2005 PerkinElmer The Study of Mass Spectrum Interferences in the Determination of High Purity 1 Rare Earth Products by using Inductively Coupled Plasm

Transcription

1 2005 ( ) PerkinElmer ICP-OES ICP-MS

2 2005 PerkinElmer The Study of Mass Spectrum Interferences in the Determination of High Purity 1 Rare Earth Products by using Inductively Coupled Plasma Mass Spectrometry Direct Determination of Rare Earth Impurities in High Purity Ytterbium Oxide 15 by Inductively Coupled Plasma Mass Spectrometry 21 ICP-MS 30 -ICP-OES 35 ICP-AES 38 ICP-OES ICP-AES ICP-OES 60 - Si Fe Na ICP-OES 15 75

3 PerkinElmer Environmental Applications o f the ELAN DRC-e ICP-MS 82 Interference Removal and Analysis of Environmental Waters Using the ELAN 86 DRC-e ICP-MS Bromine Speciation by HPLC/ICP-MS 96 Simultaneous Arsenic and Chromium Speciation by HPLC/ICP-MS in 100 Environmental Waters Radionuclide and trace element contamination around Kolaghat Thermal 103 Power Station, West Bengal Environmental implications Using Dynamic Reaction Cell ICP-MS Technology to Determine the Full Suite 111 of Elements in Rainwater Samples Experience Using Filter Paper Techniques for Whole Blood Lead Screening in 118 a Large Pediatric Population As(III) and Sb(III) Uptake by GlpF and Efflux by ArsB in Escherichia coli 126 Continuous Flow Hydride Generation Using the Optima ICP 134 Trace Level Analysis of Calcium, Magnesium, Potassium and Sodium Using 138 the Optima ICP Determination of Impurities in High-purity Gold 142 Determination of Impurities in Purified Terephthalic Acid (PTA) 145 Correction for non-spectroscopic matrix effects in inductively coupled 148 plasma-atomic emission spectroscopy by internal standardization using spectral lines of the same analyte A Drift Correction Procedure 159 Single-Element Solution Comparisons with a High-Performance Inductively 166 Coupled Plasma Optical Emission Spectrometric Method An ICP-OES Method with 0.2% Expanded Uncertainties for the 175 Characterization of LiAlO 2 Natural Remediation of Contaminants along the Forgotten River Stretch of the 183

4 Rio Grande Cardiovascular effects of Tacca integrifolia Ker-Gawl extract in rats 190 Determination of organo-zinc based fungicides in timber treatments employing 199 gas chromatographic analysis with mass selective detection and/or inductively coupled plasma atomic emission spectroscopy Structure and magnetic properties of rf thermally plasma synthesized Mn and 207 Mn Zn ferrite nanoparticles Field scale study results for the beneficial use of coal ash as fill material in 210 saturated conditions, varra coal ash burial project, weld county, colorado Digestion and Characterization of Ceramic Materials and Noble Metals 214

5 2005 PerkinElmer ICP-OES ICP-MS The Study of Mass Spectrum Interferences in the Determination of High Purity Rare Earth Products by using Inductively Coupled Plasma Mass Spectrometry Xinquan Zhang*, Jinlei Liu, Yonglin Liu, Xiang Li, Yong Yi, Yumei Jiang, Yaqin Su Jiangxi Institute of Analyzing and Testing, Nanchang , China Abstract This paper reported the mass spectrum interfering problems during analytical procedures with trace rare earth elements (REES) as impurities in fifteen high purity rare earth products by using ELAN 9000 inductively coupled plasma mass spectrometry (ICP-MS) technique. The interfering effects caused by interfering species originated from background mass spectrum, isobaric overlaps, mass spectrum overlaps and matrix-induced polyatomic ions that interfer the quantification of the specified analytes were evaluated deeply. The relationship between polyatomic ions originated from matrix-induced and the instrumental conditions was also investigated. Through quantitative analysis, the appearance concentrations of polyatomic ions arose from matrix-induced and matrix mass spectrum overlaps were presented. According to the mass spectrum interferences, the resolving methods of correction interferences were put forward. The limit of quanlitation (LOQ) for trace of REEs impurities and acquiring purity under direct determination of fifteen high purity rare earth products were also investigated. Keywords: Inductively coupled plasma mass spectrometry; High purity rare earth products; mass spectrum interferences 1. Introduction With the development of widely applications of high purity rare earth products in many fields, such as industry, nuclear technique, aviation and spaceflight, the demand for purity analysis of rare earth products with precision and accuracy is increasing. In general, several analytical techniques such as X-ray fluorescene (XRF), neutron activation analysis (NAA) and inductively coupled plasma emission spectrometry (ICP-AES)[1-3] have been employed for the quantification of REEs as impurities in relative lower purity rare earth products. However, these analytical techniques, are only effective for the lower purity of rare earth products but not adequate for rare earth products with high purity (99.99%~99.999%) in terms of sensitivity, complex spectral interference [4-5]. After Houk et al.[6] first published their methodology by using inductively coupled plasma (ICP) as an ion soure combined with mass spectrometer (MS), twenty five years have passed. Today, This analytical technique regarded as a principal measure is widely applied in the analytical fields such as high purity materials, circumstance, metallurgy, biochemistry et al. Due to its performance with high sensitivity, low detection limits, determination capability of multielements at ultra-trace level at the same time and wide dynamic range, it also becomes the major analytical method used for the determination of ultra-trace REEs as impurities in high purity rare earth products day by day[7-14]. Although ICP-MS technique possesses protrusive advantages compared with other atomic 1

6 2005 PerkinElmer ICP-OES ICP-MS spectrometric techniques, but some mass spectrum interferences produced from plasma- and matrix solvent-induced polyatomic ions as background spectrum, isobaric overlaps, matrix-induced polyatomic ions and matrix-induced spectrum overlaps that affect the neighbouring two mass numbers (M-1, M+1) analytes, limit the applications in the analytical field of high purity materials in certainly degree. Zhang et al.[11,12] utilized ICP-MS technique and reported the determination of the ultra-trace REEs as impurities in high purity CeO 2 and Sm. Through the method of matrix-match, the mass spectrum interferences generated from matrix-induced polyatomic ions and matrix-induced spectrum overlaps were overcame partly, then the determination of the two kinds of high purity rare earth products was enhanced to 99.9%~99.99%. Pedreira et al. [13, 14] accomplished the determination of rare earth analytes as impurities in high purity La 2 O 3 and Pr 6 O 11 with ICP-MS combined with high-performance liquid chromatography (HPLC) technique after matrix elements as La and Pr had been separated respectively by HPLC. Some papers evaluated the mass spectrum interferences associated with the matrix, such as polyatomic ions (MO +, MOH + ) generated in the plasma [15, 16]. But few of papers maked a systematic and quantitative investigation utilized ICP-MS technique in the determination of fifteen rare earth products during analytical procedures about a series of mass spectrum interference generated from background spectrum, isobaric overlaps, 2 matrix-induced polyatomic ions and matrix-induced mass spectrum overlaps, which interfere in the quantification of the REEs of interest with precision and accuracy. This is the exactly purpose that we will be discuss in this paper. 2. Experimental 2.1 Instrument A Perkin-Elmer Sciex ELAN 9000 ICP-MS equipped with a integrated peristaltic pump and a cross flow nebulizer combined with a scott spray chamber that was made from Ryton TM material was used. The instrumental conditions are summarized in Table 1. Before the analytical procedure, the X, Y positions of torch, the nebulizer gas flow and the lens voltage were optimized by using a quality concentration standard solution of 10µgL -1 Mg, Ce, Rh, In, Pb, Ba. After those optimization was accomplished, then optimized Auto Lens TM voltage by using another quality concentration standard solution of 10µgL -1 Be, Co, In. Finally, selected the Daily Performance. wrk from ELAN software to check the instrumental performance, confirmed that the sensitivity of spiked elements as In, Mg, Pb with quality concentration of 10µgL -1 amounted to about more than counts per second (cps), cps and Pb for cps respectively, percentage 140 Ce 16 O + / 140 Ce + and Ba 2+ /Ba + ratios about less than 3.0%, and the background of mass number 220 was also less than 10 cps. Table 1 Operating condition of ELAN 9000 ICP-MS RF Power (w) 1000 Resolution (amu) 0.70 Nebulizer Gas Flow (Lmin -1 ) 0.90 Detector dual Lens voltage (v) 6.0 Speed of Peristaltic Pump (rpm) 26 Analog Stage Voltage (v) Sweeps/Reading 3 Pulse Stage Voltage (v) 1000 Replicates 3 Discriminator threshold (v) 70 Scan Mode Peak hopping or scanning Ac rod offset (v) Reagents The superpure HNO 3 used in this work was acquired by the analytical-reagent grade HNO 3 under sub-boiling condition. H 2 O 2 was the AR grade. Sample solutions were prepared with purified water from a TYUV05 water unit with the resistivity being more than 18.3M cm ( Tao Yuan

7 2005 PerkinElmer ICP-OES ICP-MS Company, Nanchang, China). 2.3 Samples and sample preparation High purity samples with purity more than about %: La 2 O 3, Sm 2 O 3, Eu 2 O 3, Ho 2 O 3, Er 2 O 3, Tm 2 O 3, Yb 2 O 3, Lu 2 O 3 and Y 2 O 3 were acquired from Hunan Institute of Rare Earth Material (Hunan, China). High purity samples with purity more than about 99.99%: CeO 2, Pr 6 O 11, Nd 2 O 3, Gd 2 O 3, Tb 4 O 7 and Dy 2 O 3, were purchased from Johnson Matthey Company (Warel Hill, USA). 0.1g of the fifteen rare earth samples was accurately and respectively weighted into a 100mL beaker, and decomposed with ultrapure water 10mL and superpure HNO 3 2mL on a hot plate which heated at about 80 (3mL H 2 O 2 was added in addition when CeO 2 was decomposed). After the dissolving solution cooled, transferred it into a 100mL volumetric flask individually and finally diluted with ultrapure water, so the quality concentrations of each of the fifteen rare earth products were 1.0gL -1. After those dissolving solutions regarded as stock solution were accomplished with preparation procedures, then diluted them into 50mL volumetric flask with ultrapure water and added 0.5mL superpure HNO 3 in addition. Finally, The quality concentrations of the matrix of fifteen rare earth samples in the solutions were 1.0 gml -1 and 500µgmL -1 respectively. The 500 gml -1 solutions were spiked with In as internal standard with the quality concentration of 10 gl -1. Prepared blank solution at the same time. All of the beakers and volumetric flasks used in this work were carefully washed with 10% HNO 3 and then washed repeatedly with ultrapure water in order to reduce blank value. Analyte 2.4 ICP-MS determination The ELAN software provides two different modes for acquiring mass spectral data of interest during a analytical procedure. These options are the modes of peak hopping and scanning. In the solutions which fifteen rare earth matrix concentration were 500 gml -1, the appearance concentrations of the interfering ions generated from matrix-induced polyatomic ions and matrix-induced mass spectrum overlaps were obtained quantitatively by the option of selecting peak hopping mode and setting the MCA channels value to one. Otherwise, in the solutions which fifteen rare earth matrix concentrations were 1.0 gml -1, the mass spectrum were acquired individually through the option of selecting scanning mode and setting the MCA channels value to eleven. 3. Results and discussion 3.1 Plasma and matrix solvent induced polyatomic ions The presence of atmospheric gases or the argon carrier is source of potential interfering ions as the result of reaction with other analyte or matrix components [17]. These interfering ions occur as polyatomic, molecular ions. In addition, the solvent or acid in which the sample is dissolved can also be another source of interfering polyatomic ions. Chlorine from hydrochloric or perchloric acid and sulfur from sulfuric acid all form polyatomic ions with argon and other plasma gases. For organic solvents, carbon and oxygen can form polyatomic ions. Interfering ions of various types listed above can affect the determination of some analytes, These are illustrated in Talbe 2. Table 2 Plasma-induced and matrix solvent-induced polyatomic ions interferences Abundance of Isotope 28 Si S K Ca Ca Ti Interfering Ion Analyte Abundance of Isotope 12 C 16 O + 51 V O Fe ArH + 64 Zn Ar + 75 As C 16 O Se S 16 O + Interfering Ion 35 C 16 O + 40 Ar 16 O + 32 S 16 O + 2, 32 S Ar 35 Cl + 40 Ar + 2 3

8 2005 PerkinElmer ICP-OES ICP-MS Taking an analytical procedure of mass spectral scanning with blank solution, the mass spectrum as Figure 1 was obtained. It is evident from Figure 1 that those interfering ions listed in Table 2 mainly affect the determination of the analytes which mass numbers are less than 80. In the periodic table, the mass range of 14 REEs (La~Lu) varies from 138 to 176 and the lowest mass of rare earth products is 89 of yttrium, so there has no influence of spectral overlap interferences that can affect the determination of REEs as impurities in high purity products because the mass of those polyatomic ions are less than 80. It can be seen from Figure 2 of the mass spectrum of Y 2 O 3 with a solution of 1.0 gml -1. Fig. 1. Background mass spectrum with blank solution Fig. 2. Y 2 O 3 mass spectrum with concentration of 1.0µgmL Isobaric overlaps interference Since most elements have more than one naturally occurring isotope, it is possible for the mass spectrum of an isotope of one element to directly overlap that of an isotope of another element. These interferences are termed isobaric overlaps. It means that some elements can not be determinate with accuracy due to the isobaric overlaps interferences. Most of the 15 REEs have more than one naturally occurring isotope except several REEs have only one isotope, such as 89 Y, 141 Pr, 159 Tb, 165 Ho and 169 Tm. So there have some overlap interferences that affect the results of the determination of trace amount of REEs as impurities in high purity rare earth products during ICP-MS analytical procedures. Table 3 illustrates the isobaric between some REEs. Table 3 Isobaric of rare earth elements* Isobaric Analyte Isobaric Analyte 138 La(0.09), 138 Ce(0.26) 156 Gd(20.6), 156 Dy(0.06) 142 Ce(11.1), 142 Nd(27.1) 158 Gd(24.7), 158 Dy(0.1) 144 Nd(23.9), 144 Sm(3.1) 160 Gd(21.7), 160 Dy(2.3) 148 Sm(11.2), 148 Nd(5.7) 162 Dy(25.5), 160 Er(0.14) 150 Nd(5.6), 150 Sm(7.4) 164 Dy(28.2), 164 Er(1.6) 153 (26.7), 152 Gd(0.2) 168 Er(27.0) 168 Yb(0.14) 154 Sm(22.8), 154 Gd(2.2) 170 Er(15.0), 170 Yb(3.0) 176 Yb(12.7), 176 Lu(2.6) *: (x), x refers to abundance value This problem can be overcomed by selecting of isotopes carefully when developing analytical methods. For example, in the determination of trace of amount of REEs as impurities in high purity La 2 O 3 with purity varying from % to % [18], it had isobaric overlap interferences that affected the determination of the analyte 4

9 2005 PerkinElmer ICP-OES ICP-MS of Ce. The reason for this was that overlap interferences existed between 136 Ce(0.19) and 136 Xe(8.9), 138 Ce(0.26) and 138 La(0.09), 142 Ce (11.1) and 142 Nd(27.1). Otherwise, overlap interference occurred from matrix [ 139 La(99.9)]-induced affected the isotope as 140 Ce(88.5). Under the circumstance of not changing the instrumental resolution value, the concentrations of specified analytes of Ce and Nd were obtained by selecting isotopes as 142 Ce(11.1) and 143 Nd(12.2). After the concentration of isotope 143 Nd(12.2) was obtained, then the concentration of isotope 142 Ce(11.1) was also acquired by subtracting the contributing of 142 Nd(27.1) to the mass of 142, that was established a correction equation: C (142)Ce =C 142 -Abundance of 142 Nd/Abundance of 143 Nd C (143)Nd. Where C 142 was the total concentration at mass 142. Mirror isobaric overlaps can be corrected through establishing elemental equations like this. The instrumental software provides some equations and automatically corrects for known isobaric overlaps when calculating analyte concentrations in samples. But we must make a concrete analysis according to concrete condition in order to acquire accurary results. 3.3 Matrix-induced polyatomic ions interferences Polyatomic ions, such as MO +, MOH and MO + 2 may also be generated by sample matrix combination with oxygen and hydrogen [15,16]. These interfering ions create extreme influence on the measurement of the trace amount of REEs as impurities in high purity rare earth products. In order to protect the ETP detector due to extremely higher sensitivity of ELAN 9000 ICP-MS, a series of mass spectrum of matrix and matrix-induced polyatomic ions were obtained respectively with using the solutions of fifteen rare earth products In order to distinguish accurately the forms of polyatomic ions came from matrix of high purity rare earth samples and mass spectral interfering degree, the which quality concentrations were 1.0 gml -1 and taking mass spectral scanning according to mass range individually. Figure 3 and figure 4 describe typically that matrix have generated polyatomic ions with oxygen or hydrogen and formed mass spectral interferences overlap the mass of analytes of interest, such as 163, 164, 165, 166, 167, 168, 169, 170, 171 from Sm-matrix and 170, 171, 172, 173, 174, 175 from Gd-matrix. Fig. 3. Sm 2 O 3 mass spectrum with concentration of 1.0µgmL -1 Fig. 4. Gd 2 O 3 mass spectrum with concentration of 1.0µgmL -1 high purity La 2 O 3 with quality concentration of 500 gml -1 solution was selected and then executed a mass spectral scanning procedure. The mass spectrum was acquired without selecting isotopes of 5

10 2005 PerkinElmer ICP-OES ICP-MS the matrix element as 138 La or 139 La and that is illustrated in Figure 5, It can be seen from figure 5 that there have many polyatomic ions generated from La matrix combination with oxygen and hydrogen present at mass 154, 155, 156, 157, 158 and 171, with corresponding in order, these polyatomic ions are 138 LaO +, 139 LaO +, 139 LaOH +, 139 LaOH + 2, 139 LaOH + 3 and 139 LaO + 2. I mass /I In /C (138)La or C (139)La 100% (1) I mass /I In (2) Where: I mass refers to the intensity of the specified mass I In refers to the intensity of the internal reference as In of cps (n=3) C (138)La refers to concentration of 138 La[500µgmL -1 Abundance(0.09%)=0. 45µgmL -1 ] C (139)La refers to concentration of 139 La[500µgmL -1 Abundance(99.9%)=49 9.5µgmL -1 ] 10 refers to concentration of internal reference as In, gl refers to coefficients for unit of gl -1 converts as gml -1 The producing ratios and appearance concentrations for interfering ions occurred from matrix-induced polyatomic ions are shown in Table 4. Fig. 5. La 2 O 3 mass spectrum with concentration of 500µgmL -1 The producing ratios and appearance concentrations of polyatomic ions were quantitative determinated with selecting peak hopping scan mode and calculated as following equations (1) and (2): Table 4 Producing ratios and appearance concentrations of polyatomic ions generated from matrix of La with concentration of 500µgmL -1 (n=3) Polyatomic ions Intensity (cps) Producing ratio (%) Appearance concentration ( gml -1 ) 138 LaO LaO LaOH LaOH LaOH LaO Table 4 shows that the producing ratios for interfering ions described as above vary from % to 2.42%, and appearance concentrations from µgmL -1 to 12.08µgmL -1, that is the overlap interfering concentrations amount to from 0.00xµgmL -1 to xxµgml -1. The interference of polyatomic ion as MO + is the most serious interference and that for MOH + is the second. The interfering degree of these polyatomic ions is in the order: MO + >MOH + >MOH + 2 >MOH + 3 >MO + 2. It needs to notice in this example that the matrix element La has another naturally occurring isotope as 138 La(0.09). It can also form polyatomic ions as same as that of isotope of 139 La(99.9). The mass spectrum of these interfering ions overlap that of mass of 154, 155, 156, 157, and 170 in order. Since the concentration of 6

