out time between samples compared with the standard Ryton double-pass spray chamber equipped with a cross flow nebulizer. Iodine contamination from the PTFE liners of the bombs led to blank values at an average of 9 ng per ashing. The limit of detection which was based on three standard deviations of the blank was 30 ng g-1 (dry mass). By adding 3% v/v methanol to the analyte solutions and increasing the plasma power to 1200 W the signal-tonoise ratio and hence the limit of detection was improved by 50% due to the signal enhancement effect by carbon of the incompletely ionized iodine.Keywords Iodine; food ; certified reference materials ; sample introduction ; inductively coupled plasma mass spectrometry; signal enhancement Iodine is an essential element to man and the beneficial effect of iodine-containing sea weed on goitre (swollen thyroid glands due to iodine deficiency) has been known since pre-historic times.1 In the state of iodine deficiency the gland produces too little of the iodine-containing thyroidea hormones. The main source to man of iodine is food and water and for adolescents and adults a recommended dietary intake has been estimated internationally as well as in the Nordic Countries2 at 150 mg d-1. In order to assess the dietary intake of iodine in a population and in population sub-groups at risk of a too low or too high iodine intake accurate data are needed regarding the iodine content of individual food items.Such data are particularly valuable bearing in mind the relatively narrow safety margin of a factor of 2–7 between the maximum tolerable and the essential iodine intake.3 A variety of analytical methods have been used for the determination of iodine in food. Most often a sample dissolution step (wet or dry ashing) has been employed prior to the detection step which often requires that iodine has been converted to and isolated in a particular chemical or physical form amenable to the final detection. Spectrophotometry was used for the determination of iodine based on an iodidecatalyzed reduction of cerium(IV) by arsenic(III ) in a variety of food items.4 NAA has been used after direct irradiation of the sample followed by combustion of the sample in oxygen and absorption of the liberated iodine,5 or by using preconcen- Journal of Analytical Atomic Spectrometry April 1997 Vol.12 (435–439) Wet Digestion The samples were digested in high-pressure steel bombs model DAE II with PTFE liners of 50 ml volume (Berghof Tu�bingen tration or chemical separation prior to detection of the gamma activity of the 131I formed during the activation process.6–8 A related method relies on the chemisorption of radioiodide on gold after decomposition of the sample in oxygen.9 Negative TIMS was used in conjunction with quantification by the isotope dilution technique after isolation of iodine as silver iodide10,11 or after extraction of free iodine into tetrachloromethane.12 Isotope dilution analysis was also used with laser resonance ionization MS after isolation of the analyte as silver iodide.13 A method that aimed at using more simple equipment was based on cathodic stripping voltammetry following oxidative wet ashing.14 Gas chromatography was used for the determination of iodine using derivatization with pentan-3-one following alkaline dry ashing.15 Atomic spectrometric methods used include ICP-AES of generated volatile iodine species from the iodide and iodate analytes7 and ICP-MS for the direct determination of iodine in plasma and in urine.16 Iodine was determined directly by ICP-MS in milk after the addition of base17,18 or after microwave-assisted wet destruction by base.19 Some of the above mentioned analytical techniques particularly those based on radioiodine measurements require highly specialized and therefore relatively inaccessible equipment or involve multi-step procedures which are prone to iodine losses or contamination.In contrast ICP-MS which is becoming an increasingly popular technique in food research and control allows the direct determination of iodine in solution and is a highly sensitive selective and largely interference-free detector for the monoisotopic iodine (127I). Some iodine species e.g. hydrogen iodide are