11 2005 PerkinElmer ICP-OES ICP-MS 138 La(0.09) is only of 0.45µgmL -1, the concentration of 138 LaO + generated from 138 La(0.09) is about of 0.01µgmL -1 and that of 138 LaOH and 138 LaO + 2 formed by 138 La(0.09) are more lower than 0.01µgmL -1, so it can be neglected that interferences caused by these polyatomic ions as 138 LaOH and 138 LaO + 2 overlap the mass of 155, 156, 157 and 170. In the application of the analysis of fifteen high purity rare earth products using ICP-MS analytical technique, only four high purity products of Y, Ho, Tm and Lu could be directly determinate the trace of amount of REEs as impurities [10,19]. The reason for this was that it was difficult for accurate determination of some of the REEs as impurities in the eleven high purity products because of the interfering ions arose from matrix-induced polyatomic ions and else. The overlap interferences caused by matrix-induced polyatomic ions could be partly overcame by using the procedure of matrix-match [11,12], but this method could not solve completely the problem about mass spectral interferences occurring from matrix-induced polyatomic ions. Determination of trace amount of REEs in high purity rare earth products using ICP-MS after separation of matrix is another often taken measure for overcoming mass spectrum of overlap interferences generated from matrix-induced polyatomic ions. Li et al. [20] reported for the determination of trace of REEs in high purity CeO 2 by ICP-MS after separation by solvent extraction. After removal of the matrix by using 2-ethlhexyl hydrogen-2-ethylhexylphosphonate (EHEHP), the polyatomic interference in ICP-MS were negligible. 3.4 Matrix mass spectral overlap interferences In the determination of trace REEs as impurities in high purity rare earth products by using ICP-MS technique, there is another interference arose from matrix (assuming mass with M) mass spectrum overlap interferences that affect the accurate measurement of the neighbouring analytes with mass for M-1 and M+1. The appearance concentration of the neighbouring analytes relation to the matrix elements in 500µgmL -1 solutions of fifteen rare earth products were acquired quantitatively by using peak hopping scan mode and setting the instrumental resolution value to 0.7amu and the results are given in Table 5. The appearance concentrations were calculated as the same as part 3.3 of equation (2). Table 5 Appearance concentration (µgl -1 ) of mass M-1 and M+1 arose from matrix mass spectrum overlap interferences (n=3)* Matrix 139 La(99.9) Analyte (M-1) Appearance concentration ( gl -1 ) Analyte (M+1) Appearance concentration ( gl -1 ) 140 Ce(88.5) Ce(11.1) 142 Ce(11.1) 141 Pr(100) 142 Nd(27.1) 146 Nd(17.2) 139 La(99.9) Ce(88.5) Pr(100) Pr ** (100) 143 Nd(12.2) Ce(11.1), 142 Nd(27.1) Sm(15.0) Sm(15.0) 152 Sm(26.7) 154 Sm(22.8) 151 Eu(47.8) 153 Eu(52.2) 158 Gd(24.7) 160 Gd(21.7) 146 Nd(17.2) 151 Eu(47.8) Nd(5.6), 150 Sm(7.4) Eu ** (52.2) 155 Gd(14.9) 152 Sm ** (26.7), 152 Gd(0.2) 154 Sm(22.8), 154 Gd(2.2) 159 Tb ** (100) 161 Dy(18.9)

12 2005 PerkinElmer ICP-OES ICP-MS 159 Tb(100) 164 Dy(28.2) 158 Gd(24.7), 158 Dy(0.1) Gd(21.7), 160 Dy(2.3) Ho(100) Ho(100) 164 Dy(28.2), 164 Er(1.6) Er(33.4) Er(33.4) 165 Ho(100) Er(27.0) 170 Er(15.0) 169 Tm(100) 174 Yb(31.8) 175 Lu(97.4) 168 Er(27.0), 168 Yb(0.14) Yb(31.8) Tm ** (100) Yb(14.3) Er(15.0), 170 Yb(3.0) Lu ** (97.4) 3.86 Note: * : If the abundance of some isotopes belong to some matrix are less than the value of 10%, since the concentrations of these isotopes are less than 50µgmL -1, so the mass spectrum overlap interferences arose from matrix for the neighbouring analytes of mass as M-1 and M+1 can be neglect; **: some analytes suffer mass spectrum overlap interferences from the two isotopes belong to a matrix element, for example, 169 Tm**(100) suffers mass spectrum overlap interferences from 168 Er(27.0) and 170 Er(15.0). It can be seen from Table 5, we can know that the interfering appearance concentrations vary from 0.94µgL -1 to 23.29µgL -1. By referring back to Table 4, upon comparison of Table 4 and Table 5, we find that the interfering appearance concentrations arose from matrix mass spectral overlap interferences approach to that of MOH + 2~3 and MO + 2 generated from matrix-induced polyatomic ions. Taking a determination of analytes of interest by selecting other isotopes which are free from matrix mass spectrum overlap interferences is one of the methods which overcome this type of interferences. Separation matrix is the other one of the method of avoiding the interferences described as above. Authorial recently work [21] accomplished successfully with the determination of the 14 REEs as impurities in high purity metal Yb with purity of % by using ICP-MS technique through with making a direct measurement for the analytes as La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er combination with separating Yb matrix by using EHEHP for determination of the analytes as Tm, Lu. In order to resolve this interfering problem, adjusting the instrumental resolution value is the another efficient method. In this work, the variations of intensity of the isotopes of 140 Ce(88.5) in 500µgmL -1 La 2 O 3 solution and that of 168 Er(27.0) and 170 Er(15.0) in 500µgmL -1 Tm 2 O 3 solution are observed under different resolution values. These are listed in Figure 6. Intensity (10 5 cps) Ce in 500 ugml -1 La 2 O Er in 500 ugml -1 Tm 2 O Er in 500 ugml -1 Tm 2 O Resolution (amu) Fig. 6. Intensity vs. resolution for 140 Ce, 168 Er, 170 Er in matrix solutions of 500µgmL -1 of La 2 O 3 and Tm 2 O 3 respectively Fig. 7. Matrix mass spectrum of 500µgmL -1 La 2 O 3 solution with the resolution setting to 0.3amu 8

13 It is evident from Figure 6 that the intensity of 140 Ce(88.5), 168 Er(27.0) and 170 Er(15.0) decreases constantly with the adjustment of the instrumental values varying from 0.7amu to 0.3amu. The matrix mass spectrum were obtained by using a solution of 500µgmL -1 for La 2 O 3 and 500µgmL -1 for Tm 2 O 3 with the resolution values setting to 0.3amu and that are given in Figure 7 and Figure 8. It is apparent that the mass spectrum of the neighbouring isotope of 140 Ce(88.5), 168 Er(27.0) and 170 Er(15.0) at mass of 140,168, 170 are completely separate from that of matrix of 139 La(99.9) and 169 Tm(100) at mass of 139 and 169. Through this experiment described as above, we can get an conclusion that it is possible that we can determinate directly the trace neighbouring analytes with concentrations vary from 0.xµgL -1 to xµgl -1 in some high purity (99.999%) rare earth products with a solution of matrix of 500µgmL -1 without using a complex procedure of separating matrix. For example, in the determination of trace amount of REEs as impurities in high purity Er 2 O 3 (purity was %) by using ICP-MS technique, since the analyte of Ho and Tm both have only one naturally occurring isotope of 165 Ho(100) and 169 Tm(100), that is the measurement of monoisotopes of 165 Ho(100) and 169 Tm(100) suffer the interferences arose from Er matrix mass spectrum overlap interferences, so the determination of high purity Er 2 O 3 (99.999%) was exactly difficult except for using a complex procedure of separating matrix. In this work, we measured directly the trace amount of 12 REEs in high purity Er 2 O 3 (99.999%) by using ELAN 9000 ICP-MS with the instrumental resolution value setting to 0.7amu and obtained that of the monoisotopes of 165 Ho(100) and 169 Tm(100) with the resolution value setting to 0.3amu. We also measured directly the concentrations of isotopes of 165 Ho(100) and 169 Tm(100) with the resolution value setting to 0.7amu, the concentrations of 165 Ho(100) and 169 Tm(100) were 22.7µgL -1 and 3.7µgL -1 in Er matrix solution of 500µgmL -1. Fig. 8. Matrix mass spectrum of 500µgmL -1 Tm 2 O 3 solution with the resolution setting to 0.3amu As is clear from Table 6, the concentrations with the determination the for monoisotopes of 165 Ho(100) and 169 Tm(100) are 1.72µgL -1 and 0.39µgL -1 respectively in a 500µgmL -1 Er 2 O 3 (99.999%) solution with the resolution value setting to 0.3amu and it is much lower than that obtained respectively with the resolution value setting to 0.7amu by using ELAN 9000 ICP-MS. 139 La Table 6 Results for trace of 14 REEs (µg g -1 ) in high purity Er 2 O 3 (99.999%) 140 Ce 141 Pr 143 Nd 147 Sm 151 Eu 155 Gd 159 Tb Dy 165 Ho 169 Tm 172 Yb 175 Lu 89 Y REEs

14 2005 PerkinElmer ICP-OES ICP-MS 3.5 Relationship between the interfering effects and instrumental conditions By referring back to the parts of 3.1 to 3.4, we can find it is the most serious interference of the MO + generated from matrix-induced polyatomic ions with the producing ratio amount to about from 2.0% to 3.0%. Generally speaking, the concentration of a matrix solution is controlled 500µgmL -1 for the determination of trace amount of REEs in high purity rare earth products by using ICP-MS, so the concentrations of interfering MO + vary from 10µgmL -1 to 15µgmL -1. We must take some measures for reducing these interferences that affect the determination with accuracy of REEs of interest. In this work, it was investigated that the relationship between the intensity of matrix-induced polyatomic ions and the instrumental conditions with a 500µgmL -1 La 2 O 3 solution Relationship between matrix-induced polyatomic ions and RF power In the presence of the resolution value setting to 0.7amu and nebulizer gas flow to 0.9 Lmin -1, adjusting RF power to 800, 900, 1000, 1100 and 1200w respectively, the intensity of the polyatomic ions, such as 139 LaO +, 139 LaOH +, 139 LaOH + 2, 139 LaOH + 3 and 139 LaO + 2 changing with the adjustments was observed with a 500µgmL -1 La 2 O 3 solution. These are given in Figure 9. Intensity (cps) RF Power (w) 139 LaO + / LaOH + / LaOH + 2 / LaOH + 3 / LaO + 2 /10 4 Fig.9. Intensity vs. RF power of polyatomic ions in 500µgmL -1 La 2 O 3 solution It can be seen from Figure 9 that the 139 LaO + intensity (up to x 10 8 cps) continuous decreases with RF power constantly increasing. At the end of RF power equals to 1200w, the 139 LaO + intensity arrives at the most lowest value. In opposition to this, the intensity of 139 LaOH + (up to x 10 7 cps) continuous increases with RF power constantly increasing. When the value of RF power equals to 1100w, the intensity of 139 LaOH + remains approximately constant. The phenomenons of the intensity of 139 LaOH + 2 (up to x 10 6 cps) and 139 LaOH + 3 (up to x 10 4 cps) value with adjusting RF power value are very similar to that of 139 LaO + and 139 LaOH + respectively. However, the intensity of 139 LaO + 2 (up to x 10 4 cps) continuous increases with RF power constantly increasing. When the value of RF power equals to 1000w, the intensity of 139 LaO + 2 reaches the most highest value and then arrives at the most lowest value while RF power equals to 1100w, at the end, the intensity of 139 LaO + 2 continuous increases with RF power constantly increasing. As mentioned above, in a practical analysis procedure, it is favour of decreasing the polyatomic ions interferences of MO + and MOH + 2 with increasing RF power and is also favour of decreasing that of MOH +, MOH + 3 and MO + 2 with decreasing RF power Relationship between matrix-induced polyatomic ions and nebulizer gas flow Nebulizer gas flow is another important factor that affects the polyatomic ions interferences which interfere the determination of analytes of interest by using ICP-MS technique. In case of resolution value setting to 0.7amu and RF power to 1000w individually, adjusting nebulizer gas flow to 0.90, 0.91, 0.92, 0.93, 0.94 and 0.95 Lmin -1 respectively, the intensity of polyamotic ions, such as 139 LaO +, 139 LaOH + 1-3, 139 LaO + 2 change with adjustment of nebulizer gas flow was investigated by using a 500µgmL -1 La 2 O 3 solution. These are shown in Figure 10. From Figure 10 it is apparent that the intensity of polyatomic ions of 139 LaO +, 139 LaOH LaO and 2 continuous increases with nebulizer gas flow constantly increasing. Therefore, it is favour of decreasing the polyatomic ions interferences arose from matrix-induced with decreasing nebulizer gas flow. 10

15 2005 PerkinElmer ICP-OES ICP-MS Intensity (cps) Fig. 10. Intensity vs. Nebulizer gas flow of polyatomic ions in 500ugmL -1 La 2 O 3 solution 3.6 Appearance concentration of polyatomic ions arose from matrix-induced In the presence of setting the resolution value to 0.7amu, RF power to 1000w and nebulizer gas flow to 0.9Lmin -1, we took a quantitative determination of appearance concentrations of polyatomic ions generated from matrix-induced in nine high purity rare earth products by using ELAN 9000 ICP-MS. These results were calculated with the same equation as party 3.3 of equation (2), and then the concentrations are given in Table 7. The results of polyatomic ions arose from matrix-induced of high purity of Y, Ho, Er, Tm, Yb and Lu in a 500µgmL -1 solution are not given in Table 7, the reason for this is that the mass range of polyatomic ions generated from the matrix like that does not cover the mass range of trace of amount REEs which will be measured in this samples. It can be seen from Table 7 that the appearance concentrations of polyatomic ions arose from matrix-induced are from to 17.49µgmL -1. Table 7 Appearance concentration (µgml -1 ) of polyatomic ions generated from matrix (n=3) Matrix Polyatomic ion Interfered Mass Appearance concentration La Ce Pr Nd Nebulizer gas flow (Lmin -1 ) 139 LaO + / LaOH + / LaOH + 2 / LaOH + 3 / LaO + 2 / LaO LaOH LaOH LaOH LaO CeO CeOH CeO CeOH CeOH CeOH CeOH CeO CeO CeOH PrO PrOH PrOH PrOH Pr O NdO NdO NdOH NdO NdOH NdOH NdO NdOH NdOH NdO NdOH NdOH NdOH NdOH NdOH NdO NdOH NdOH NdOH NdOH NdO NdOH NdOH NdOH NdOH NdOH

16 2005 PerkinElmer ICP-OES ICP-MS Sm Eu Gd Tb Dy 144 SmO SmOH SmOH SmO SmOH SmO SmOH SmO SmOH SmOH SmO SmOH SmOH SmOH SmOH SmOH SmOH SmO SmOH SmOH SmOH SmOH SmO SmOH SmOH SmOH SmOH SmOH EuO EuOH EuO EuOH EuOH EuOH EuOH EuOH GdO GdO GdOH GdO GdOH GdOH GdO GdOH GdOH GdOH GdO GdOH GDOH GdOH GdOH GdOH TbO DyO DyO DyOH DyOH LOQs and acquiring purity under direct determination of fifteen high purity rare earth products The LOQs and the acquiring purity under direct determination for trace of REEs as impurities in fifteen high purity rare earth products by using ELAN 9000 ICP-MS technique were presented in this part. The LOQs for trace of REEs impurities are defined as the corresponding concentration values of ten times of the stand deviation (10δ) in eleven measurements of 1% HNO 3 blank solution and with consideration of dilution factors of the samples. On the other hand, the LOQs for those trace of REEs impurities that suffer the overlap interferences arose from isobaric, polyatomic ions and mass spectrum overlaps generated from matrix in some high purity rare earth products under direct determination using ICP-MS are defined as the corresponding concentration values of ten times of the stand deviation (10δ) in eleven measurement of 500µgmL -1 matrix solutions. Higher abundant isotopes of REEs of interest in fifteen high purity rare earth products were also selected along with consideration of isotopes free from the overlap interferences arose from that described as above in case of direct determination using ICP-MS without separation procedure of matrix. The LOQs and acquiring purity under direct determination are given in Table 8. The acquiring purity refers to here the capabilities of the method under direct determination of the trace of REEs impurities in high purity rare earth products by using ICP-MS technique with consideration of total LOQs of all of the 12

17 2005 PerkinElmer ICP-OES ICP-MS REEs impurities. The values are calculated as the following equation: 100-(Σ LOQ x 10-6 ) 100% Where : Σ LOQ x refers to total LOQs for the fourteen trace of REEs impurities in a high purity rare earth product 10-6 refers to the coefficient of unit for µg converts as g 4. Conclusions A series of mass spectrum interferences, the appearance concentrations of overlap interferences and polyatomic ions arose from matrix-induced and overcoming these interfering problems et al. in the determination of trace of amount of REEs as impurities in fifteen high purity rare earth products by using ELAN 9000 ICP-MS technique were reported in this paper. No influence of mass spectrum overlap interferences of polyatomic ions arose from lasma- and matrix solvent-induced that can affect the accurate determination of REEs in high purity products because the mass rage of these interfering ions is less than 80 Isobaric overlap interferences between some REEs can be overcame by selecting isotopes that free from overlap interference or establishing correction equations. The interferences caused by matrix-induced polyatomic ions can be partly overcame by using the method of matrix-match, but separation of matrix is the most efficient method to avoid these interferences. The overlap interferences caused by matrix can be resolved by selecting isotopes that free from overlap interference or by separation of matrix, but through adjusting the resolution value is a simple method to avoid these interferences sometimes. Table 8 LOQ (µg g -1 ) and acquiring purity (%) under direct determination of fifteen high purity rare earth products Matrix La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Y Analyte Isotope La --- LOQ Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ Isotope LOQ