gaseous at room temperature and therefore volatile. The risk of loss of volatile species was reduced by adjusting the iodine-containing analyte solution at alkaline pH and hence suppressing the volatility of hydrogen iodide.17–19 Volatilization of iodine led to severe memory effects which may otherwise cause problems in the sample introduction system of the ICP mass spectrometer instrument.19 Alternatively a wet ashing step that makes use of a strong oxidizing agent such as perchloric acid converted volatile iodine to non-volatile species.11–14 The aim of this paper is to describe the development and evaluation of an analytical method which is based on sample dissolution by wet ashing using a mixture of nitric and perchloric acids in closed steel bombs in combination with ICP mass spectrometer detection.Potential sources of error during sample introduction into the ICP mass spectrometer have been given special attention and the method has been used for the iodine analysis in a range of certified reference materials (CRMs).EXPERIMENTAL 435 Germany). Prior to using new PTFE liners for analytical work the amount of iodine that contaminated the PTFE was reduced by treatment with 4 ml of nitric acid at 160 °C for 4 h. The PTFE liners which were submitted to this procedure were reserved for the analysis of iodine in order to keep the risk of additional external contamination by iodine to a minimum. When not in use the liners were filled with a mixture of 2 ml of nitric acid and 48 ml of water. Instrumentation and Instrumental Settings The iodine content of the diluted wet-ashed residues was determined by a Perkin-Elmer SCIEX ELAN 5000 ICP-MS instrument (Perkin-Elmer SCIEX Concord Ontario Canada).The instrument was run at normal resolution and set to detect the signal intensity at m/z 127 (127I+) in the quantitative and in the graphics data acquisition modes which allowed quanti- fication and recording of the signal intensity versus time respectively. The instrument optimizations included physical positioning of the plasma relative to the mass spectrometer the lens voltage settings rf power and nebulizer gas flow. A 20 ng ml-1 aqueous standard solution of iodine as iodate normally produced a signal of approximately 60000 counts s-1. An AS-90 autosampler with polypropylene sample vials was used in conjunction with a peristaltic pump for the introduction of the sample solutions via two separately tested nebulizer/spray chamber assemblies.Further details of the instrumental settings are given in Table 1. Table 1 Instrumental settings and calibration ICP-MS instrument— Rf power Sampler and skimmer cones Argon flow rates Outer Intermediate Nebulizer Mass-to-charge ratio detected Quantitative mode Dwell time per mass Sweeps per reading Readings per replica Number of replicates Scanning mode Graphics mode (signal intensity versus time) Dwell time per mass Sweeps per reading Readings per replicate Number of replicates Estimated time Scanning mode Sampling system— Autosampler Wash time between samples Read delay Peristaltic pump speed Spray chamber and nebulizer assemblies Calibration— Type Working standard solution Added volumes 436 1000–1300W Platinum 15 l min-1 0.8 l min-1 1 l min-1 (variable) m/z 127 1 3 80 1000 ms Peak hop 1 5 50 ms 1000 278 s Peak hop 120 s 80 s 2.5 ml min-1 (a) Ryton double-pass (Scotttype) with a gem-tipped cross-flow nebulizer (b) Glass cyclonic with a Meinhard (type TR-30-K3) concentric nebulizer Standard additions 10 mg Iml-1 as potassium iodate 50 ml and 100 ml (variable) Journal of Analytical Atomic Spectrometry April 1997 Vol.12 Blank Values and Limit of Detection In spite of thorough cleaning of the utensils by nitric acid prior to the analytical work the iodine concentration level of the blanks significantly exceeded zero.Rigorous testing of each of the possible sources of contamination (the PTFE material pipette tips acids etc.) showed that the PTFE was the only notable source. The iodine blank concentration (average±one Standard Substances and Chemicals An aqueous standard stock solution at 1000 mg ml-1 of iodine was prepared from potassium iodate volumetric standard which contained 99.95–100.05% iodine (Merck Darmstadt Germany). Additionally