18 2005 PerkinElmer ICP-OES ICP-MS Yb Lu Y Isotope LOQ Isotope LOQ Isotope LOQ ΣLOQ x Acquiring purity 6N 4N 4N 2N 3N 3N 2N 2N 4N 6N 4N 6N 4N 6N 6N Note: 2N refers to the purity for 99% of a high purity rare earth product, 3N for 99.9%, 4N for 99.99% and 6N for %. --- Acknowledgements This work was supported by the important item [No. (2004) 210] of Jiangxi provincial Department of Science and Technology. Reference [1] C. R. Lalonde, J. I. Dalton, Can. J. Spectrosc. 26 (1982) 163. [2] E. S. Gladney, N. W. Bower, Geostand. Newsl. 9 (1985) 261. [3] J. G. Crock, F. E. Lichte, Anal, Chem. 54 (1982) [4] S. S. Biswas, R. Kaimal, A. Sthumadhavan, P. S. Murty, Anal. Lett. 24 (1991) [5] N. Daskalova, S. Velichkov, N. Krasnobaeva, P. Slavova, Spectrochim. Acta 47(1992) E1595. [6] R. S. Houk, V. A. Fassel, G. D. Flesch, H. J. Svec, A. L. Gray, C. E. Taylor, Inductively coupled argon plasma as an ion source for mass spectrometric determination of trace elements, Anal. Chem. 52 (1980) [7] K. E. Jarvis, A. L. Gray, R. S. Houk, Handbook of inductively coupled plasma mass spectrometry, Blackie, Glasgow, [8] C. J. Park, G. E. M. Hall, Analysis of geological matericals by inductively coupled plasma mass spectrometry with sample introduction by electrothermal vaporization, J. Anal. Aton. Spectrom, 3 (1998) [9] K. G. Heumann, Trace determination and isotope analysis of the elements in life sciences by mass spectrometry, Biomed, Mass spectrum, 12 (1985) [10] J. L. Liu, Y. D. Tong, X. Q. Zhang, Determination of trace impurities of rare earth elements in high purity yttrium oxide by ICP-MS, J. Rare Earths. 13 (1995) [11] X. Q. Zhang, Y. D. Tong, J. L. Liu, Determination of trace rare earth elements in high purity cerium oxide by ICP-MS, Fenxi Ceshi Xuebao, 17 (1998) [12] X. Q. Zhang, J. L. Liu, Y. M. Jiang, Y. Yi, T. D. Tong, P. Lin, Determination of impurities in high purity samarium by ICP-MS, Fenxi Ceshi Xuebao, 24 (2005) [13] W. R. Pedreira, J. E. S. Sarkis, C. Rodrigues, I. A. Tomiyeshi, C. A. da siva Queiroz, A. Abrao, Determination of trace amounts of rare earth elements in high pure praseodymium oxide by double focusing inductively coupled plasma mass spectrometry and high-performance liquid chromatography, J. Alloys and compounds, (2001) [14] W. R. Pedreira, J. E. S. Sarkis, C. A. da siva Queiroz, C. Rodrigues, I. A. Tomiyeshi, A. Abrao, Determination of trace amounts of rare-earth elements in high pure neodymium oxide by ICP-SFMS and HPLC techniques. J. Solid sate Chemistry, 171 (2003) 3-6. [15] M. A. Vaughn, G. Horlick, Oxide, hydroxide and doubly charged species in inductively coupled plasma mass spectrometry. Appl. Spectrosc, 40 (1986) [16] A. L. Gray, Mass spectrometry with an inductively coupled plasma as an ion source: the influence on ultra trace analysis of background and matrix response. Spectrochim, Acta. 41B (1986) [17] S. H. Ta, G. Horlick, Background spectral features in inductively coupled plasma mass spectrometry. Appl. Spectrosc. 40 (1986) [18] X. Q. Zhang, J. L. Liu, Y. M. Jiang, Y. Q. Suo, Y. L. Liu, X. Li, Determination of trace REEs and Non-REEs impurities in high purity lanthanum oxide by ICP-MS, J. Chinese mass spectrometry society.25 (2004) [19] M. Yin, B. Li, T. F. Fu, Determination of rare earth impurities in high purity Sc 2 O 3, Y 2 O 3 Ho 2 O 3, Tm 2 O 3 and Lu 2 O 3 by ICP-MS, J. Analytical science. 11(1995) [20] B. Li, Y. Zhang, M. Yin. Determination of trace amount of rare earth elements in high-purity cerium oxide by ICP-MS after separation by solvent extration, Analyst. 122 (1997) 543~547. [21] X. Q. Zhang, J. L. Liu, Y. Yi, Y. M. Jiang, P. Lin, Y. D. Tong, Determination of trace impurities of trace impurities in high purity ytterbium by ICP-MS, Fenxi Ceshi Xuebao, 23 (2004)

19 2005 PerkinElmer ICP-OES ICP-MS Direct Determination of Rare Earth Impurities in High Purity Ytterbium Oxide by Inductively Coupled Plasma Mass Spectrometry Xinquan Zhang*, Jinglei Liu, Xiang Li, Yonglin Liu, Yong Yi, Yumei Jiang, Yaqin Su Jiangxi Institute of Analyzing and Testing, Beijing East Road 171, Nanchang , Jiangxi Province, China Abstract A novel methodology was developed for the direct determination of trace quantities of rare earth elements (REEs) in high purity ytterbium oxide by inductively coupled plasma mass spectrometry (ICP-MS) in this work. The mass spectra overlap interferences arose from Yb matrix on the neighbouring analytes of 169 Tm(100) and 175 Lu(97.4) were eliminated by adjusting instrumental peak resolution value from 0.7amu to 0.3amu. The matrix suppression effect of Yb on the ion peak signals of REEs impurities was compensated effectively for the addition with internal standard as In. The limit of quantitation (LOQ) of REEs impurities was from µgg -1 to 0.022µgg -1, the recoveries of spiked sample for REEs were found to be in the rage of 91.2%~105% through the use of proposed method and relative standard deviation (RSD) varied between 3.1% and 7.2%. The novel methodology had been found to be suitable for the direct determination of trace REEs impurities in %~ % high purity Yb 2 O 3. Keywords: Direct determination; Rare earth impurity; High purity ytterbium oxide; Inductively coupled plasma mass spectrometry 1. Introduction With the development of widely applications of high purity rare earth products in many material fields, such as phosphor materials in a display, gain media in a solid state laser, optomagnetic materials in an optical isolator and high temperature superconductor, etc., the REEs are increasing their importance in many industrial fields. Therefore, a sensitive, rapid and reliable analytical method for trace analysis of REEs impurities in high purity rare earth products is strongly required for quality control, product certification and evaluation of material performance. After Houk et al. [1] first published their methodology by using inductively coupled plasma (ICP) as an ion source combination with mass spectrometer (MS), twenty five years have passed. Today, this analytical technique regarded as a principal measure is widely applied in the analytical fields such as high purity materials, environment, metallurgy, biochemistry et al. due to its attractive and excellent performance in terms of high sensitivity, low determination 15

20 2005 PerkinElmer ICP-OES ICP-MS limits, determination capability of multielements at trace or ultra-trace at the same time and wide dynamic rage compared with other analytical techniques such as inductively coupled plasma atomic emission spectrometry (ICP-AES), so, it becomes the major analytical method used for the determination of trace of REEs impurities in high purity rare earth products day by day [2-9]. Although ICP-MS technique possesses protrusive advantages described as above, some major problems encountered in the application of this measure to the determination of trace REEs in high purity rare earth products are the matrix suppression effect and the mass spectra interferences arose from isobaric overlap, polyatomic ions and overlap interferences generated from parent matrix except for the measurement of trace of REEs impurities in high purity Y, Ho, Tm, Lu products employed with ICP-MS technique [5, 7]. The matrix suppression effect could be compensated for adding internal standard [2, 5, 6], but these overlap interferences may restrict its direct applicability for the determination of trace of REEs impurities in high purity rare earth products. From the literature reports, many attempts had been made to solve these interferences with the procedure of separation of trace quantities of REEs impurities from the sample matrix. Pedreira et al [10,11] investigated the determination of rare earth analytes as impurities in high purity La 2 O 3 and Pr 6 O 11 with ICP-MS combined with high-performance liquid chromatography(hplc) technique after parent matrix as La and Pr had been separated respectively by HPLC. Otherwise, the techniques of ion chromatography (IC) and eletrothermal vaporation(etv) combined with ICP-MS were employed for separation of the trace of REEs impurities from parent matrix to overcome these overlap interferences[12,13]. In case of %~ % high purity Yb 2 O 3 analytical procedure employed with ICP-MS, overlap interferences generated from Yb matrix lead to the failure in direct measurement of Tm, Lu. Several papers evaluated for the determination of trace of REEs in high purity Yb 2 O 3 using ICP-MS technique. Cai et al. [14] determined thirteen trace quantities of REEs in high purity Yb 2 O 3 using ICP-MS based on Ga internal standard, but the important analyte value of Lu was not reported because of the mass spectra overlap interferences arose from parent matrix of Yb. Zhang et al. [15] accomplished recently with the determination of 14 REEs as impurities in % high purity metal Yb using ICP-MS technique through with making a direct measurement for the analytes as La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er combination with separating Yb matrix by using 2-ethlhexyl hydrogen-2-ethylhexylphosphonate (EHEHP) as solvent extraction for the determination of analytes of Tm and Lu. However, in these multi-stage combined procedures, a risk of increasing contamination and worsening LOQs may be encountered. It can also spend a long time and is difficult to grasp sometimes, because the concentration large difference between matrix (1000µgmL -1 ) and trace quantities of REEs impurities (0.00xµgmL -1 ) can amount to about or more. Direct determination of trace quantities of REEs impurities in high purity rare earth products has the advantages of minimizing sample preparation, reducing the potential for sample contamination and simplest manipulation comparing with those procedures described as above of separation parent matrix. In this work, with the aid of adjusting the instrumental peak resolution, the overlap interferences arose from Yb matrix on the neighbouring analytes of Tm and Lu were overcame, it was realized that the direct determination of trace quantities of 14 REEs in %~ % high purity Yb 2 O 3 employed with ICP-MS technique, to our best knowledge, this was not reported. 2. Experimental 2.1 Instrument A Perkin-Elmer Sciex ELAN 9000 ICP-MS equipped with a integrated peristaltic pump and a cross flow nebulizer combined with a scott spray chamber which was made from Ryton material was used. The instrument conditions are summarized in Table 1. 16

21 2005 PerkinElmer ICP-OES ICP-MS Table 1 Operating condition of ELAN 9000 ICP-MS RF Power (w) 1000 Resolution (amu) 0.30~0.70 Nebulizer Gas Flow (Lmin -1 ) 0.90 Detector dual Lens voltage (v) 6.0 Speed of Peristaltic Pump (rpm) 26 Analog Stage Voltage (v) Sweeps/Reading 3 Pulse Stage Voltage (v) 1000 Replicates 3 Discriminator threshold (v) 70 Scan Mode Peak hopping Ac rod offset (v) Reagents The superpure HNO 3 used in this work was acquired by the analytical-reagent grade HNO 3 under sub-boiling condition. H 2 O 2 was the AR grade. Sample solutions were prepared with purified water from a TYUV05 water unit with the resistivity being more than 18.3M cm ( Tao Yuan Company, Nanchang, China). 2.3 Standard calibration solution preparation Specpure rare earth standard samples which were purchased from Johnson Matthey Materials Technology Company (Orchard, Royston, England) were used to prepare the mixed standard calibration solutions containing all REEs except Yb 2 O 3, the concentrations for the mixed standard solutions were 5.0µgL -1 and 10µgL -1 respectively, each of the standard solutions was spiked with In as internal standard, the concentration of In was 10µgL -1, the final acid concentration being adjusted to 1% HNO 3 (V/V). 2.4 Sample preparation Two samples (with purity of % and %, respectively) of high purity Yb 2 O 3 which were acquired from Hunan Insitute of Rare Earth Material (Hunan, China) were weighted (0.1000g) into 100mL beakers individually, and decomposed with ultrapure water 10mL and superpure HNO 3 2mL on a hot plate which heater at about mL H 2 O 2 was added in addition when Yb 2 O 3 was decomposed. After the dissolving solution cooled, transferred it into a 100mL volumetric flask individually and finally diluted with ultrapure water, so the quality concentrations of Yb 2 O 3 in solutions were 1.0gL -1 and the solutions were all spiked with In as internal standard with the quality concentrations of 10µgL -1. prepared blank solution at the same time. Prepared another four solutions of the two samples of %~ % high purity Yb 2 O 3 as same as the procedure described as above, respectively. These solutions were all spiked with 14 REEs and the quality concentrations of 14 REEs were 1.0µgL -1 and 2.0µgL -1 individually. In was also spiked with concentrations of 10µgL -1. Prepared four solutions using % high purity of Yb 2 O 3, the Yb 2 O 3 matrix concentrations were 300, 500, 700, 1500µgmL -1, respectively. And these solutions were also spiked with 14 REEs and the quality concentrations were 2.0µgL -1, spiked In as internal standard with quality concentrations being 10µgL -1. All of the beakers and volumetic flasks used in this work were carefully washed with 10% HNO 3 (V/V) and then washed repeatedly with ultrapure water in order to reduce blank value. 2.5 ICP-MS determination In the analytical procedure for the direct determination of trace quantities of 14 REEs impurities in high purity Yb 2 O 3 employed with ELAN 9000 ICP-MS techniques, setting the instrumental peak resolution value to 0.7amu for the measurement of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er and that to 0.3amu for the determining of Tm and Lu. The solutions of 1.0gL -1 Yb 2 O 3 spiked with 14 REEs of 1.0µgL -1 and 2.0µgL -1 respectively were used to take the recovering experiments. The matrix solutions with concentrations of 300, 500, 700, 1000, 1500µgmL -1 individually were used to take matrix suppression effect experiments. Otherwise, higher abundant isotopes of the analytes were selected along with consideration of isotopes free from the overlap interferences of isobaric. 17

22 2005 PerkinElmer ICP-OES ICP-MS 3. Results and discussion 3.1 Mass spectra interferences arose from Yb matrix There are many works have evaluated the interferences generated from isobaric overlaps, polyatomic ions and overlap interferences arose from matrix-induced in case of determination of trace of REEs impurities in high purity rare earth products using ICP-MS technique [5, 6, 7, 10, 11, 14, 15]. Yb has seven naturally occurring isotopes, such as 168 Yb(0.14), 170 Yb(3.0), 171 Yb(14.3), 172 Yb(21.9), 173 Yb(16.2), 174 Yb(31.8) and 176 Yb(12.7). Due to its mass range varying from 168 to 176, the mass range of polyatomic ions as YbO +, YbOH and YbO + 2 arose from Yb matrix combination with oxygen and hydrogen is from 184 to 208, that doesn t cover on the mass range of the REEs which we will determine using ICP-MS in the analytical procedure. It means that there is no influence of polyatomic ion interferences originated from Yb matrix that can affect the determination of REEs impurities in high purity Yb 2 O 3 using ICP-MS with precision and accuracy. However, it exists another interferences arose from Yb matrix overlap, such as 169 Tm(100) suffering the overlap interference from 168 Yb(0.14) and 170 Yb(3.0), 175 Lu(97.4) also from 174 Yb(31.8) and 176 Yb(12.7). Experimental results present that the interferences can produce about 5µgL -1 and 10µgL -1 overlap concentrations on the analyte of 169 Tm(100) and 175 Lu(97.4) respectively with the matrix solution of 1.0gL -1 Yb 2 O 3. Adjusting the instrumental peak resolution value can correct the overlap interferences originated from Yb matrix on the analytes of 169 Tm(100) and 175 Lu(97.4). Table 2 presents the results of Tm, Lu acquired from setting the peak resolution values from 0.7amu to 0.3amu using ELAN 9000 ICP-MS. Table 2 Results (µgg -1 ) of Tm and Lu obtained under different resolution value (amu) with % 1.0gL -1 Yb solution (n=3) Resolution Analyte Tm(100) Lu(97.4) It is evident from Table 2 that the results of Tm and Lu continuous decrease with the peak resolution values continuous increase. When the peak resolution value equal to 0.3amu, the results of 169 Tm(100) and 175 Lu(97.4) are 0.21µgg -1 and 0.26µgg -1 individually. This agree with the % purity value of high purity Yb 2 O Matrix suppression effect and internal standard Matrix suppression effect is another main problem in high purity rare earth products analysis employed with ICP-MS [2, 5, 6, 7, 10, 11, 14, 15]. This signal suppression effects are regarded significantly as the factor of space charge effect [16]. In this paper, six solutions with Yb 2 O 3 matrix concentrations varying from 0, 300, 500, 700, 1000 and 1500µgmL -1 spiked with 14 REEs of 2.0µgL -1 were used to observe the matrix suppression effects. The relationship between the intensity of some REEs and the matrix concentrations are given in Figure 1. Intensity (cps) Y + (*10 4 cps ) 161 Dy + ( * 10 4 cps ) Matrix concentration (ugml -1 ) 140 Ce + (*10 4 cps ) 151 Eu + ( * 10 4 cps ) 115 In + ( * 10 5 cps ) Figure 1. Intensity vs. matrix concentration 18

23 2005 PerkinElmer ICP-OES ICP-MS It can be seen from Figure 1 that the intensity of analyte of Y, Ce, Eu, Dy continuous decrease with the matrix concentration continuous increase. When the matrix concentration equals to 500µgmL -1, the intensity of those analytes remains approximately constant. The phenomenon of In intensity changing with Yb matrix is very similar to that for Y, Ce, Eu and Dy. In this work, sample dilution factors of 1000 were used for direct determination of REEs impurities, we have noticed that the signal intensity decreasing ratios for the analytes were from 60% to 70%. However, the signal intensity decreasing ratio for In was also from 60% to 70%, so, using In as a internal standard can effectively compensate the matrix suppression effect. 3.3 Recovery test Using the four solutions of 1.0gL -1 Yb 2 O 3 (99.999%~ %) spiked with 1.0µgL -1 and 2.0µgL -1 respectively of 14 REEs, following the sample treatment procedure, whole procedure recoveries for all the interest analytes were determined. The results are listed in Table 3. By referring to Table 3, we can know that recoveries of all the 14 REEs are satisfactory being from 91.2% to 105%. Thus, according to Table 3, trace quantities of 14 REEs impurities in high purity Yb 2 O 3 with purity of % and % could be accurately and direct determined using ELAN 9000 ICP-MS with the proposed method. Table 3 Spiking recovery (%) test Analyte % Yb 2 O 3 solution(1.0gl -1 ) % Yb 2 O 3 solution(1.0gl -1 ) 1.0µgL µgL µgL µgL Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Lu Precision and limit of quantitation (LOQ) According to the IUPAC definition, the LOQ is defined as the corresponding concentration values of ten times the standard deviation (10δ) in eleven measurements of 1% HNO 3 (V/V)blank solution and with consideration of dilution factors of the samples. The LOQ values are summarized in Table 4. As listed in Table 4, LOQs for all of 14 REEs impurities were from µgg -1 to 0.022µgg -1, that can fully satisfy the requrement for analysis of %~ % high purity Yb 2 O 3, the relative standard deviation (RSD) for all of 14 REEs impurities vary from 3.1% to 7.2%. Table 4 Method precision (%) and limit of quantitation (µgg -1 ) Analyte La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Lu Y LOQ RSD

24 2005 PerkinElmer ICP-OES ICP-MS 3.5 Sample analysis We direct determined the trace quantities of 14 REEs impurities in % and % high purity Yb 2 O 3 samples using ELAN 9000 ICP-MS with the established method. Analytical results are listed in Table 5. It is evident from Table 5 that the total of 14 REEs impurities were low 10µgg -1 and 1.0µgg -1 individually. Table 5 Analytical results (µgg -1, n=3) Analyte La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Lu Y 1 # # Conclusion A novel, rapid and reliable methodology for direct determination of trace of 14 REEs impurities in %~ % high purity Yb 2 O 3 employed with ELAN 9000 ICP-MS was established. The mass spectra overlap interferences arose from Yb matrix that affect the measurements of Tm and Lu with precision and accuracy were eliminated through selecting different peak resolution values according to different analytes of interest. In was spiked as a internal standard element to compensate effectively the Yb matrix suppression effects. The proposed method is especially useful for direct determination of trace of REEs impurities in Yb 2 O 3 samples with purity of %~ %. Acknowledgements This work was supported by the important item [No. (2004) 210] of Jiangxi Provincial Department of Science and Technology, Republic of China. References [1] R. S. Houk, V. A. Fassel, G. D. Flesh, H. J. Svec, A. L. Gray, C. E. Taylor, Anal. Chem. 52 (1980) [2] K. E. Javis, A. L. Gray, R. S. Houk, Handbook of inductively coupled plasma mass spectrometry, Blankie, Glasgow, [3] D. J. Douglas, E. S. K. Quan and R. G. Smith, spectrochim Acta 38B (1983) [4] H. T. Delves and M. J. Campbell, J. Anal. Atom. Spectrom, 3 (2) (1998) [5] J. L. Liu, Y. D. Tong, X. Q. Zhang, J. Rare Earths, 13 (1995) [6] X. Q. Zhang, J. L. Liu, Y. M. Jiang, Y. Q. Su, Y. L. Liu, X. Li, J. Chinese mass spectrometry society, 25 (2004) [7] M. Yin, B. Li, T. F. Fu, J. Anal. Science, 11 (1995) [8] C. J. Pickford and R. M. Brown, Spectrochim Acta 41 B (1986) [9] V. K. Panday, J. S. Becker, Fresenius, J. Anal. Chem., 352 (1995) 327. [10] W. R. Pedreira,J.E. S. Sarkis, C. Rodrigues, I. A. Tomiyoshi, C. A. da Silva Queiroz, A. Abrão, J. Alloys and Compounds, 344 (2002) [11] W. R. Pedreira, J.E.S. Sarkis, C. Rodrigues, I. A. Tomiyoshi, C. A. da Silva Queiroz, A. Abrão, J. Alloys and Compounds, 323~324 (2001) [12] D. S. Braverman, J. Anal. At. Spectrom., 7 (1992) 43. [13] N. Shitaba, M. Fudagawa, M. Kubota, Anal. Chem., 63 (1991) 636. [14] S. Q. Cai, N. Zhang, X. S. Liu and P. An, J. Anal. Science, 11 (1995) [15] X. Q. Zhang, J. L. Liu, Y. Yi, Y. M. Jiang, P. Lin, Y. D. Tong, Fenxi Ceshi Xuebao, 23 (2004) [16] J. A. Olivares and R. S. Houk, Anal. Chem., 58 (1986)