a commercial 1000±0.5 mg ml-1 iodine standard as sodium iodide in water was acquired from the producer (Teknolab Drøbak Norway). Working standard solutions at 10 mg ml-1 were prepared daily from these stock solutions by dilution with water. Nitric acid pro analysi (Merck) which was sub-boil distilled in an all-quartz apparatus (Hans Ku�rner Rosenheim Germany) and perchloric acid pro analysi were used for the wet ashings.Water (>18 V cm-1) was produced in a Millipore Super-Q apparatus (Millipore Milford MA USA). Samples A variety of CRMs produced by NIST (National Institute of Science and Technology MD USA) and BCR (Community Bureau of Reference Brussels Belgium) certified for total iodine (Table 2) were analysed. The samples were continuously stored in a desiccator at room temperature. The residual water content in the CRMs stored under these conditions was around 2–5%. Procedure Ten to fourteen bombs were taken through the procedure in parallel. Two randomly selected bombs were filled with the digestion acids mixture and taken through the entire procedure to monitor the average and variation of the iodine blank value.Sub-samples of 0.1–0.5 g (dry mass) were weighed into the PTFE liners of the bombs followed by addition of first 3.5 ml of nitric and then 1.5 ml of perchloric acid. After addition of the acids the bomb was immediately closed to prevent the risk of volatilization of iodine and then heated at 160°C for 4 h during the night. After cooling the bomb was opened in a fume hood and the built-up pressure was gently released. Each of the wet-ashed residues was taken to 20 g by water which was added directly to the tared PTFE liner. The volume of this diluted sample residue was then calculated after determination of the density which was (average and standard deviation) 1.115±0.016 g ml-1 (n=24) by weighing of a separate aliquot of the residue.The iodine content of this solution remained non-volatile for at least 5 days. Four millilitres of the solution which was spiked with appropriate volumes of the iodate working standard solution for the purpose of quantification by the standard additions procedure was taken to 10 ml by water in the autosampler vials. This spiked and diluted analyte solution must be analysed on the same day as the iodine may be converted to volatile forms upon storage. Prior to performing quantitative analysis the iodine signal intensities of the diluted sample solutions were recorded versus time in the graphics mode. The shape of the signal profile was used to monitor that the iodine was present as non-volatile species.Following this evaluation the samples were measured in the quantitative mode. For each set of samples two blanks were taken in parallel through the entire procedure. RESULTS AND DISCUSSION Table 2 Quantitative results for iodine in certified reference materials Certified values 0.167±0.024 Reference material (acronym) Hay powder 1.29±0.09 BCR 129 NIST 1570 BCR 150 BCR 186 NIST 1572 NIST 1566a Spinach Milk powder Pig kidney Citrus leaves Oyster tissue Cod muscle 1.84±0.03 4.46±0.42 4.95±0.49 BCR 422 * Indicative value (BCR). standard deviation) was 0.18±0.11 ng ml-1 (n=13) in the final solution for measurement or equivalent to 9 ng per ashing.The LOD based on three standard deviations of the blank is 30 ng g-1 (dry mass) when 0.5 g sample is taken for analysis. This corresponds to an LOD value of approximately 5 ng g-1 for the equivalent amount of fresh sample. New liners which were not previously cleaned led to a contamination level by iodine approximately ten times that found after cleaning. However an attempt to further reduce the blank by other cleaning agents (alkaline detergent) and procedures (sonication) was unsuccessful. Conversion of Volatile Iodine to Non-volatile Species The composition of the acid mixture for the wet ashing was selected with the aim of mineralizing the organic matter of the sample and of converting volatile iodine species e.g.hydrogen iodide in the acidic sample solution to non-volatile species e.g. iodate. This conversion was necessary to prevent the risk of analyte loss and to prevent severe memory problems in the sample introduction system of the ICP-MS detector. Perchloric