25 2005 PerkinElmer ICP-OES ICP-MS 杨 振 宇 ICP-MS 7 As Pb Ni Cd Cr Sb Zn 80% 125% r CV <10% ICP-MS 6 mg/kg Na Ca Mg Si µg/kg Hg Inductively Coupled Plasma Mass Spectrometry ICP-MS ICP-MS 1 HPLC HPCE 21

26 2005 PerkinElmer ICP-OES ICP-MS GC ICP-MS 1 Table 1 National sanitary and analytical standards of elements lipped from some kinds of food containers GB Pb As Cr Cd Ni GB/T GB Pb Cd GB/T GB Pb Cd Sb GB/T GB Pb As Cr Zn GB/T Milli-Q Millipore 2~3 ELAN6000 ICP-MS Perkin-Elmer a SCIEX 10% 3 b ml MΩ cm Zn Cr Ni As Cd Sb Pb 3 ICP-MS ICP-MS 1000 mg/l Ba Cd Co Cu Ge Pb 4 5 Mg Rh Sc Tb Tl 10 µg/l Perkin-Elmer a 4% % 4% 1 b 4% v/v 2 GB/T c

27 2005 PerkinElmer ICP-OES ICP-MS ICP-MS ICP-MS ICP-MS Table 2 Instrumental conditions and data acquisition parameters of ICP-MS RF 1300 W 1.0 ml/min 1.00 L/min 0.85 L/min 20 L/min W 1 (18±1) Ni 3 3 Table 3 Selected isotopes mass and internal material /% /ms Zn Cr Ni As Cd Sb Pb Cd As 10 µg/ml 20 µg/ml 1000 µg/ml GB Table 4 Specific sanitary standards of the elements lipped from food containers (mg/l) GB/T Pb<2.5 Cd<0.25 As<0.5 Pb<7.0 Cd<0.5 Sb<0.7 GB/T Pb<7.0 Cd<0.5 Sb<0.7 GB/T Zn<1.0 Cd<0.02 As<0.04 Pb<0.2 GB/T Pb<1.0 Cd<0.02 Cr<0.5 Ni<3.0 As<0.04 Pb<1.0 Cd<0.02 Cr<0.5 Ni<1.0 As<

28 2005 PerkinElmer ICP-OES ICP-MS 5 Table 5 Elements concentration of the standard solutions ρ / mg L -1 ρ/ µg L -1 Zn Cr Ni As Cd Sb Pb ICP-MS 2.2 ICP-MS Ar 35 Cl + 75 As 5 75 As Se Se ICP-MS GB/T ICP-MS 5% HNO 3 v/v ICP-MS ICP-MS ICP-MS

29 2005 PerkinElmer ICP-OES ICP-MS 6 Table 6 Final results of the elements content in the different concentrations of acetic acid ρ/ µg L -1 As Pb Ni Cd Cr Sb Zn % % % % % % As Pb Ni Cd Cr Sb As As Ni Cd Zn Zn y=ax+b 1 Fig.1 Response of As, Ni, Cd, Zn in difference r>0.999 concentrations of acetic acid Pb Cr Sb 2 Fig.2 Response of Pb, Cr, Sb in difference concentrations of acetic acid

30 2005 PerkinElmer ICP-OES ICP-MS 7 Table7 Elements concentration of samples ρ/ µg L -1 As Pb Ni Cd Cr Sb Zn ~ %~125% Table 8 8 Zn Results of recovery in porcelain enamel Cr / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni / / / µg L -1 µg L -1 µg L -1 /% Ni As Cd Sb Pb As Cd Sb Pb / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb Pb Table 9 Results of recovery in aluminium / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb Pb

31 2005 PerkinElmer ICP-OES ICP-MS / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb Pb / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb Pb / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb Pb Pb %~125% 9 80%~120% Table 11 Results of recovery in glass Table 10 Results of recovery in stainless steel / / / µg L -1 µg L -1 µg L -1 /% / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb Pb Cr Ni As Cd Sb Pb

32 2005 PerkinElmer ICP-OES ICP-MS / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Sb Pb / / / µg L -1 µg L -1 µg L -1 /% Zn Cr Ni As Cd Table 12 Detection limit of method µg/l ρ/ µg L -1 Zn Cr Ni As Cd Sb Pb Sb Pb CV 80%~120% 10% 13 Table 13 Results of precision experiment of porcelain enamel ρ / µg L -1 CV/% ρ / µg L -1 CV/% ρ / µg L -1 CV/% Zn Cr Ni As Cd Sb Pb

33 2005 PerkinElmer ICP-OES ICP-MS 3 ICP-MS As Cr 4% CV 10% 4 [1] K.E... Zn As [M]. Cr %~120% Zn [2] A.R., A.L. Zn. [M]. Zn Determination of Multi-elements in Food Container using a Rapid Analytical Method by ICP-MS YANG Zhen-yu (Shanghai Imp.& Exp. Commodity Inspection & Quarantine Bureau Shanghai , China) Abstract: A rapid analytical method for the simultaneous determination of multi-elements in food container was established by ICP-MS. The samples such as porcelain enamel, chinaware, aluminium, stainless steel, glass were dipped in by 4% acetic acid. The proposed method was applied to the analysis of seven elements in five food containers. The recoveries were from 80% to 125%, and linear correlation r>0.999, and the precision CV<10%. The method has the advantages of excellent sensitivity, precision, accuracy and simplicity of sample pretreatment, and the interferences were low. It can meet demands of concerning standards for quality analysis. Key words inductively coupled plasma-mass spectrometry (ICP-MS) food container elements analysis. 29

34 2005 PerkinElmer ICP-OES ICP-MS ICP-MS 欧 阳 荔 1, 王 小 燕 1, 解 清 1, 刘 雅 琼 1, 王 京 宇 1,*, 张 卫 2, 李 竹 ICP-MS ICP-MS ICP-MS [1~4] B 1.2 1:1 ICP-MS

35 2005 PerkinElmer ICP-OES ICP-MS MWS-2 Perkin-Elmer ELAN 3 mm DRC ICP-MS 0.9 L/min 1.8 L/min 15L/min Rh 2.3 GBW a %Triton TX % h 75 mg 1.5 ml BV-III ml ml W 10 min W 15 min W 10 min Berghof 1 GBW09101 a n=6 ρ/ mg kg -1 ρ/ mg kg -1 P As Cr Y Mn Mo Fe Ag Ni Cd Cu La Sr Pr Ba Nd Pb Sm Zn Gd Mg Dy Hg Co

36 2005 PerkinElmer ICP-OES ICP-MS AAS AFS Zn 88 As Hg Cd 97 [ = / *100%] % % Zn As Hg Cd % Zn As Hg Cd Zn As Hg Cd

37 2005 PerkinElmer ICP-OES ICP-MS 3 29 µg g -1 Al Y P Mo Cr Ag Mn Cd Fe Cs Ni La Cu Pr Sr Nd Ba Sm Pb Gd Zn Dy Mg Tl Li U Co Hg As ng g -1 Al Y P Mo Cr Ag Mn Cd Fe Cs Ni La Cu Pr Sr Nd Ba Sm Pb Gd Zn Dy Mg Tl Li U Co Hg As Mn Ni Sr Pb Co Cd Ag 33

38 2005 PerkinElmer ICP-OES ICP-MS 5 Al Y P Mo Cr Ag Mn Cd Fe Cs Ni La Cu Pr Sr Nd Ba Sm Pb Gd Zn Dy Mg Tl Li U Co Hg As Al As Tl Y Pr Nd Sm Gd Dy 20% Mn Ni Sr Cs Mn Ni Ba Co Cd Cs Hg 20% Cs Gd Ni Sr 5 [1]. Ni Sr Mg Tl U [2].. Mn Ni Co As Ag Hg [3]. Sr Mg Li Tl Cs Tl U As Cd Hg Tl [4] :

39 2005 PerkinElmer ICP-OES ICP-MS -ICP-OES 孙 启 文, 周 标 ICP-OES 0.03 mg/l 96%~103% 1.54% LABCONCO 18.2 M Ω cm g/l RF 1.3 kw 15 L min 0.2 L/min 0.8 L/min 1.5 ml/min 30 s 2 s g 105 -ICP-OES 6 ml 1:1 HCl 2 ml 1:1 HNO 3 1mL HF 10 ml Perkin-Elmer Optima ICP WX-3000 [1] % Water Pro PS 35

40 2005 PerkinElmer ICP-OES ICP-MS nm ~5 mg/l mg/l g 800 W 10 min [2] ICP 2.5 mg/l 2 2 1%~10% v/v 10% v/v 10% 1 ρ /(mg L -1 ) n=8 RSD/% ρ/(mg L -1 ) /% ICP-OES 3 FAAS 36

41 2005 PerkinElmer ICP-OES ICP-MS 3 ρ Na/(mg L -1 ) -ICP-OES FAAS FAAS 4 -ICP-OES 3 [1]. [J] ICP-OES [2] M JN. ICP [M] mg/l 44. Determination of Impurity of Sodium in Catalyst of Coal Liquifaction by Microwave Digestion and ICP-OES SUN Qi-Wen ZHOU Biao (Yankuang Energy R&D CO. LTD. Shanghai China) Abstract: Determination of impurity of Na in the catalyst of coal liquifaction by microwave digestion and ICP-OES is described. The detection limit for Na is 0.03 mg/l The recovery is 96%~103% and The relative standard deviation is 1.54%. The result is satisfactory. Key words: microwave digestion; ICP-OES; catalyst; coal liquifaction; Na 37

42 2005 PerkinElmer ICP-OES ICP-MS ICP-AES 刘 崇 华, 钟 志 光, 卞 群 洲, 卢 焯 冬, 贺 柏 龄 As Al Bi Cd Cu Fe Pb Se Sb Zn 92%-105% RSD 1.5% (Echelle) Scott Meinhard, SCD [1] 0.5~10 s GB [2] Pb Cu Zn 1.2 AAS A.R. [3] ICP-AES Cd [4] As S Sb [5] 1.3 ICP-AES As Al Bi Cd Cu Fe Pb Se Sb Zn Perkin-Elmer OPTIMA XL As 10 µg/ml Al Bi Cd Cu Fe Pb Se Sb Zn ICP-AES g 50 ml 5mL ICP-AES, ~3 1 10% µg Perkin-Elmer OPTIMA 3000XL /ml 1000 µg/ml WinLab 38

43 2005 PerkinElmer ICP-OES ICP-MS Examine 1 1 As Bi Cd Cu Fe Pb Se Sb Zn Al λ/nm RF 900~1400 W 1300 W 1300 W 1.0~0.4 L/min 0.4 L/min As Bi Cd Pb 0.6 L/min % 1.0 L/min ml/min As Bi Cd Cu Fe Pb Se Sb Zn Al (µg/ml) ρ/(µg. ml -1 ) /% RSD/% As Bi Cd Cu Fe Pb Se Sb Zn Al

44 2005 PerkinElmer ICP-OES ICP-MS BY As Bi Cd Cu Fe Pb Se Sb Zn Al % % GB ~ [1] :, [4]. ICP-AES [2].. GB 728 [5]. ICP-AES 84., [3] Direct Determination of As, Al, Bi, Cd, Cu, Fe, Pb, Se, Sb, Zn in Tin Ingot by ICP-AES LIU Chong-hua, BIAN Qun-zhou, LU Zhuo-dong, HE Bai-ling, ZHONG Zhi-guang Guangdong Import & Export Commodity Inspection Bureau,Guangzhou,510623, China Abstract ICP-AES was used for the direct determination of As, Al, Bi, Cd, Cu, Fe, Pb, Se, Sb, Zn in Tin ingot. The matrix effect at was studied and the preferable experimental conditions are investigated. The recovery rates of this procedure are between 92% and 105%. The RSD is within 1.5%. The proposed method is simple and rapid with lower determination limits than chemical method. Key words ICP-AES; tin ingot; SCD; matrix effect 40

45 2005 PerkinElmer ICP-OES ICP-MS ICP-OES 沈 金 根, 李 波, 顾 元 溥 ICP-OES PE Optima 2000 DV Perkin-Elmer Optima y=x x y DV ~22 85% x 1.3 U x 1 x GSB(G) g/l P=95% K=1.96 U x x U x ~5.00 mg/l 1 x U(x 1 ) Mn( nm) 1.5 Mn r

46 2005 PerkinElmer ICP-OES ICP-MS 1 x i(mg/l) y i(i) y(i) r ρ/(mg. L -1 ) 2 /% /% n K=3 1/2 ( x x ) /( n 1 ) S= n-1 / x U rel (x 1 ) /K 2.9 /3 1/2 =1.7% V 1 =1/2[ U rel (x 1 )/ U rel (x 1 )] -2 [ U rel (x 1 )/ U rel (x 1 )] 0.1 V 1 =50 m 1 2 S p = S j = m 1.2% j= 1 2 x U(x 2 ) V 10 4 n-1 42 S p mv V i =9 8=72J=1

47 2005 PerkinElmer ICP-OES ICP-MS x x 3 U rel (x 2 ) Sp/3 1/2 =0.7% V V=n-1=9 ρ/(mg. L -1 ) n-1 S/% [ U rel (x 3 )/ U rel (x 3 )] 0.2 x U(x 3 ) V 3 = x x 0.3 U rel (x) = P=95 K 1.96 U rel ( x1) + U rel ( x 2) + U rel ( x3) (0.3/1.96)% 0.153% = =1.9% 1.7% + 0.7% % 100 ml x V x 50 A /3 1/ x U c rel = P i U rel (x)=1.9% /3 1/2 3 x V eff =V x =50 U rel (x 3 ) 4 = % =0.24% % % 2 43

48 2005 PerkinElmer ICP-OES ICP-MS 5 U rel(x i) U rel(x i) P i P i U (x) V i x U rel(x) U rel(x 1) % U rel(x 2) 0.7% 72 U rel(x 3) 0.24% P=95% 8 V eff =50 t K p =t 95 (50)=2.00 U 95,rel = t 95 (50)U c,rel = %=3.8% 7 U 95,rel 3.8 V eff =50 K p =2.00 [1].. [2].. 44

49 2005 PerkinElmer ICP-OES ICP-MS 宋 晓 年, 冯 天 培 Al-1%Si Al-1%Si 4M DRAM Al-1%Si 16M DRAM M DRAM Al-1%Si GB/T [1] -1% 1 GB 1% 4~ ~ [2] 1% % 1 % Al Si Fe Cu Al-1%Si 0.85~ GB ~ % 45

50 2005 PerkinElmer ICP-OES ICP-MS Optima 3000 Perkin-Elmer -1% Scott Gem Tip TM SCD 40 MHz CFT % MOS Al-1%Si M cm 100 µm 0.3~0.5 g g 2 Al-1%Si 2.1 Al-1%Si Al-1%Si 0.5 g 0.5 g -1% 2-1% 0.1 g 50 ml (g) (ml) (wt%) GB ~ Al-1%Si 2 1:5 10mL [3] 3 1 ml 1:1 1 1% 1:1 0.5 ml 10 46

51 2005 PerkinElmer ICP-OES ICP-MS [4] ml 50 ml g g Al-1%Si 100 ml 1 ml 1:1 1 ml 1:1 2 ml 0.5 ml 3 10 ml Si Fe Cu RF 1300 W L/min :15 : mm 5 s ml/min RSD DL [ / 100%] 4 4 /(µg L -1 ) /(µg L -1 ) /% RSD/% DL/(µg L -1 ) Si Fe Cu

52 2005 PerkinElmer ICP-OES ICP-MS IVA22 99%~101% RSD Al-1%Si 1.4% 5 5 Al-1%Si Si Fe Cu Al-1%Si 6 Al-1%Si w/ Si Fe Cu KS IVA KS IVA IVA Al-1%Si 4 [1] GB/T [S] -1%. [2] GB ~ [S]. [3] Al-1%Si [4]

53 2005 PerkinElmer ICP-OES ICP-MS 宋 晓 年, 冯 天 培 HCl+H 2 O 2 +HNO 3 GB/T [1] GB3260 [2] GB OPTIMA 3000 Perkin-Elmer Scott Gem Tip TM SCD 40 MHz CFT-33 CCD MOS 12M cm 2 49

54 2005 PerkinElmer ICP-OES ICP-MS ml 2.2 Cu Fe Sb As Al Cd HCl+H 2 O 2 Pb Bi Zn +H 2 SO 4 HCl+H 2 O 2 +HNO g g 20 ml 1:1 HCl 10 ml HCl 1 ml H 2 O 2 5 ml HNO As Bi Cu Fe Pb Sb Cd Zn Al RF 1300 W L/min :15 : mm 5s ml/min RSD DL [ / 100%] 2 5% µg/l 50

55 2005 PerkinElmer ICP-OES ICP-MS 0.01% 98%~102% RSD 4.15% 3 RSD 2.1% GB3260 GB %~0.0001% 0.01% 8% 20% 0.001%~0.0001% 2 /(µg L -1 ) /(µg L -1 ) /% RSD/% DL/(µg L -1 ) As Bi Cu Fe Pb Sb Cd Zn Al RSD 1 2 /% RSD/% /% RSD/% Cu Fe Pb Zn Al w/% As Bi Cu Fe Pb Sb Cd Zn Al

56 2005 PerkinElmer ICP-OES ICP-MS 4 5 GB3260 [1] GB/T [S]. [2] GB [S]. Determination of Trace Impurities in High Purity Metal Tin by Inductively Coupled Plasma Emission Spectrometry SONG Xiao-nian, FENG Tian-pei Xi an microelectronic Technology Institute, Xi an , China Abstract: This paper introduced a method of determination of trace impurities in high purity metal tin. The method was through adding HCl+H 2 O 2 +HNO 3 to digest sample, analyzed by inductively coupled plasma emission spectrometry. The experiment optimized element-wavelength and operation-parameter, eliminated influence by matrix matching. The recovery of the experiment proved: this method was accurate and reliable, can be used in rapidly determination of trace impurities in high purity metal tin. Key words: inductively coupled plasma emission spectrometry; high purity metal tin, impurities 52

57 2005 PerkinElmer ICP-OES ICP-MS ICP-AES 陆 伟 星, 王 鑫 艳 ICP-AES ICP L/min 0.5 L/min 12 mm [1] 2% 1000 µg/ml 0~1000 µg/ml [2,3] 0~100 µg/ml [4~6] X 1.2 ICP-AES l g 0.1 mg 100 ml l0 ml 20 min 5 ml 5 ml 5 ml 1 l ml ml 10.0 ICP Optima3300 Perkin-Elmer W 15 L/min ml 5 ml 53

58 2005 PerkinElmer ICP-OES ICP-MS 1:1 ICP 10.0 ml W 12 mm L/min [7] ICP [8] [9] 2.2 [8] Y(II) nm l nm GSBH nm µg/ml /µg 1 (n=6) RSD/% K Na Zn