acid is well suited for such an oxidation at the elevated temperature and pressure which occurs during the bomb ashing. The minimum amount of perchloric acid that was necessary to ensure the oxidation of iodide was about 1.5 ml when used in combination with 3.5 ml of nitric acid for 0.5 g of dry sample. The closed bomb system additionally reduces the risk of loss by evaporation of the analyte during the wet ashing procedure. When using perchloric acid extreme care should be taken to prevent explosions which are likely to occur if concentrated hot perchloric acid is in contact with biological (oxidizable) materials.The use of the closed bomb system however prevents the evaporation of the nitric acid from the acid mixture hence the risk of explosion is reduced to a minimum. When analysing food samples high in fat such as cheese or fatty meat the amount of sample taken for analysis should be reduced to 0.1–0.2 g to prevent excessive pressure build-up due to violent oxidation or ultimately an explosion. The successful conversion of volatile iodine to non-volatile species was indicated by a signal intensity profile (Fig. 1A) recorded in the graphics mode identical to that normally observed in ICP-MS studies of non-volatile elements.In contrast the ICP-MS signal intensity profile showed pronounced fronting and did not reach a steady-state level when volatile iodine species were present in solution (Fig. 1B) and hence quantification became impossible. Furthermore the pronounced tailing of the signal required long wash-out times and increased the risk of analyte carry-over. This unusual signal profile was probably caused by volatile iodine species which adhered to the surfaces of the sample introduction system and tubing etc. In this study the closed high-pressure steel bombs were used to prevent the possible loss of iodine by evaporation during the wet ashing procedure. The use of microwave-assisted wet Journal of Analytical Atomic Spectrometry April 1997 Vol. 12 Sample Introduction When analysing a potentially volatile element like iodine optimum and interference-free sample introduction is of great importance.With the aim of optimizing the iodine signal intensity and minimizing the wa-out time between samples the standard double-pass spray chamber with a gem-tip cross- flow nebulizer was compared with a low-volume glass cyclonic spray chamber equipped with a concentric nebulizer. The ICP-MS signal intensities obtained with these nebulizer-spray chamber assemblies upon aspiration of a 20 ng ml-1 aqueous solution of iodine as iodate (Fig. 2) show that the signal intensity was improved by a factor of two when using the cyclonic spray chamber. With this type of sample introduction device a markedly shorter aspiration time was necessary to arrive at a steady-state signal.Furthermore the iodine signal intensity versus time in Fig. 3 shows that the wash-out time necessary to arrive at the base-line level with the cyclonic Iodine concentration/mg g-1 Other values This study 0.15; 0.17 1.089; 1.166 0.145* 1.09; 1.02 1.15 0.16 1.76; 1.65 4.53; 4.54; 5.07 4.59 Fig. 1 ICP-MS signal intensity versus time of iodine measured in the wet-ashed residue of 100 mg Cod Muscle (BCR CRM 422) diluted to 50 ml with water A measured immediately after dilution and B measured five days after dilution. ashing with the same acid mixture might be feasible.6 However the risk of losing volatile iodine species during venting at elevated pressure in this type of system requires utmost attention and was not tested.437 Fig. 2 ICP-MS signal intensity obtained when using a cyclonic spray chamber with a concentric nebulizer for 20 ng ml-1 of I as iodate A in water; B in 3.5 ml HNO and 1.5 ml HClO diluted to 50 ml by water; C in 3.5 ml HNO 4 3 and 1.5 ml HClO diluted after bomb ashing to 50 ml by water; D in 4 3 3.5 ml HNO and 1.5 ml HClO diluted after bomb ashing of 500 mg potato starch 4 3 to 50 ml by water; and E same as A but using a standard double-pass spray chamber with a cross-flow nebulizer. Fig. 3 ICP-MS signal intensity profiles (tails) for iodine in Cod Muscle CRM (cf. Fig. 1) using A cyclonic spray chamber with a concentric nebulizer; and B double-pass spray chamber with a