59 2005 PerkinElmer ICP-OES ICP-MS 3 10 /% RSD/ 3 /% n=8 % K 2 2O GSBH 8 BH W 2 /nm µg/ml K Na Zn ICP-AES [4]. 2000, l20(1):52. [5] (4):85. [6] (5):41. [7] H.M.Kuss,Applicatious of Microwave Digestion Technique for Elemental Analysis Fresenius J Anal Clem [8]. [M] [1] :20. GSBH Na 2O Zn K 2O Na 2O Zn [2] GB/T [S]. Zn K 2O GBW Na 2O K 2O Na 2O GBW Zn [3] GB/T [S]. [9] :1222. Determination of K, Na and Zn in Iron Ore and Coke Ash by ICP-AES LU Wei-xing, WANG Xin-yan (Testing Laboratory of Technology Center Shanghai Meshan Co.Ltd Baosteel Group., China) Abstract A method for the determination of K, Na and Zn in iron ore and coke ash. In which the dissolution of sample using hydrochloric acid, nitric acid and hydrofluoric acid, after dissolution, add perchloric acid to treat the sample, and the determination use ICP-AES method. This article made the sample treatment and measurement parameter test, and investigated the precision of the method. The experimentation data shows that the method is simple in operating and the result is reliable. Key words ICP-AES K Na Zn iron ore and coke ash 55

60 2005 PerkinElmer ICP-OES ICP-MS 曹 丽 玲 Multiwave <1.0 g <0.5 g 5.0 ml 1.0 ml 1 P\T Multiwave 3000 / Power 800 W Ramp 10 min Hold 20 min P-Rata 0.7 bar IR-Lim 180C PE Multiwave

61 2005 PerkinElmer ICP-OES ICP-MS 3 1/

62 2005 PerkinElmer ICP-OES ICP-MS - 何 燕, 曹 丽 玲 GBW µg/ml 1% 100 µg/ml g 5 ml 1 ml Power 800 W Ramp 10 min Hold 20 min P-Rate:0.7 bars IR-lim L/min 0.2 L/min W ml 1.1 Perkin-Elmer Multiwave Perkin-Elmer Optima 2000DV 1% 10% 24 h µg/ml 1 r= µg/ml

63 2005 PerkinElmer ICP-OES ICP-MS X 1 = (A 1 A 2) * V 1 *1000/ m 1 * X 1 mg/kg A µg/ml A g µg/ml V 1 2 ml m 1 g 2 ρ/(µg. g -1 ) /% g 3 3 µg/g (µg/g) RSD/%

64 2005 PerkinElmer ICP-OES ICP-MS ICP-OES 杨 桂 芳, 厉 海 英, 朱 琴 ICP-OES ICP-OES MSF PE ICP-OES VLSI Perkin-Elmer Optima 2000DV ICP-OES ICP-MS GDMS 1 1 / / / / /W /MHz /MPa /s L min -1 L min -1 L min -1 m L min HCl HNO 3 H 2 SO g/ml 2 60

65 2005 PerkinElmer ICP-OES ICP-MS 2 v/v ρ / ρ/(µg. L -1 ) (mg. ml -1 ) Ca Cu Fe In Ni Pb Sn Zn % % g ml 3 ml ml 100 ml 3 3 Ca Cu Fe In Ni Pb Sn Zn λ/nm mg/mL MSF 1mg/mL 10mg/mL 1 Ni Cu MSF MSF 61

66 2005 PerkinElmer ICP-OES ICP-MS Ca Cu Fe In Ni Pb Sn Zn mg mg /% Ca Cu Fe In Ni Pb Sn Zn RSD/% ICP-OES 0.10 ~5.29% 96 ~105.8% 62

67 2005 PerkinElmer ICP-OES ICP-MS - Si Fe Na 李 明 利, 袁 园 Si Fe Na FAAS GFAAS AFS ICP-OES ICP-OES ICP-MS Milestone ETHOS E / Perkin-Elmer Optima 2100DV ICP-OES Fe Merck Fe mg/l 4% Si Si 10 [1] mg/l 4% Na NaCl h mg/l 1 CsCl 4% 65% 40 N 2 O-C 2 H 2 N 2 O-C 2 H [2] 1 10mL 1 63

68 2005 PerkinElmer ICP-OES ICP-MS 2 10mL 5 ml 2 50 ml 3 10 ml 1 Teflon / min / W / 5 ml 2 ml mL mL 4 ICP-OES 2 50mL Fe Na Si 4 2 ICP-OES / / / /W L min -1 L min -1 L min -1 Fe Si Na H 2 SiO 3 ICP-OES ICP-OES 3 3 HF 1 Na HF HF Si 2 Fe SiF 4 3 Si 2.2 [1,3] HF H 2 SiF 6 HF 85.0%~98.0% 64

69 2005 PerkinElmer ICP-OES ICP-MS 3 ICP-OES Fe Na Si ρ/ mg L HNO 3 Si Fe Na Fe Na Si GB/T GB/T [4,5] Si ICP-OES 43.6 mg/l Fe 2.78 mg/l Na 16.3 mg/l 3 3 ICP-OES 4 [1] [2] B [3] [4] GB/T [5] GB/T

70 2005 PerkinElmer ICP-OES ICP-MS 王 雅 宁 ICP-AES ICP-AES ICP-AES 10 ml 1 ml 200 ml ICP-AES Ni Mn Mo Si Cu P Perkin-Elmer Optima 2000DV 1300 W 15 L/min 0.2 L/min 0.8 L/min 1.5 ml/min 15 mm 30 s s MHz 1:1 4 Mn Si Cu Ni Mo P 1 mg/ml g 150 ml

71 2005 PerkinElmer ICP-OES ICP-MS 1 /nm Ni L/min Mn Mo P Si µg/ml 10 3 Cu Ni Mn Mo P Si Cu / µg ml /% /% /% SD RSD/% /% Ni Mn 0.66 Mo 3.06 P Si Cu

72 2005 PerkinElmer ICP-OES ICP-MS 4 /% /% SD RSD/% Ni Mn Mo P Si Cu [1] PerkinElmer 90%~103% 5.89%~0.30%.Optima. [2]. [M]. RSD n=10. 68

73 2005 PerkinElmer ICP-OES ICP-MS 鲁 颖 ICP-AES Al Sn Cu Fe 94.2 ~102.4 ICP-AES 1.2 Al Sn Cu Fe ml 100 ml 5 ICP-AES ml 5 ml Al Sn Cu Fe NaCl g NaCl 5 ml 15 NaCl 5 ml Perkin-Elmer Optima 2000DV kw 15 L/min ρ/ mg ml L/min 0.8 L/min L/min 15 mm 30 s Al s Sn ρ=1.84 g/cm 3 ρ=1.42 Cu g/cm 3 1mg/mL 5:95 Fe mg/mL 5: mg/mL 5:95 1mg/mL 5:95 69

74 2005 PerkinElmer ICP-OES ICP-MS ~10 ml 5 ml 2 2 Al Sn Cu Fe 2.4 λ/nm s n=10 Al Sn Cu Fe / µg ml R n=10 ρ/ mg ml -1 RSD/% /% Al Sn Cu Fe ICP-AES Al Sn Cu Fe [2] [1]. [M]... [M].. 70

75 2005 PerkinElmer ICP-OES ICP-MS 王 静 ICP-AES 94.5%~108% RSD 1.06%~4.08% ICP-AES 10 s, 1.2 1:1 1:1 99.9% Al Mn Zn Zr Nd Si Cu Fe Al Mn Zn Si Cu Fe Ni Be Ni Zr Zn Be 1 mg/ml 1.3 ICP-MS ~ g 100 ml 10 ml Perkin-Elmer Optima 2000DV kw 15 L/min 0.2 L/min 0.8 L/min 1.5 L/min 15 mm 30 s 71

76 2005 PerkinElmer ICP-OES ICP-MS ~15 ml 15 ml 10 ml /nm /nm Al Ni Mn Fe Zn Be Si Zr Cu Nd ml 0.1 g ml Al Mn Zn Si Cu /(mg L -1 ) Ni Fe Be Zr Nd /(mg L -1 ) Al Mn Zn Zr Nd BY Mn Zr Si 94.5%~108% RSD 1.06%~4.08% 72

77 2005 PerkinElmer ICP-OES ICP-MS 4 m/mg Al Mn Zn Si Cu Ni Fe Be Zr Nd 100mgMg+ 10mgAl 100mgMg+ 0.5mgMn 100mgMg+ 5mgZn 100mgMg+ 1mgZr 100mgMg+ 2mgNd w/% x RSD/% w/% /% Al Mn Zn Si Cu Ni * Fe Be * Zr * Nd * Ni Be Zr Nd

78 2005 PerkinElmer ICP-OES ICP-MS Al Mn Zn Si Cu Ni Fe Zr Be Nd 3 ICP-AES Al Mn Zn Si Cu Ni Fe Zr Be Nd 6 w/% RSD% Al Mn BY Zn Si Fe Cu Al Mn BY Zn Si Fe Cu

79 2005 PerkinElmer ICP-OES ICP-MS ICP-OES 15 ( ) :,, 15 10% 87% 110% : (ICP-OES); ; [1] 0.01 ug/ml 10% 87% 110% PE Optima 2000DV 160nm 900nm CCD PE Optima % 10% 3% ug/ml Cr Fe Sn Zr Cd Co Al Cu Mn Mo Ni Ti V Ta Nb Pb Y ug/ml % Zr Cr Fe Sn mg/mL mg/mL

80 2005 PerkinElmer ICP-OES ICP-MS RF 1200W 1mg/mL 18L/min 0.8L/min 10% 0.2L/min 1.5mL/min 10% 3 100mg/mL nm g 5mL40% 0.7mL Al Cd mL40% Co Ta ml3% Cu Ti mL 10ug/mL 25 Mn V ml Mo Cr Al Cd Co Cu Mn Mo Nb Ni Pb Nb Fe Ta Ti V Ni Sn mL 10mL Pb Cr Fe Sn 10% ICP 160nm 900nm ICP 4 4 nm nm Cd , , , Al , , , , Co , , , , Cu , , , , Mn , , , , Mo , , , Ni , , , , Ti , , , , V , , , , Ta , , , , Nb , , , , Pb , , , ,

81 2005 PerkinElmer ICP-OES ICP-MS Cr , , , Fe , , , Sn , , , [2] 15 PE Optima2000DV nm 2.3 Al Cd Mn Nb Pb nm ug/ml Cd Al Co Mn Mo Cu Ni Ti V Ta Nb Pb

82 2005 PerkinElmer ICP-OES ICP-MS Cd Al Co Mn Mo Ti V Ta Nb Pb Cu Ni ug/ml 10 ug/ml 6 6 n=6 ug/ml RSD ug/ml RSD ug/ml *ug/ml % % ug/ml *ug/ml % % Cd Cd Al Al Co Co Cu Cu Mn Mn Mo Mo Ni Ni Ti Ti V V Ta Ta Nb Nb Pb Pb * mg/mL

83 2005 PerkinElmer ICP-OES ICP-MS 7 nm 5% 10% 15% 20% 25% 30% 35% Cd Al Co Cu Mo Mn Ni V Ti Ta Nb Pb Cr Fe Sn % 35% 0mg/mL 1mg/mL 5% 10% 2mg/mL 3mg/mL 4mg/mL 5mg/mL 6mg/mL 7mg/mL 8mg/mL 9mg/mL 10% 10mg/mL mg/ml nm ug/ml Cd Al Co Cu Mo Mn Ni V Ti Ta Nb Pb Cr Fe Sn

84 2005 PerkinElmer ICP-OES ICP-MS 7mg/mL mg/mL Cr Fe Sn 5mg/mL 3.7 6mg/mL 5mg/mL Cr Fe Sn 1mg/mL 9 12 nm Cd Std (ug/ml) Al Co Cu Mo Mn mL Ni % 5mL 3% V Nb Pb mg/mL 5.5ug/mL 11.5 ug/ml 65 ug/ml 10 ug/ml 10 Cr Fe Sn 12 nm Std (ug/ml) 12 Cr Fe Sn mg/mL Ti Ta n=11 nm ug/ml ug/ml % ug/ml ug/ml % Cd Al Co Cu Mo Mn Ni V

85 2005 PerkinElmer ICP-OES ICP-MS Ti Ta Nb Pb nm % % % RSD% % % % RSD% Cr Fe Sn [1] GB/T [2] ICP-AES. J Determination of Impurity Elements in Zirconium Alloys by ICP-OES Method Song Jun-wu, Dong Shi-zhe,Dai Yan,Han Xiang-jun,Zhou Yi-fang (The Physical And Chemical Testing And Measuring Centre China National Nuclear Industry Corp.202 Factory,Baotou ) Abstract: In this report, the Zirconium alloys sample was dissolved by HNO 3, HF and H 2 SO 4.The 15 impurity elements in Zirconium alloys sample was determined directly by inductively coupled plasma emission spectrometry. Yttrium was used as internal standard. In order to find out the influence for determination, amount added of internal standard and acid was studied. And the influence of the concentration of matrix was also studied.the chemical station was used effectively to prevent the interference during analysis. By this method, the lower detection limits were obtained. Virtue of the method is rapid and accurate. The relative standard deviation of the method for various elements is less than 10% and the recovery of the elements studied was 87% to 110%. Keywords: ICP-OES; Zirconium alloys; Trace elements. 81

86 FIELD APPLICATION REPORT ICP-MASS SPECTROMETRY Environmental Applications of the ELAN DRC-e ICP-MS Author: Ewa Pruszkowski PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT Introduction The ELAN product line consists of three ICP-MS models: ELAN 9000, ELAN DRC-e and ELAN DRC II. The ELAN 9000 is a rugged, reliable system designed for routine analysis in high throughput laboratories. It uses proven methodology for environmental, clinical, geochemical and other applications. The ELAN DRC II, with its patented Dynamic Reaction Cell (DRC ) technology, is designed for ultratrace analyses and is routinely used in a variety of application areas, including semiconductor, clinical, metallurgical, environmental and geological. It is also an ideal tool for research-oriented applications and offers exciting potential for additional application areas. The ELAN DRC-e was introduced in 2002 as a result of the needs of routine, high-throughput labs performing trace analysis in difficult matrices. Like the DRC II, it uses Dynamic Reaction Cell technology to lower detection limits and reduce interferences compared to conventional ICP-MS systems. The ELAN DRC-e is available with one or two reaction-gas mass-flow controllers which allow the use of all typical reaction gases except ammonia. An optional upgrade kit is available for ammonia operation. The system is equipped with nickel cones and an HF corrosion resistant sample introduction system. The ELAN DRC-e is upgradeable to full DRC II functionality through the purchase of an upgrade kit (which includes the corrosive purge assembly, second reaction gas channel with getter, platinum cones, quartz sample introduction kit and vacuum and mass-flow controller enhancements). The Dynamic Reaction Cell with the innovative Axial Field Technology makes the DRC-e the instrument of choice for many applications, including chemical, environmental, nuclear, and geochemical, where optimal performance in challenging matrices is required. Experimental This field application report summarizes the instrumental conditions used for analyzing environmental samples using the ELAN DRC-e (see Table 1) and gives examples of typical results that can be obtained. For analysis of drinking water Standard Reference Materials (SRMs), ICSA and ICSAB standards (Interference Check, Standards A and AB from U.S. EPA Method 6020), NASS-5 (seawater SRM from Canadian Research Council), and NIST 2711 (soil SRM from National Institute of Standards and Technology, Gaithersburg, MD, U.S.A.), two reaction gases were used: methane and oxygen.

87 Table 1. General Conditions for ELAN DRC-e Instrument ELAN DRC-e Samples Drinking water SRM, ICS A, ICS AB, NASS-5, NIST 2711 RF Power Nebulizer Flow 1400 W 0.94 L/min Reaction Gases CH 4, O 2 Calibration External Calibration Standards Blank, 10, 20 µg/l in 1% HNO 3 Internal Standards Measurement Time Rh, Ir (20 µg/l) 1 second per isotope Replicates 4 Results In the Trace Metals in Drinking Water SRM (High Purity Standards, Charleston, SC, U.S.A.), four elements benefit from DRC technology: Cr, Ni, As and Se. Instead of using elemental equations required by traditional ICP-MS systems to correct for polyatomic spectral interferences, the reaction gas (methane and/or oxygen) used in the ELAN DRC-e removes the interferences from the ion beam through predictable and controllable gas-phase ion-molecule reaction chemistry. The result is improved detection limits and better accuracy, as shown in Table 2. Table 2. Trace Metals in Drinking Water SRM Analyte Reaction Gas RPq Experimental Certified Value Recovery Flow (ml/min) Result (µg/l) (µg/l) (%) Cr 52 CH Ni 60 CH As 75 CH Se 78 CH Only DRC-mode elements shown In wastewater and soil analysis using the U.S. EPA method 6020 protocol, analysis of Interference Check Standard A and AB (ICS A, ICS AB) is required. The interfering elements such as Al, Ca, Fe, Mg, Na, K, C, Cl, P, and S are present in both standards at the level of mg/l. Even through no specific criteria are given in the method, our goal was to measure values less than 1 µg/l in the ICS A and +/- 20% of the spike value in the ICS AB. As and Se were analyzed with both CH 4 and O 2 to check which gas was most suitable (Table 3). The seawater SRM (NASS-5) has very low concentrations of several elements and is often used as a seawater blank. Our goal was to measure sample values lower than 1 µg/l in a 10x dilution and obtain spike recoveries +/- 20%. Table 4 shows the reaction cell conditions used for the analysis and the results. 2

88 Finally, an analysis of a leached soil, NIST SRM 2711 was performed on a sample prepared several years ago. Due to sample evaporation over time, the measured concentrations were slightly high, but the spike recoveries at low levels were good, as shown in Table 5. Table 3. Interference Check Standard A and AB Analyte Reaction Gas Flow (ml/min) RPq Experimental Value (µg/l) Expected Value (µg/l) Spike (µg/l) Recovery (%) Cr 52 CH < Ni 60 CH < Cu 65 CH < Se 80 CH < Se 78 O < AsO 91 O < Only DRC elements shown Concentration of interfering elements = mg/l Table 4. Seawater SRM NASS-5 Analyte Reaction Gas Flow (ml/min) RPq Experimental Value (µg/l) Expected Value (µg/l) Spike (µg/l) Recovery (%) Cr 52 CH < Fe 56 CH < Ni 60 CH < Cu 65 CH < Se 78 CH < Se 78 O < AsO 91 O < Only DRC elements shown SRM was diluted 10x Table 5. Leached Soil, NIST SRM 2711 Analyte Reaction Gas Flow (ml/min) RPq Experimental Value (µg/l) Expected Value (µg/l) Spike (µg/l) Recovery (%) Cr 52 CH Ni 60 CH Cu 63 CH Se 78 CH (0.1) Se 78 O (0.1) AsO 91 O Only DRC elements shown SRM was diluted 1000x 3

89 Conclusion In summary, the ELAN DRC-e is an excellent instrument for routine trace analyses where there is a need for lower As, Se, Cr, Fe, and Ni detection limits in high matrix environmental samples. This work has demonstrated the ELAN DRC-e can successfully measure a number of elements which suffer from interferences in a variety of environmental samples, including drinking waters, seawaters, soils and interference check standards. The ELAN DRC-e is a rugged, easy-to-use system for routine labs wanting to switch from graphite furnace atomic absorption (GFAA) to a more productive tool. PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA Phone: (800) or (+1) For a complete listing of our global offices, visit PerkinElmer, Inc. All rights reserved. The PerkinElmer logo and design are registered trademarks of PerkinElmer, Inc. PerkinElmer is a registered trademark and Dynamic Reaction Cell and DRC are trademarks of PerkinElmer, Inc. or its subsidiaries, in the United States and other countries. ELAN is a registered trademark of MDS Sciex, a division of MDS, Inc. All other trademarks not owned by PerkinElmer, Inc. or its subsidiaries that are depicted herein are the property of their respective owners. PerkinElmer reserves the right to change this document at any time without notice and disclaims liability for editorial, pictorial or typographical errors. The data presented in this Field Application Report are not guaranteed. Actual performance and results are dependent upon the exact methodology used and laboratory conditions. This data should only be used to demonstrate the applicability of an instrument for a particular analysis and is not intended to serve as a guarantee of performance _01

90 Interference Removal and Analysis of Environmental Waters Using the ELAN DRC-e ICP-MS Introduction Drinking-water regulations have been established in most industrialized nations. Although the list of elements for each country is typically very similar, the regulatory limits can vary, as shown in Table 1. ICP-MS has become a popular technique for the analysis of drinking water because of its very low detection limits and short analysis time. Performing an analysis for as many as 30 elements by ICP-MS generally takes less than 5 minutes per sample, compared to several hours per sample using other techniques with similar detection capabilities, such as graphite furnace atomic absorption. ICP MASS SPECTROMETRY A P P L I C A T I O N N O T E Table 1. Typical regulated elements and allowable limits for drinking water in the United Kingdom (NS30), European Union (EU) and United States of America (USA) Element NS30 EU USA (µg/l) (µg/l) (µg/l) Silver (Ag) Aluminum (Al) Arsenic (As) Boron (B) 2,000 1,000 Barium (Ba) 1,000 2,000 Berilium (Be) 4 Calcium (Ca) 250,000 Cadmium (Cd) Chromium (Cr) Copper (Cu) 3,000 1,000 1,000 Iron (Fe) Mercury (Hg) Potassium (K) 12,000 Magnesium (Mg) 50,000 Manganese (Mn) Molybdenum (Mo) Sodium (Na) 150, ,000 Nickel (Ni) Phosphorous (P) 2,200 Lead (Pb) Antimony (Sb) Selenium (Se) Thallium (Tl) 2 Uranium (U) 30 Vanadium (V) 50 Zinc (Zn) 5,000 5,000 Authors Kenneth R. Neubauer, Ph.D. Ruth E. Wolf, Ph.D. PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA

91 However, depending on the matrix content and sample-preparation strategies, some elements (such as arsenic, selenium, chromium, iron and nickel) may suffer from polyatomic interferences in ICP-MS. These interferences may be even more troublesome in other environmental water samples, such as bottled water or wastewater because of the higher matrix content. These interferences can be overcome in most cases by applying elemental correction equations or monitoring an alternative isotope. Unfortunately, for monoisotopic elements such as arsenic, an alternative isotope is not available. Additionally, some matrices have such high levels of interferences that the use of interference corrections becomes difficult. Table 2 lists some commonly occurring interferences in ICP-MS that may affect the analysis of environmental samples such as drinking water. An alternative approach to applying interference correction equations is to remove the interferences prior to analysis. This can be accomplished with the ELAN DRC-e ICP-MS (PerkinElmer SCIEX, Concord, ON, Canada). The ELAN DRC-e uses patented Dynamic Reaction Cell (DRC ) technology to chemically remove the interferences from the ion beam before they enter the analyzer quadrupole. The DRC consists of a quadrupole mass filter in an enclosed cell located between the ion optics and the analyzer quadrupole of an ICP-MS. The enclosed cell can be pressurized with a reaction gas that chemically reacts with the interfering species to remove them from the ion beam. In addition, the active quadrupole inside the DRC provides the ability to establish a mass bandpass window, with both a high-mass and low-mass cutoff, inside the cell. Ions that are not stable within this mass bandpass window are ejected from the cell before they can enter the analyzer quadrupole. The bandpass window can be automatically changed with the analyte mass via Dynamic Bandpass Tuning (DBT) to carefully control the chemistry occurring inside the cell. This unique feature provides control of the reaction chemistry to eliminate the possibility of new interferences forming within the reaction cell. In addition, Axial Field Technology Table 2. Common interferences in ICP-MS (AFT) provides optimal performance in any matrix. With the ELAN DRC-e, interferences are eliminated by this combination of chemical resolution and DBT inside the Dynamic Reaction Cell, minimizing the need to use correction equations. The data presented in this paper demonstrate the capabilities of the ELAN DRC-e for running all of the elements in U.S. Environmental Protection Agency (EPA) Method 200.8, the standard method used in the U.S.A. for the analysis of drinking waters and wastewaters Interfering Species Affected Elements 40 Ar 35 Cl +, 40 Ar 37 Cl +, 40 Ca 35 Cl +, 40 Ca 37 Cl + 75 As +, 77 Se + 79 BrH +, 81 BrH + 80 Se +, 82 Se + 35 Cl 16 O +, 37 Cl 16 O + 51 V +, 53 Cr + 40 Ar 12 C + 52 Cr + 40 Ar 14 N + 54 Fe + 40 Ar 16 O +, 40 Ca 16 O + 56 Fe + 40 Ar 16 OH +, 40 Ca 16 OH + 57 Fe + 42 Ca 16 O +, 44 Ca 16 O + 58 Ni +, 60 Ni + 40 Ar 23 Na + 63 Cu + Table 3. ELAN DRC-e instrumental parameters Parameter Value RF Power 1300 W Nebulizer Gas Flow ~ L/min (set for < 3% oxides) Sample Introduction Rate 1 ml/min Nebulizer Cross-flow Gem-tip Spray Chamber Ryton Scott-type Double-pass Reaction Gas Methane (99.999%) DRC Pressurization Time 30 sec DRC Gas Flow Change Time 15 sec DRC Vent Time 30 sec Detector Mode Dual Lens Scanning Sampler/Skimmer Cones Nickel Dwell Time 50 ms Points per Peak 1 Sweeps per Reading 20 Readings per Replicate 1 Replicates 3 2

92 by ICP-MS. Although the data was based on Method 200.8, it can also be used to demonstrate performance improvements for these elements under the methods required by other countries. Although the majority of elements specified in U.S. EPA Method are determined without much difficulty by conventional ICP-MS, a few may suffer from interferencerelated problems in certain matrices. In these cases, an alternative samplepretreatment step or alternate analytical technique may be needed to confirm the results, which increases the time and cost of the analysis. It would be highly advantageous to Table 4a. Isotopes monitored and applicable parameters in standard mode Analyte Isotope RPq Correction Equations Al Sb 121, As * [ArCl77 - (0.815 * Se82)] Ba 135, Be Cd 111, Cr Co Cu 63, Fe 54, Pb * Pb * Pb207 Mn Mo Ni Se * Kr83 Ag Tl Th U V Zn Ca 43, Mg Na K RPq is the Dynamic Bandpass Tuning parameter Table 4b. Isotopes monitored and applicable parameters in DRC mode Element m/z CH 4 Flow RPq (ml/min) Cr Ni As Se Se RPq is the Dynamic Bandpass Tuning parameter run all the elements in a single multielemental analysis in order to improve sample throughput and decrease the cost of analysis. The data in this paper show that by using the ELAN DRC-e, all of the elements in a typical drinking-water, bottled-water or wastewater sample analyzed using U.S. EPA Method can be run in a single analysis using a single reaction gas even those with problematic interferences. Experimental All work discussed here was performed on an ELAN DRC-e in a normal laboratory setting (i.e., non-cleanroom conditions). The guidelines outlined in the U.S. EPA Method turnkey application note were followed. 1 The objective of this work was to develop a single, robust analytical method using one reaction gas for the analysis of water samples in order to maximize sample throughput, overall laboratory productivity and ease-of-use. The instrument conditions used for this method are displayed in Table 3. While the stated goal has been achieved, it should be noted that, if necessary, superior performance for some elements determined in DRC mode may be achieved using individually optimized conditions for each element or a different reaction gas. Tables 4a and 4b show the elements measured, the mode used, and the appropriate reaction-cell conditions. DRC-mode analysis indicates that the DRC was pressurized with a reaction gas and Dynamic Bandpass Tuning applied. For all the DRCmode analyses, % methane (Matheson Gas Products, East Rutherford, NJ, USA) was used as the reaction gas. Standard mode 3

93 indicates that the DRC cell was not pressurized (i.e., the reaction gas was turned off) and is approximately equivalent to running the elements on an ELAN 9000 instrument. All of the elements analyzed in DRC mode were also analyzed in standard mode for comparison purposes. By analyzing samples this way, each sample is run one time with one method, and both DRC- and standardmode analyses are accomplished with one optimization file. This scheme eliminates the need to analyze each sample more than once, to use two methods or to use two optimization files. For various elements, multiple isotopes were monitored for comparison purposes. Table 4a also shows any interference corrections (if used) for standard-mode analysis. Calibration standards were made in 1% nitric acid at the levels indicated in Table 5 and run as an external calibration curve. The calibration blank was a 1% nitric acid solution, and blank subtraction was used. An internal standard mixture was added to the blank, the standards and each sample to yield a final concentration of 20 µg/l 6 Li, 45 Sc, 71 Ga, 103 Rh, 115 In, 165 Ho and 193 Ir. A wastewater reference material (High Purity Standards, Charleston, SC, USA) and a municipal wastewater were digested using U.S. EPA Method Spikes were added prior to digestion. Results Interference Reduction The major elements in environmental water that suffer from interferences include Cr, Ni, As and Se; their associated interferences are ArC +, CaO +, ArCl + and Ar 2+, respectively. These interferences originate from Table 5. Calibration standard concentrations Analytes Standard 1 Standard 2 Standard 3 (µg/l) (µg/l) (µg/l) Al, Sb, As, Ba, Be, Cd, Cr, Co, Pb, Mn, Mo, Ni, Se, Ag, Tl, Th, U, V, Zn Na, Mg, K, Ca, Fe 200 1,000 10,000 Table 6a. Instrument detection limits (IDLs) and Matrix-IDLs in standard mode Analyte m/z IDL Matrix-IDL 200Matrix-IDL (µg/l) (µg/l) (µg/l) Be Al V Cr Mn Co Ni Cu Zn As Se Mo Ag Cd Sb Ba Tl < Pb Th U 238 < Na Mg K Ca N.A. Fe N.A. = Not Applicable (Ca in matrix) Table 6b. Instrument detection limits (IDLs) and Matrix-IDLs in DRC mode Analyte m/z IDL Matrix-IDL 200Matrix-IDL (µg/l) (µg/l) (µg/l) Cr Ni As Se

94 matrix and/or plasma species. To reduce these interferences, methane was used as the reaction gas. Although Fe can suffer from ArN +, ArO +, CaO +, ArOH + and CaOH + interferences at low levels, it is usually present at elevated levels in North American drinking waters, and therefore can be determined in standard mode. 1 The ELAN software performs an optimization routine that allows the correct gas flow to be chosen. This optimization is performed once, and the optimum gas-flow settings are then stored in the analytical method for use in routine analysis. An example showing the reactioncell gas eliminating an interference appears in Figure 1. In this example, methane is the reaction gas and the matrix consists of 50 mg/l Ca, 0.5% HCl and 0.2% MeOH. In this figure, the red line represents the signal due to the matrix, and the blue line is the signal resulting from the matrix + 1 µg/l Ni plotted as a function of methane flow. The difference between these lines represents the signal due to 1 µg/l Ni. From this plot, it is evident that methane removes the CaO + interference on Ni. The appropriate cell gas flows for the other elements were determined in a similar manner. The final methane flow and bandpass parameters used for the analysis are shown in Table 4b. Method Validation Both instrumental detection limits (IDLs) and matrix instrument detection limits (Matrix-IDLs) were determined (Table 6). As defined by the U.S. EPA, detection limits were calculated by measuring a sample eight times, using three replicates per measurement. The standard deviation of these eight measurements was then multiplied by three Figure 1. Cell gas optimization plot of Ni in a calcium matrix. The red line (squares) represents the signal at m/z 60 at various CH 4 flows from a matrix consisting of 50 mg/l Ca + 0.5% HCl + 0.2% methanol. The blue line (diamonds) is for the same solution, but spiked with 1 µg/l Ni. Table 7. Analysis of a drinking-water certified reference material (Trace Metals in Drinking Water, High Purity Standards) Standard Mode (unless indicated) Analyte m/z Certified Experimental Recovery Value Value (%) (µg/l) (µg/l) Be Al V Cr Cr DRC-e Mn Co Ni Ni DRC-e Cu Zn As As DRC-e Se Se DRC-e Mo Ag Cd Sb Ba Tl Pb Th U Na 23 6,000 6, Mg 24 9,000 9, K 39 2,500 2, Ca 43 35,000 31, Fe

95 to obtain the detection limit for each element. The instrument detection limit (IDL) for each element was determined by measuring a 1% nitric acid blank; the Matrix-IDLs were determined in the same manner, except a solution consisting of 50 µg/l Ca, 0.2% methanol and 0.1% HCl was analyzed. This matrix solution was devised to test the detection capabilities of the ELAN DRC-e in a sample matrix containing the common interfering species of calcium, chloride and carbon which can lead to the interferences listed in Table 2. In addition, since many natural waters and wastewaters contain high levels of calcium that lead to increased levels of CaO + interference and false positives for nickel in these types of samples, the Ni detection limit was also evaluated in a solution containing 200 mg/l calcium (Matrix200-IDL). The advantage of DRC-mode analysis is evident in the results in Tables 6a and 6b. As Tables 6a and 6b illustrate, the detection limits determined in DRC mode are superior to those determined in standard mode for most elements and are adequate for the determination of analytes at normal levels in drinking-water samples. Even at high calcium concentrations, the DRC-mode Matrix200-IDL for nickel is substantially better than that determined in standard mode and comparable to the Matrix-IDL determined in the solution containing calcium, chloride and carbon. Thus, the use of DRC mode will result in a more robust determination for nickel using the mass 60 isotope (the most abundant practical isotope) in various types of calciumcontaining matrices, such as hard waters and treated wastewaters. Table 8. Analysis of trace metals in a natural-water certified reference material (NIST 1640) Standard Mode (unless indicated) Analyte m/z Certified Experimental Recovery Value Value (%) (µg/l) (µg/l) Be Al V Cr Cr DRC-e Mn Co Ni Ni DRC-e Cu Zn As As DRC-e Se Se DRC-e Mo Ag Cd Sb Ba Tl 205 < Pb Th U Na 23 29,350 28, Mg 24 5,819 5, K Ca 43 7,045 6, Fe To test the method performance, a drinking-water standard reference material, Trace Metals in Drinking Water (High Purity Standards), was analyzed. The results are presented in Table 7. The data in Table 7 show that the experimental values for the analytes in Method are generally within 10% of the certified values for all the elements determined in both standard and DRC modes. This is considered an acceptable recovery for U.S. EPA Method 200.8, which requires the measured value of an external quality-control sample to be within ±10% of the stated value. The alkali metals are not listed as analytes in Method 200.8, but are included for informational purposes. The results for K and Ca are somewhat low, probably due to the comparatively low value of the high calibration standard for these elements as compared to the levels present in the sample. Table 8 shows similar results for a certified natural-water sample and Table 9 for a wastewater 6

96 material. The Al and V values in the wastewater sample are higher than expected. The high Al value is likely due to contamination during the digestion step. The high V measurement may be due to the presence of another interference, such as ClO +, arising from the HCl used in the digestion step. To evaluate the method on real samples, drinking-water, bottledwater and wastewater samples were analyzed. The results for these samples, as well as recoveries for a 10 µg/l spike, are presented in Tables 10, 11 and 12. These results show that the major matrix species are present at different levels in the samples, as evidenced by the differing Ca, Na, K and Mg concentrations. Nevertheless, spike recoveries are within ±10% for all elements present in the samples at concentrations under 50 µg/l. The wastewater sample spikes were added prior to digestion, so the acceptable limits for the spike recovery are broader at ±20%. For samples where the unspiked concentrations are greater than ten times the spike value, recovery calculations were not performed. In the public-water sample (Table 10), a difference of about 8 µg/l is observed in the nickel values measured under standard and DRC conditions. The higher value results from standard-mode analysis and indicates an interference contributing to the apparent 60 Ni signal, most likely CaO +. Applying DRC-mode conditions eliminates the interference and leads to a much better spike recovery (98%). The same effect is also evident for Ni in the bottled-water sample (Table 11). Table 9. Analysis of a wastewater certified reference material (Trace Metals in Wastewater E, High Purity Standards) Standard Mode (unless indicated) Analyte m/z Certified Experimental Recovery Value Value (%) (µg/l) (µg/l) Be Al V Cr Cr DRC-e Mn Co Ni Ni DRC-e Cu Zn As As DRC-e Se Se DRC-e Mo Ag Cd Sb Ba Tl Pb Th 232 U 238 Na 23 Mg 24 K 39 Ca 43 Fe Figure 2a. Stability analysis of low-level elements ( µg/l) in tap water over 9.5 hours. In the legend, s= standard mode, D=DRC mode. 7

97 These results indicate that the DRC mode removes interferences, thereby allowing lower concentrations to be measured. The stability of the method was evaluated by analyzing a tap-water sample for 9.5 hours against an external calibration curve. Figures 2a and 2b show stability plots of several DRC- and standard-mode elements; Figure 2a shows data for low-level analytes (concentrations below 0.7 µg/l), and Figure 2b shows data for high-level elements (greater than 1 mg/l). These plots demonstrate the stability of the method while switching between standard- and DRC-mode analysis by showing how stable the measurements are, both at very low and very high levels. Interferences can greatly affect an analysis, but the ELAN DRC-e eliminates them, leading to more stable analyses. This is demonstrated in Figure 3, for the determination of 52 Cr in a tap-water sample over 9.5 hours. In standard mode, the 52 Cr signal is initially approximately 0.5 µg/l but continuously decreases with time to slightly less than 0.1 µg/l. This indicates that an interference must be present on 52 Cr signal in standard mode, leading to the instability of the signal, as well as the high baseline values. The most probable explanation for the decline in this standard-mode mass 52 signal is the decreasing carbon content of the water, resulting from CO 2 outgassing to establish an equilibrium. 2 As the carbon-dioxide content of the water decreases with time, the amount of carbon available in the sample matrix to form ArC + in the plasma decreases; hence, the decrease in the background signal at mass 52 Table 10. Analysis of a Connecticut public water supply with a 10 µg/l spike Standard Mode (unless indicated) Analyte m/z Sample Sample + Spike Spike Recovery (µg/l) (µg/l) (%) Be Al V 51 < DL Cr 52 < DL Cr DRC-e 52 < DL Mn Co Ni Ni DRC-e Cu * * Zn * * As As DRC-e Se Se DRC-e Mo Ag Cd 114 < DL Sb Ba Tl Pb Th U Na 23 14,393 * * Mg 24 2,818 * * K 39 1,984 * * Ca 43 14,977 * * Fe * * *Spike was too low relative to the native concentration Figure 2b. Stability analysis of high-level elements (2-20 mg/l) in tap water over 9.5 hours. These elements were determined in standard mode. 8

98 from ArC +. Since this interference is eliminated using DRC mode, the background for the signal is substantially lower at about 0.1 µg/l and very stable (with the exception of a small plateau around the 5-hour mark due to Cr contamination in one of the autosampler vials). It is also interesting to note that after approximately 7.5 hours, the standard-mode background signal level at mass 52 is at the same level as in DRC mode. Presumably, this indicates that equilibrium has been established and carbon dioxide is no longer being liberated from the sample. Conclusion The results of this study indicate that the ELAN DRC-e can be used to analyze environmental-water samples using both standard-mode and DRC-mode analysis in a single analytical run using a single reaction gas. The total time of analysis per sample, including sample uptake, read delay and rinse times, was just under 6 minutes per sample for a total of 45 isotopes, although fewer isotopes would be used in a typical analysis (multiple isotopes were used for method-development and validation purposes). The data shown were analyzed following the guidelines in U.S. EPA Method for drinking-water and wastewater analysis in the United States. Following this method, the IDLs, Matrix-IDLs and recoveries for a certified reference material are reported. The use of a reaction gas in the ELAN DRC-e provides the ability to measure lower levels of certain elements in samples due to the elimination of matrix- and plasmabased polyatomic interferences. The data also show that the stability of the system is maintained, even while Table 11. Analysis of a bottled water and 10 µg/l spike Standard Mode (unless indicated) Analyte m/z Sample Sample + Spike Spike Recovery (µg/l) (µg/l) (%) Be 9 < DL Al V Cr 52 < DL Cr DRC-e Mn Co 59 < DL Ni Ni DRC-e Cu 63 < DL Zn 66 < DL As As DRC-e Se Se DRC-e Mo Ag Cd 114 < DL Sb Ba * * Tl Pb 208 < DL Th U Na 23 6,300 * * Mg 24 26,232 * * K * * Ca 43 60,120 * * Fe * * *Spike was too low relative to native concentration Figure 3. Chromium signals in standard (blue squares) and DRC (red triangles) modes for a 9.5 hour analysis of tap water. The lower, steadier signal in DRC mode indicates removal of the ArC + interference. The decreasing Cr signal in standard mode results from dissolved CO 2 outgassing over time, resulting in a decreasing carbon (ArC + ) background. 9