cross-flow nebulizer.spray chamber is about 40 s in comparison with 100 s when using the double-pass spray chamber. The pulsed noise which was observed with the latter spray chamber (Fig. 3B) may be ascribed to pressure pulsations from the peristaltic pump20 which was used for draining the spray chamber. Even though this pump was also used for draining the cyclonic spray chamber the pulsed noise did not show. The use of the low dead-volume cyclonic spray chamber with the concentric nebulizer is advantageous as it provides better analyte sensitivity without increasing the base-line noise which results in an Journal of Analytical Atomic Spectrometry April 1997 Vol. 12 Signal Enhancement by Methanol The signal intensity of elements with first IE values in the 9–11 eV range (e.g.iodine arsenic and selenium) may be enhanced by introduction of carbon as methanol22 or as other carbon-containing molecules23 into the ICP. In order to optimize this enhancement effect for iodine the S/N for the aqueous and methanol–water (3+97 by volume) solutions of iodate were recorded at m/z 127 at different rf power settings as shown in Fig. 4. The S/N increased only slightly for the aqueous solution when an optimum rf power of 1100W was applied (curve B). However in the 3% methanol solution the S/N was improved by 54% at the 1200 W optimum power setting. At the 1300 W power setting the iodine sensitivity was even greater but the noise increased at the same time and the Fig. 4 ICP-MS S/N-ratio at m/z 127 at different rf power settings for 20 ng ml-1 of I as iodate in A methanol–water mixture (3+97 by volume); and B aqueous solution.The nebulizer gas flow settings for optimum signal intensity (ordinate axis to the right) are given for C methanol–water mixture (3+97 by volume); and D aqueous solution. 438 improved S/N value and hence improved LOD as well as shorter times of analysis. Matrix Effects The first ionization energy (IE) for iodine of 10.3 eV is relatively high compared with most other elements. Consequently the ionization of this element in the argon ICP is incomplete and has been estimated at 29%.21 The sensitivity of iodine obtained with the ICP-MS instrument is however still favourable because iodine is monoisotopic (127I). The sensitivity (Fig.2) obtained for a 20 ng ml-1 solution of iodine as iodate in an acid matrix containing 4.4% nitric acid and 2.0% perchloric acid (curve B) is 74% of that obtained for the same iodate concentration in aqueous solution (curve A). This acid matrix is close in composition to that of the ashed and diluted sample solutions. When the same amount of iodate was subjected to the bomb ashing procedure the sensitivity (curve C) dropped further to 67% of that for the aqueous solution. The sensitivity was exactly the same (curve D) when the iodate was bombashed in the presence of 0.5 g of a biological material (potato starch). The reduction in iodine sensitivity may be caused by a higher liquid viscosity of the acid-containing solutions which reduces the analyte uptake rate.A reduced ionization efficiency of iodine in the chlorine-rich solution may additionally explain the observed effect. The results show that external calibration using an aqueous standard solution or using a matrix-matched acid solution led to inaccurate (erroneously low) results. Therefore the method of standard additions with iodate was used for calibration. resulting S/N decreased slightly (curve A). The nebulizer gas flow necessary for optimum analyte sensitivity varied with changing rf power settings and solvent composition22 (curves C and D). In the presence of 3% methanol in solution the optimum nebulizer gas flow is reduced by approximately 0.05 l min-1. For iodine analyses where an optimum LOD value is required the addition of 3% methanol to the aqueous analyte solution and the use of 1200 W rf power is therefore recommended for the ICP-MS instrument used.Under these analytical conditions nickel sampler and skimmer cones degrade rapidly and the use of platinum cones is recommended.22 CONCLUSIONS An analytical method based on ICP-MS has been presented for the accurate determination of iodine in food-related CRMs. Quantitative Determination of Iodine in