99 switching between DRC and standard modes of analysis within a method. These results demonstrate that the ELAN DRC-e ICP-MS can be used for effective multielemental analysis of drinking water, bottled water and a typical municipal wastewater, even those with significant calcium content. It is superior to conventional ICP-MS for several key elements because common polyatomic interferences are removed. References: 1. U.S. EPA Method for the Analysis of Drinking Waters and Wastewaters, Application Note D-6527, Ruth E. Wolf, Eric Denoyer and Zoe Grosser, PerkinElmer Instruments, Environmental Chemistry, 5th Edition, Stanley E. Manahan, Lewis Publishers, 1991, pages 40-42, 156. Table 12. Analysis of a municipal wastewater and 10 µg/l pre-digestion spike Standard Mode (unless indicated) Analyte m/z Sample Sample + Spike Spike Recovery (µg/l) (µg/l) (%) Be Al * * V Cr Cr DRC-e Mn * * Co Ni Ni DRC-e Cu Zn As 75 < DL As DRC-e Se Se DRC-e Mo Ag Cd Sb Ba Tl Pb Th U Na 23 * * Mg 24 9,470 * * K 39 6,000 * * Ca 43 37,000 * * Fe * * *Spike was too low relative to native concentration PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA Phone: (800) or (+1) For a complete listing of our global offices, visit PerkinElmer, Inc. All rights reserved. The PerkinElmer logo and design are registered trademarks of PerkinElmer, Inc. Dynamic Reaction Cell and DRC are trademarks of PerkinElmer, Inc. or its subsidiaries, in the United States and other countries. ELAN is a registered trademark of MDS Sciex, a division of MDS, Inc. All other trademarks not owned by PerkinElmer, Inc. or its subsidiaries that are depicted herein are the property of their respective owners. PerkinElmer reserves the right to change this document at any time without notice and disclaims liability for editorial, pictorial or typographical errors _01 ELEC070400

100 Bromine Speciation by HPLC/ICP-MS Introduction Water for public consumption must be purified prior to distribution. A number of processes are used for water purification, including ozone treatment to kill bacteria. While this method is effective, ozonolysis can also convert bromide (Br - ), a natural component of many waters, into bromate (BrO - 3 ), a carcinogen. Therefore, the need exists to measure bromate in drinking waters, which means that bromate must be determined separately from other forms of bromine. Current methods for measuring bromate and bromide involve separating the bromine-containing components by ion chromatography and using ICP-MS as a detector, monitoring bromine at m/z 79. This is the protocol stated in U.S. Environmental Protection Agency (EPA) Method and also used in Europe. The goal of this work was to develop a faster chromatographic method for bromine speciation than that achieved in Method Experimental HPLC conditions Separation was accomplished using the PerkinElmer Series 200 Quaternary HPLC Pump, Autosampler, Vacuum Degasser and Peltier Column Oven. The column used for this separation was a PRP-x100 anion exchange column (Hamilton, Reno, NV, USA), 15 cm, 4.6 mm ID, 5 µm particles. 50 µl aliquots of samples were injected using a 200 µl PEEK injection loop (Rheodyne L.P., Rohnert Park, CA, USA). The column temperature was maintained at 35 C and inch ID PEEK tubing was used throughout the system. The mobile phase consisted of 18 mm nitric acid (GFS Chemicals, Columbus, OH, USA) and 34 mm ammonium hydroxide (Fisher Scientific, Pittsburgh, PA, USA). The ph was adjusted to 4.0 using dilute nitric acid and ammonium hydroxide. All solutions were made with ASTM Type 1 deionized water. The HPLC conditions are summarized in Table 1. Authors Pamela A. Perrone, Ph.D. Wilhad M. Reuter, Ph.D. Kenneth R. Neubauer, Ph.D. Cynthia P. Bosnak Gerald A. Hall Zoe A. Grosser, Ph.D. PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA HPLC / ICP-MS A P P L I C A T I O N N O T E

101 Prior to analysis, the column was conditioned for 30 minutes with the mobile phase flowing at 1.5 ml/min to establish an equilibrium between the mobile phase and column packing. At the end of each day, the column was washed for 20 minutes with a 95/5 water/methanol solution at 1.5 ml/min to remove any salts. This was followed by a 15-minute rinse with a 70/30 water/methanol rinse, in order to prevent the column from drying out. The column was capped for overnight storage. ICP-MS conditions For speciation analysis, detection of the HPLC eluent was accomplished with an ELAN DRC II ICP-MS system (PerkinElmer SCIEX, Concord, ON, Canada). m/z 79 ( 79 Br + ) was monitored in standard mode (i.e., no reaction gas was used). The instrumental conditions are shown in Table 2. For total-bromine analysis, the same instrumental conditions as those shown in Table 2 were used, except that a one-second integration time was used, and three replicates were acquired for each sample. Standards and samples All standards were made by appropriate dilution of 1000 mg/l stock solutions of bromide and bromate (High Purity Standards, Charleston, SC, USA) with ASTM Type 1 deionized water. External calibrations were used for quantitative determinations of speciated and total bromine. For total-bromine analysis, gallium (10 µg/l) was used as an internal standard. Water samples were collected from various sources and included tap water (both from municipal water supplies and private wells) and bottled water. Acid was not added to any samples, and the only sample preparation involved degassing those bottled water samples which were carbonated. For analysis, samples were not treated or diluted in any manner. Data analysis All chromatographic post-run data analysis was performed with Chromera software (PerkinElmer Life and Analytical Sciences, Shelton, CT, USA). Quantitative results were obtained from peakarea measurements using external calibration curves generated from standards made in deionized water. Table 1. HPLC Conditions. HPLC System PerkinElmer Series 200 Quarternary Pump Autosampler Peltier Column Oven Column Hamilton PRP-x100; 15 cm, 4.6 mm ID, 5 µm Mobile Phase 18 mm HNO mm NH 4 OH ph 4.0 ph Adjustment Dilute HNO 3, NH 4 OH Injection Volume 50 µl Flow Rate 1.5 ml/min Sample Preparation None Samples Various waters (non-acidified) Table 2. ICP-MS Conditions. Instrument Nebulizer Spray Chamber ELAN DRC II Quartz Concentric Quartz Cyclonic RF Power 1500 W Analyte 79Br + Dwell Time 500 ms (speciation) 50 ms (total Br analysis) Figure 1. Chromatograms of a) 10 µg/l bromate/bromide standard and b) bottled water containing bromate and bromide. 2

102 Results and discussion Figure 1 shows chromatograms of a 10 µg/l standard of bromate and bromide and a bottled water sample containing both bromate and bromide. These chromatograms demonstrate that the species can be separated in seven minutes and that there is no retention-time shift when a real sample is analyzed. To test the stability of the chromatographic method, 10 consecutive injections of a bottled water sample containing BrO 3 - and Br - were made. Figure 2 shows these 10 chromatograms overlaid and indicates excellent reproducibility, with standard deviations of 0.5 (BrO 3 - ) and 0.3 (Br - ). These results demonstrate that the method is stable from sample to sample. Calibration curves for both species are shown in Figure 3 and demonstrate the linearity of the method. The curves were applied to quantitative measurements of various water samples. Table 3 shows quantitative chromatographic results for five water samples over 4 days. It should be noted that these samples only contained Br - and no BrO 3 -. Also shown in this table are the results from total Br analysis (i.e., Br analysis with the ICP-MS without an HPLC). Similar data is shown in Table 4 for two water samples Figure 2. Ten overlaid chromatograms from consecutive injections (8 minutes between injections) of a bottled water sample. Table 3. Quantitative Determination of Bromide in Water Samples (µg/l). Day Sample Total Br Bottled Water Well Water Well Water Well Water Tap Water No BrO 3 - detected which contained both BrO 3 - and Br -. In both tables, the day-to-day reproducibility of the chromatographic results demonstrates the stability of the method, while the agreement between the speciated and total-bromine results demonstrates accuracy. Figure 4 shows a chromatogram of a standard containing 1 µg/l bromate and bromide. The peaks for both species are clearly visible above the baseline, demonstrating that 1 µg/l levels of both species can be determined. Conclusions This work has shown the ability to separate bromide and bromate with ion exchange chromatography. By selecting the appropriate mobile phase and column, the separation can be accomplished in seven minutes. Detection was accomplished with ICP-MS monitoring m/z 79 ( 79 Br + ) in the standard mode, which yields detection limits of 1 µg/l for both bromate and bromide. Because the standard mode of analysis was used, comparable results can be expected using the ELAN 9000 ICP-MS. The methodology is demonstrated to be rugged and stable through replicate injections and by comparison of results for drinkingwater samples over several days. The accuracy of the method is shown by the similarity of the speciated results with those from total-bromine analysis. References 1. J.T. Creed, C.A. Brockhoff, T.D. Martin U.S. EPA Method Determination of Bromate in Drinking Waters by Ion Chromatography Inductively Coupled Plasma-Mass Spectrometry,

103 Table 4. Quantitative Determination of Bromide and Bromate in Water Samples (µg/l). Day Bottled Water Br BrO Total Br (sum) Total Br (expt) 27.4 Day Bottled Water Br BrO Total Br (sum) Total Br (expt) 28.3 Figure 3. External calibration curves for bromate and bromide. Figure 4. Chromatograms of a standard containing 1 µg/l each of bromate and bromide. PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA Phone: (800) or (+1) For a complete listing of our global offices, visit PerkinElmer, Inc. All rights reserved. The PerkinElmer logo and design are registered trademarks of PerkinElmer, Inc. Chromera is a trademark and PerkinElmer and TotalChrom are registered trademarks of PerkinElmer, Inc. or its subsidiaries, in the United States and other countries. PerkinElmer SCIEX is a trademark of PerkinElmer, Inc. ELAN is a registered trademark of MDS Sciex, a division of MDS, Inc. All other trademarks not owned by PerkinElmer, Inc. or its subsidiaries that are depicted herein are the property of their respective owners. PerkinElmer reserves the right to change this document at any time without notice and disclaims liability for editorial, pictorial or typographical errors _01 KG Printed in USA

104 20 Simultaneous Arsenic and Chromium Speciation by HPLC/ICP-MS in Environmental Waters Kenneth R. Neubauer, Wilhad Reuter, Pamela Perrone, Zoe Grosser PerkinElmer Life and Analytical Sciences INTRODUCTION Many elements can exist in a variety of oxidation states which have differing impacts on health and the environment. As a result, it is necessary to quantify the individual oxidation states of these elements within a sample rather than simply the total element content for an accurate assessment of their impact. Two elements in this category are chromium and arsenic. Trivalent chromium (Cr III) is an essential nutrient, while hexavalent chromium (Cr VI) is toxic, does not occur naturally, and results only from anthropogenic activities. Arsenic exists in a variety of forms with the trivalent form (As III) being the most toxic, followed by the pentavalent form (As V). Other common forms of arsenic include monomethyl arsenic (MMA), dimethyl arsenic (DMA) and arsenobetaine (AsB). The speciation and quantification of both chromium and arsenic are best accomplished by using HPLC to separate the species and ICP-MS to detect them. However, HPLC separations can take ten minutes per element and require different mobile phases to separate different elements. Under these constraints, each sample would have to be analyzed twice to determine both the chromium and arsenic species present. A significant problem encountered when attempting to measure low levels of chromium and arsenic in environmental waters is matrix interference. Carbon, chloride and calcium are common elements typically found in these types of samples. Carbon (ArC + ) interferes with the chromium isotopes while chloride (ArCl +, CaCl + ) interferes with the only arsenic isotope. Consequently, elimination of these interfering matrix components is required to enable lowlevel detection of chromium and arsenic. The goal of this work was to develop a method for simultaneous chromium and arsenic speciation in environmental water samples in less than five minutes. With this approach, each sample would only have to be analysed once. ICP-MS with a Dynamic Reaction Cell (DRC ) was chosen as the detector because of its ability to provide the lowest possible detection limits by eliminating the effects of common interferences. It should be noted that a current concern regarding speciation is the preanalysis preservation of species in a sample. It is known that several species readily interconvert between oxidation states so that the concentration of species analysed is not necessarily representative of that which exists in the environment. This is a separate area of research and debate and beyond the scope of this work. EXPERIMENTAL HPLC Conditions Separation was accomplished using the PerkinElmer Series 200 Quaternary HPLC Pump, Autosampler and Vacuum Degasser. A 3 cm Pecosphere column with 3 µm C8 packing (Part No , PerkinElmer Life and Analytical Sciences, Shelton, CT USA) was used for the separation. Other details of the HPLC conditions are given in Table 1. The mobile phase consisted of 1mM tetrabutylammonium hydroxide (TBAOH) and 0.5 mm ethylenediaminetetraaceticacid dipotassium salt dihydrate (EDTA) (Aldrich Chemical Company, Milwaukee, WI USA), and 5% methanol (Fisher Scientific, Pittsburgh, PA USA). The ph was adjusted to 7.2 using dilute nitric acid and ammonium hydroxide (Fisher Scientific). All solutions were made up in 18 M-ohm distilled deionised water. These conditions were chosen because they provided the best compromise between peak shape and retention time for all of the chromium and arsenic species investigated. Prior to analysis, the column was conditioned for 30 minutes with the mobile phase flowing at 1.5 ml/min; this was required to properly equilibrate the column. Upon completion of analyses at the end of each day, the column was washed for 15 minutes with a 5/95 mixture of methanol/water to remove the buffer/salts from the column, followed by a 15 minute wash with a 70/30 methanol/water mixture to prevent the column from drying out. For overnight storage, the column remained connected to the HPLC system; the ends of the column were capped for long-term storage. ICP-MS Conditions Detection of chromium and arsenic species was accomplished with an ELAN DRC II (PerkinElmer SCIEX, Concord, ON Canada); detailed instrumental conditions are given in Table 2. Oxygen was chosen as the reaction gas because it reduces the ArC + interference on Cr + at m/z 52 and reacts readily with As + to form AsO + (m/z 91), a new species which does not suffer from ArCl + and CaCl + interferences at m/z 75. Total chromium and arsenic concentrations were also measured without speciation, using conventional nebulization into the ICP-MS. For these analyses, all instrumental parameters were the same as for the speciation analysis except for the reaction cell conditions, which are shown in Table 3. The reaction cell conditions differ between analyses because of the different matrices involved: for speciation, the matrix is the mobile phase; for total analyses, the matrix is the water samples.

105 21 Standard and Sample Preparation All standards and samples were prepared in a solvent mixture similar to the mobile phase (1 mm TBAH mm EDTA; ph = 7.2) and allowed to sit for at least 30 minutes prior to analysis. Samples were diluted by at least a factor of two. Higher dilution factors were used for certain samples containing high analyte levels, so that the final concentration was within the range of the calibration curve. It was necessary for the diluted samples to sit for at least 30 minutes in mobile phase so that all species were equilibrated with the mobile phase. All standards were made by dilution of the following 1000 mg/l stock solutions: chromium (III) and arsenic (V) (PE Pure, PerkinElmer Life and Analytical Sciences, Shelton, CT USA), and chromium (VI) and arsenic (III) (SpexCertiprep, Metuchen, NJ USA). Samples were obtained from various sources (see Tables 4 and 5) and collected in plastic bottles (polyethylene or polypropylene), without acid preservation. Lake and river waters were filtered by gravity through Whatman 40 filter paper prior to analysis to remove particulate material. Bottled waters were purchased in a local grocery store and allowed to degas at room temperature. Residential waters were collected directly from faucets of private homes. Calibration curves were established with 0.5, 1.0, and 5.0 µg/l standards; each species yielded an R 2 > 0.999, thus demonstrating the linearity of the technique. Peak areas were used for all quantitative measurements. Sample chromatograms appear in Figures 2-4 which are representative of the samples analyzed. Figure 2 shows a municipal water supply sample that contains a high level of chromium and low levels of arsenic; Figure 3 is a residential well-water sample containing elevated arsenic and low-level chromium; Figure 4 is another municipal water supply sample containing low levels of both chromium and arsenic. Figure 2. Chromatogram of Glendale, CA municipal water sample. The concentrations shown are those read by the instrument; they are not corrected for dilution. An interesting aspect of Figure 6 is that the As (V) peaks do not occur at the same retention time, a reproducible phenomenon. When a total-metal analysis was performed on these samples, it was noticed that they contained differing levels of minerals (i.e. Na, Mg, K, Ca, Fe) with Water 1 having the highest mineral content and Water 3 containing the lowest level. Figure 5. Chromium chromatograms from three residential well waters in Danbury, CT. RESULTS AND DISCUSSIONS Figure 1 shows a chromatogram of 1 µg/l chromium and arsenic standards. These selected ion chromatograms were acquired simultaneously by monitoring two different masses on the ICP-MS: m/z 52 for Cr+ and m/z 91 for AsO+. This figure clearly shows that all four species can be detected simultaneously within a single analysis in less than three minutes. Figure 3. Chromatogram of a residential well-water sample. The concentration shown is that read by the instrument; it is not corrected for dilution. Figure 6. Arsenic chromatograms from three residential well waters in Danbury, CT. A subsequent study was performed to explore the effect of salt concentration on the retention times of the chromium and arsenic species. Chromium and arsenic standards (at 1µg/L) were made in 25, 100 and 500 mg/l sodium chloride solutions and analysed. Figure 1. Chromatogram of four chromium and arsenic species separated and quantitated in a single analysis. Each species is present at 1µg/L. For these studies, oxygen was chosen as the reaction gas for the removal of ArC +. Although oxygen reacts readily with ArC +, it is known that ammonia is much more efficient at removing this interference. However, for this study, ammonia was not used since it also reacts strongly with As+, thus removing it from the ion stream. Therefore, the elevated chromium baseline that was observed in Figure 1 resulted from the incomplete removal of ArC+ as well as chromium contamination in the mobile phase. Although the elevated baseline may appear to be problematic, it did not affect the signalto-noise ratio (S/N) for the chromatographic peaks. Subsequent data shows that 100 ng/l Cr could still be detected. Figure 4. Chromatogram of Shelton, CT municipal water sample. The concentrations shown are those read by the instrument; they are not corrected for dilution. Figures 5 and 6 show chromium and arsenic chromatograms obtained from 3 residential well waters in the same city (Danbury, CT USA). Figure 5 shows that the amount and species of chromium vary among the residences, while Figure 6 demonstrates that each water supply contained only As (V), all at about the same quantity. These results show how the levels and species of chromium and arsenic can vary within a community.