Food-related CRMs The performance of the analytical method was tested using CRMs which covered a variety of sample types and iodine concentrations. Prior to running the quantitative analyses all samples were run in the graphics mode to ensure that the ICP-MS signal stabilized at a steady-state intensity within 5–10 s as indicated in Fig.1A. The two additions of iodate for the standard additions calibration procedure were 50–150% and 150–250% respectively in concentration of the expected iodine concentration in the diluted sample solution. Calibration curves constructed this way showing a coefficient of correlation less than 0.999 were not used for quantification as they reflected a deviation from linearity which led to inaccurate results. The slopes of the standard addition calibration curves were close in value for different sample types and the within-day RSD value was 3.4% (n=8). For routine work a larger sample throughput may therefore be obtained by calibrating several unknown samples against one common standard additions calibration curve.When used in the quantitative mode the RSD values of the recorded ICP-MS signal for unknown samples were 0.4–2.9%. The highest values were associated with the lowest measured iodine concentrations. Signal intensity RSD values within this range were used to additionally control the stability of signal intensities during the quantitative measurements. In the case of a non-steady state signal intensity for iodine (cf. Fig. 1B) this RSD value would increase markedly. The quantitative results for iodine in CRMs (Table 2) are in good agreement with the certified or literature values. The proposed method will be used in the near future for an extensive study of iodine in fish and dairy products on the Danish market. Further data on the reproducibility and accuracy will be provided during the course of these investigations.Journal of Analytical Atomic Spectrometry April 1997 Vol. 12 The method fills a gap in the use of ICP-MS for the determination of iodine in solid biological materials. The key to a successful analysis was to ensure that volatile iodine was oxidized to non-volatile species. This was achieved by using perchloric acid and nitric acid at elevated temperature in closed bombs. REFERENCES 1 Orr J. B. and Leitch I. Iodine in nutrition. A review of existing information up to 1927 Special Report No. 123 Medical Research Council London 1929. 2 Nordic Council of Ministers Nordic Nutrition Recommendations Report 198952 Copenhagen 1989. 3 Frey H. Iodine in Risk Evaluation of Essential T race Elements — Essential Versus T oxic L evels of Intake ed.Okarsson A. Nordic Council of Ministers Copenhagen 1995 pp. 119–132. 4 Fischer P. W. L’Abbe� M. R. and Giroux A. J. Assoc. Off. Anal. Chem. 1986 69 687. 5 Dermelj M. Slejkovec Z. Byrne A. R. Stegnar P. Stibilj V. and Rossbach M. Fresenius’ J. Anal. Chem. 1990 338 559. 6 Rao R. R. and Chatt A. Anal. Chem. 1991 63 1298. 7 Rao R. R. and Chatt A. Analyst 1993 118 1247. 8 Langenauer M. Kra�henbu�hl U. and Wyttenbach A. Anal. Chim. Acta 1993 274 253. 9 Mantel M. Analyst 1988 113 973. 10 Heumann K. G. and Schindlmeier W. Fresenius’ Z. Anal. Chem. 1982 312 595. 11 Gramlich J. W. and Murphy T. J. J. Res. Nat. Inst. Stand. T ech. 1989 94 215. 12 Schindlmeier W. and Heumann K. G. Fresenius’ Z. Anal. Chem. 1985 320 745. 13 Fasset J. D. and Murphy T. J. Anal. Chem. 1990 62 386. 14 Holak W. Anal. Chem. 1987 59 2218. 15 Mitsuhashi T. and Kaneda Y. J. Assoc. Off. Anal. Chem. 1990 73 790. 16 Allain P. Mauras Y. Douge� C. Jaunault L. Delaporte T. and Beaugrand C. Analyst 1990 115 813. 17 Baumann H. Fresenius’ J. Anal. Chem. 1990 338 809. 18 Stu�rup S. and Bu�chert A. Fresenius’ J. Anal. Chem. 1996 354 323. 19 Vanhoe H. Van Allemersch F. Versieck J. and Dams R. Analyst 1993 118 1015. 20 Luan S. Pang H. Shum S. C. K. and Houk R. S. J. Anal. At. Spectrom. 1992 7 799. 21 Jarvis K. E. Gray A. L. and Houk R. S. Handbook of Inductively Coupled Plasma Mass Spectrometry Blackie Glasgow 1992. 22 Larsen E. H. and Stu�rup S. J. Anal. At. Spectrom. 1994 9 1099. 23 Allain P. Jaunault L. Mauras Y. Mermet J.-M. and Delaporte T. Anal. Chem. 1991 63 1497. Paper 6/07581I Received November 7 1996 Accepted Janua