106 22 Figure 7. Effect of salt concentration on chromium speciation. Each chromium standard is present at 1 µg/l. Figure 8. Effect of salt concentration on arsenic speciation. Each arsenic standard is present at 1 µg/l. The results appear in Figures 7 and 8 and show that the salt concentration of a sample affects the retention times of Cr (III) and As (V). As the salt concentration increases, the retention times of Cr (III) and As (V) increase, while the retention times of Cr (VI) and As (III) are unaffected. This trend explains the observation in Figure 6. The effect of salt concentration can probably be mitigated by increasing the ionic strength (i.e., adding a buffer) to the mobile phase. This will be the focus of a future study. Two other common arsenic species usually studied along with As (III) and As (V) are monomethyl arsenic (MMA) and dimethyl arsenic (DMA). Using a chromatographic method which separates these four arsenic species, it was found that MMA and DMA were not present in any of the water samples analyzed. Therefore, the focus of this work included only the arsenic species which were observed: As (III) and As (V). Table 1. HPLC System HPLC Conditions PerkinElmer Series 200 Quartenary Pump column oven, Autosampler and Vacuum Degasser Column Temp 35 C Column Pecosphere C8; 3 µm particles; 3 cm Mobile Phase 1 mm TBAOH mm EDTA (potassium salt) + 5% methanol ph 7.2 ph Adjustment Dilute HNO 3, NH 4 OH Injection Volume 50 µl Flow Rate 1.5 ml/min Auto Sampler Flush Solvent 5% methanol Table 2. ICP-MS Conditions Instrument ELAN DRC II (PerkinElmer SCIEX) Nebulizer Quartz Concentric Spray Chamber Quartz Cyclonic RF Power 1500 W Analytes Cr + (m/z 52); AsO + (m/z 91) Reaction Gas O 0.6 ml/min RPq 0.55 Dwell Time 500 milliseconds (per analyte) Analysis Time 150 seconds Table 3. Reaction Cell Conditions for Total As and Cr Determination Analyte m/z Reaction Gas Flow (ml/min) RPq Cr 52 NH AsO 91 O Table 4. Chromium Results Sample Cr (III) Cr (VI) Total Cr (µg/l) (µg/l) (µg/l) Connecticut River 0.07 Lake Mohegan Shelton, CT Water Glendale, CA Water Oxford, CT Water Bottled Water A Bottled Water B Bottled Water C = None detected Table 5. Arsenic Results Sample As (III) As (V) Total As (µg/l) (µg/l) (µg/l) Connecticut River Lake Mohegan Shelton, CT Water Glendale, CA Water Oxford, CT Water Bottled Water A Bottled Water B Bottled Water C = None detected CONCLUSIONS This study demonstrated the feasibility of rapid, simultaneous chromium and arsenic speciation in environmental waters using HPLC/ICP-MS. By careful selection of chromatographic conditions, Cr (III), Cr (VI), As (III) and As (V) can be separated in a single run in under three minutes. Low-level detection is accomplished using DRC ICP-MS to eliminate the effects of interferences on chromium and arsenic. This method was then applied to a variety of environmental water samples from rivers, lakes, municipal water supplies and residential well waters, as well as from bottled waters. Chromium and arsenic levels below 100 ng/l were measured, as shown in the chromatograms of water samples. Judging from the signal-to-noise ratios observed in these chromatograms, lower levels can probably be detected. The results from this work demonstrated that this method can serve as a rapid screening tool to quantitatively determine these species in a number of water sources. Illuminating Concepts AstraNet, the Cambridge-based suppliers of fibre optic instrument-ation, have dramatically extended their range of technologies by an exclusive distribution agreement with Sciencetech Inc, of Canada. Founded in 1985, Sciencetech specialise in novel spectroscopic instrumentation based on their unique "Building Block" concept and an impressive catalogue of light sources, mono-chromators, detectors and other components. Individual modules or complete, integrated systems to customer specification can be supplied. The extraordinary flexibility available to end users or self-build instrument designers is apparent from the Sciencetech range of arc lamps, a wide choice of Xenon, Mercury and Mercury-Xenon lamps with outputs from 250W to 1.5 KW, complemented by lamp housings, power supplies and a choice of condensing systems. For fibre optics and other applications that require imaging the arc into a small, bright focus, a range of refractive condensers is available. Alternatively, ellipsoidal reflectors with a range of focal lengths can be supplied. One application where the company has particular expertise is that of solar simulation, where a high pressure Xenon arc is combined with special filters to approximate to solar radiation, for applications such as artificial ageing, photovoltaics or photochemistry. Four Solar Simulation Systems are available. The lower power ( W) versions can be configured to produce either a collimated 1" diameter beam or for fibre optic illumination. The 1KW and 1.5KW models produce an 8" diameter illuminated area as standard. A unique optical design uses mirrors to "fold" the light from lamp to target, providing 30% more output power than competitive systems using diffusers. Circle no. 748

107 RESEARCH COMMUNICATIONS Radionuclide and trace element contamination around Kolaghat Thermal Power Station, West Bengal Environmental implications A. Mandal and D. Sengupta* Department of Geology and Geophysics, Indian Institute of Technology, Kharagpur , India The Kolaghat Thermal Power Station (KTPS), located at Mecheda in Midnapur District, West Bengal is one of the largest thermal power stations in eastern India. Combustion of coal in the KTPS generates large quantities of ash that are disposed-off in vast areas of land adjoining the power station. Geochemical and radiometric analysis of the pond ash has been undertaken to assess the quantity of toxic elements that are likely to contaminate the soil and the groundwater system. Trace element analysis reveals that toxic elements (Pb, Cu, Ni, Fe, As) are sufficiently enriched in pond ash than their crustal abundances, and preferably in the lighter size fractions. Radionuclides (U, Th) also show enrichment of 3 5 times in coal ash compared to their crustal average and are much higher than in the pond ashes of other thermal power plants in India. Chemical analysis of the water samples collected from tubewells near the ash ponds reveals high concentration of trace elements (Al, Li, Ni, Fe, As, Zn, B, Ag, Sb, Co, Si, Mo, Ba, Rb, Se, Pb V, Cr, Cu, Cd, Mn, Sr), whose distribution is mainly controlled by the ash deposited in the area. Among these elements, Al, Li, As, Zn, Ag, Sb, Si, Mo, Ba, Rb, Se and Pb show higher concentration in the tubewell waters near the ash pond, implying significant input from the ash pile. The enrichment of some elements (Al, Fe, As and Mn) above WHO guidelines for drinking water denotes significant contamination of the groundwater from the toxic elements leached from the ash pile. THERMAL power generation contributes to more than 70% of the power generation in the country 1. Indian coal is of bituminous type, with 55 60% ash 1. Combustion of coal thus generates huge amount of ash which are disposed-off either in dry or slurry form. The Kolaghat Thermal Power Station (KTPS) uses the wet disposal method. Ash generated from coal combustion has a greater tendency to absorb trace elements that are transferred from coal to waste products during combustion due to its small size and hence, large surface area 2. Coal is also radioactive due to primordial 238 U, 232 Th and 40 K. Earlier work 3 on coal and ash has shown that Indian coals contained ppm 238 U and ppm of 232 Th. But recent studies 4 have shown that pond ash generated from coal contains as high as 50 ppm 232 Th and 10 ppm of 238 U. *For correspondence. ( [email protected]) Groundwater is vulnerable to metal contamination due to waste disposal and leachate percolation 5 7. Coal-fired power stations have recently come under scrutiny as potential sources of mercury and other heavy metal pollutants. When large quantities of ash accumulate for long periods of time in the disposal site, hazardous substances are likely to be released by leaching, percolate through the soil layers and eventually reach the groundwater. The KTPS, situated in Midnapur District, West Bengal, India is the second largest power station in the state, with an installed capacity of 1260 MWe/day, generating 1210 MWe/ day of electricity. Figure 1 a gives the location map of the study area along with the power station. This station comprises six units of 210 MWe each. The ash is deposited in four large ash ponds (1A, 1B, 4A, 4B), located 4 5 km south of the power station (Figure 1 b). The major source of potable water for the villages surrounding the ash pond is the Kasai river, and the tributaries of the Rupnarayan river (Figure 1 a). In the present investigation an attempt has been made to study the trace element geochemistry, radioactivity in coal and ash, enrichment of radionuclides from coal to ash and trace elements in water to assess the potential impact of ash disposal on the quality of groundwater. Coal samples were collected from inside the power station, shortly before being fed into the boiler. The ash samples were collected from the ash ponds 1A and 1B, since the other two ash ponds, i.e. 4A and 4B were already full of water and the newly produced ash from the power station was being dumped in slurry form. After collection, the ash samples were air-dried at C for about 24 h. Water samples were collected in 200 ml polyethylene bottles from tubewells located near the ash ponds 1A, 1B, 4A, 4B (numbered as 1A1, 1A2, 1B, 4A1, 4A2, 4B) and from the surrounding villages of Bahala (B1, B2), Raksha (R1, R2, R3, R4), and Mecheda (M1, M2, M3). The sampling bottles were soaked in 10% HNO 3 for 24 h and rinsed several times with deionized water prior to use. The samples were filtered in the laboratory with a Whatmann 42 filter paper. After filtration, each sample was acidified with 2 ml HNO 3 to prevent any precipitation of metals. Elemental analysis of each water sample was carried out using the Inductively Coupled Plasma Mass Spectroscopy (ICP MS) MODEL Perkin-Elmer Sciex ELAN DRC II (Toronto, Canada) at the Central Research Facility available at National Geophysical Research Institute (NGRI), Hyderabad. Radon measurements were performed by LSC technique with the ultra low-level Quantulus 1220 TM (Wallac Qy). Samples were prepared with 15 ml water samples and 5 ml RADONS (Etrac Lab.) following the procedure at our laboratory. The counting time was 200 min. Laboratory measurements of 238 U, 232 Th and 40 K in coal and ash samples were undertaken using a low-level gamma-ray spectrometric set-up at NGRI. The detector is a 5 6 NaI (Tl) crystal coupled to a 5 diameter photomultiplier tube. In closed systems, the activity concentrations of nuclei in the decay chain reach secular equilibrium 8, meaning that CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY

108 RESEARCH COMMUNICATIONS a b Figure 1. Location map of (a) study area and (b) sites of collection of water samples. the activity concentrations of all nucleides are the same. As a result, measuring the activity concentration of one member in a closed system provides information on the presence of all the members 9. After drying, the samples were sealed 618 and stored an airtight container to prevent escape of radiogenic gases 222 Rn and 220 Rn, and to allow attainment of radioactive equilibrium in the decay chain. After attainment of secular equilibrium between 232 Th, 238 U and their daughter CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY 2005

109 RESEARCH COMMUNICATIONS products, the samples were subjected to gamma-ray spectrometric analysis. Natural radionuclides of relevance to this work are mainly 232 Th, 238 U and 40 K. While the concentration 40 K can be measured directly by its own gamma-rays, 232 Th and 238 U are not gamma-ray emitters. Concentration of 238 U and 232 Th was assessed from the intensity of the gamma lines of their daughters: 214 Bi (1.76 MeV) and 208 Tl (2.62 MeV), assuming all daughter products are in equilibrium with their parents. Concentration of 40 K was measured from the 1.46 MeV peak. Assaying was carried out by the method followed at NGRI 10. The relationship between the number of atoms of a certain species, N, and its activity, A, is defined as N = T 1/2 A/ln2, (1) where T 1/2 is the half-life of the radionuclide. Since activities are expressed in becquerel (1 Bq corresponds to one decaying atom/s), T 1/2 has to be expressed in seconds. Using eq. (1) and Avogadro s number 9,11, one notes that 1 ppm 238 U and 232 Th corresponds to 12.3 and 4.0 Bq kg 1 respectively, of 238 U and 232 Th. Similarly, one notes that 1% 40 K corresponds to 310 Bq kg 1 40 K. In this study activity concentrations are expressed in Bq kg 1 using the conversion factors for conversion of ppm of 238 U and 232 Th to Bq kg 1. Results of chemical analysis show that the feed coal is dominated by SiO 2, Al 2 O 3 followed by Fe 2 O 3, K 2 O, TiO 2, CaO, MgO, and Na 2 O (Table 1). Coal from KTPS is thus aluminosilicious in nature. Chemical analysis of the ash sam- Table 1. Bulk chemical composition of feed coal and pond ash from KTPS Average Average Compound Unit coal (pond ash) Na 2O wt% MgO wt% Al 2O 3 wt% SiO 2 wt% P 2O 5 wt% K 2O wt% CaO wt% TiO 2 wt% MnO wt% Fe 2O 3 wt% C wt% S wt% Be ppm Sc ppm V ppm Co ppm Ni ppm Cu ppm Zn ppm As ppm Rb ppm Sr ppm Zr ppm Ba ppm Pb ppm ples of Kolaghat shows that the ash is dominantly composed of oxides of silica, alumina and iron with minor amounts of TiO 2, K 2 O, CaO, MgO, MnO and Na 2 O (Table 1). Trace element concentration of the ash shows that the ashes of Kolaghat contain sufficient amounts of As, Cu, Pb, Ni, Zn, Co, V, Sc, Be, Cs and Zr relative to the feed coal (Table 1). Combustion of coal causes decomposition of organic and inorganic matter. Trace elements associated with the organic material get released and accumulate in the refractory phases, e.g. mullite and other aluminous phases as clays are the highest repositories of trace elements. Enrichment of trace elements in ash relative to the crustal abundance has been determined by the enrichment factor (EF) which is defined as the ratio of the concentration of an element in coal with respect to its concentration in the crust 12. In the present study EF was calculated by normalizing the concentration of elements with that of Al, since it is considered to be non-volatile under combustion conditions 13. Thus EF (Ash) relative to the crust is defined as: EF (Ash) = (C X /C Al ) Ash /(C X /C Al ) Crust, (2) where (C X /C Al ) Coal = (Concentration of X /Concentration of Al) Ash/Crust, X = Element. EF of the elements in pond ash relative to their crustal abundance was also determined using eq. (2) and are shown in Table 2. It is seen that among the trace elements, As, V and Pb show maximum enrichment compared to their crustal abundances. These are volatile elements and are generally associated with organic matter in coal. Combustion of coal results in release of these elements, which on cooling condense down the tract and get deposited on the ash particles. Radioactivity of coal and ash samples from KTPS are given in Table 3. The data show that the activity concentrations of 238 U and Th 232 range from to Bq kg 1 and from Bq kg 1 to Bq kg 1 in ashes, with a mean of and Bq kg 1 respectively. 238 U and 232 Th in feed coal range from to Bq kg 1 and from to Bq kg 1, with a mean of and Bq kg 1 respectively. The activity concentrations of 232 Th are much higher than those of 238 U in KTPS coal. Combustion of coal enhances the radioactivity in ash, with 232 Th activity concentrations exceeding those of 238 U. 40 K in ashes ranges from to Bq kg 1, with a mean of Bq kg 1. The activity concentrations of radionuclides are high compared to those in fly ash from the thermal power stations in other parts of India, e.g. in Uttar Pradesh 14, where the activity concentrations of 226 Ra (daughter product of 238 U) is Bq kg 1 and that of 232 Th is Bq kg 1. The enrichment factor (EF R ) of radionuclides in ash as computed by eq. (2) is 3.5 for 238 U and 5 for 232 Th. Thus thorium concentrations in ash from Kolaghat are five times higher compared to their crustal abundance. The total gamma dose emitted from the ash pond has also been calculated CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY

110 RESEARCH COMMUNICATIONS Table 2. Enrichment factors of elements in pond ash from KTPS with respect to their crustal abundance Element Be As V Co Ni Cu Zn Rb Sr Zr Pb Ba Sc Enrichment factor (ash) Table 3. Radioactivity in pond ash and coal samples from KTPS 238 U 232 Th S. No. Location (Bq/kg) (Bq/kg) (Bq/kg) IIT-KAP 1 Ash pond IIT-KAP 2 Ash pond IIT-KAP 3 Ash pond IIT-KAP 4 Ash pond IIT-KAP 5 Ash pond IIT-KAP 6 Ash pond IIT-KAP 7 Ash pond IIT-KAP 8 Ash pond IIT-KAP 9 Ash pond IIT-KAP 10 Ash pond IIT-KAP 11 Ash pond IIT-KAP 12 Ash pond IIT-KAP 13 Ash pond IIT-KAP 14 Ash pond No. of samples Max Min Average Standard deviation IIT-KC 1 Kolaghat (coal) IIT-KC 2 Kolaghat (coal) IIT-KC 3 Kolaghat (coal) IIT-KC 4 Kolaghat (coal) IIT-KC 5 Kolaghat (coal) No. of samples Max Min Average Standard deviation using dose conversion factors of UNSCEAR 15. The total gamma dose rate in air at 1 m above the ground at each location in the ash pond is calculated using the following equation: D = (0.462C U C Th C K )ngyh 1, (3) where C U, C Th, C K are the activity concentrations (Bq/kg) of 238 U, 232 Th and 40 K in the ash. The mean dose rate as computed from the above equation was found to be ngyh 1, which is about three times higher than the world average 15 of 51 ngy 1. Thus the population within 80 km radius of the ash pond is exposed to a high dose rate. Details of the concentration of the major and trace elements in the groundwater samples near the ash ponds and at the surrounding villages are given in Table 4. ph of the water samples ranged from 7.02 to 8.70, indicating alkaline nature of the water. The elements found in highest concentration 40 K in the tubewell waters were Ca followed by Na, Mg and K. Na and K are present in higher amounts in the tubewell waters in the villages than those near the ash pond. Ca, on the other hand, is enriched in the tubewell waters of the ash pond, implying a significant input from the ash pile. A comparison (Figure 2) of data is also made between the concentration of the trace elements in the waters of the tubewells near the ash pond and those of the tubewells in the villages. In Figure 2, TAP denotes the average concentration of elements in the ash pond tubewells (1A1, 1A2, 1B, 4A1, 4A2, 4B) and TV is the average concentration of the elements in the tubewells (B1, B2, R1, R2, M1, M2, M3) at the villages Bahala, Raksha and Mecheda. Among the trace elements, Al, Li, As, Zn, Ag, Sb, Si, Mo, Ba, Rb, Se and Pb show higher concentration in the water samples collected from the tubewells (TAP) than those collected from Bahala, Raksha and Mecheda (TV; Figure 2). Other trace elements V, Cr, Cu, Cd, Mn and Sr show higher concentration levels in TV than in TAP (Figure 2). The concentrations of Ni, Fe and Co are almost the same in the tubewell waters in all locations (Figure 2). The results of the radiometric analysis of some of the water samples are shown in Table 5. The data show that the tube wells (4A1) and (4B) located near the ash ponds 4A and 4B yield high activity of radon than the tube wells (R1 and M1) located at some distance from the ash ponds. SQP(E) is the quenching parameter which gives us an idea of the presence of molecules such as organic or suspension material in the groundwater sample that could affect the measurements. It is done by the Quantulus and high values in the studied samples mean absence of these undesirable molecules. Radon is a daughter product of 238 U. The 238 U concentration in ash samples has been found to exceed the crustal abundance by a factor of three. Since uranium is a soluble radionuclide, it might have been leached out from the ash pile and hence the high concentration of radon in the well waters of the ash pond implies significant input from the ash disposal site. Table 6 shows the maximum permissible limits for various elements in drinking water. Table 7 shows the respective EFs of the different elements compared to the concentration of the corresponding elements in the WHO guidelines 16. Mn, Pb show maximum enrichment in all the tubewell waters, while As and Al exceed the limit at a few places. The enrichment of Fe, Ni, Ba is close to 1 in most places indicating that their concentrations are well within the WHO limits. The tubewell waters near the ash ponds show maximum enrichment in the elements compared to the other places. This is especially true for Al, Mn, Pb and As, implying significant input from the ash pile. For other locations although the enrichment values are close to 1 and slightly > 1, they CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY 2005

111 RESEARCH COMMUNICATIONS Table 4. Concentration of different elements (ppm) in tubewell waters near ash ponds and surrounding villages 4B 4A1 4A2 1A1 1A2 1B B1 B2 R1 R2 R3 R4 M1 M2 M CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY

112 RESEARCH COMMUNICATIONS Figure 2. Variation in concentration of trace elements in water collected from tube wells located near the ash pond (TAP) and those in the nearby villages (TV). 622 Table 5. Radon activity concentration and quenching parameter Rn activity SQP(E) Sample (Bq/l) (quenching parameter) 4A ± R ± M ± B ± Table 6. Element WHO guidelines for drinking water 17 Maximum permissible limit (ppm) Al 0.2 As 0.01 Ba 0.7 Cr 0.05 Cu 2 Fe 0.3 Mn 0.1 Ni 0.02 Pb 0.01 Zn 3 are likely to exceed the WHO 16 guidelines in future with unabated input of ash particles rich in Fe, Ni and Ba. The high concentration of trace elements in waters of the tube wells (1A1, 1A2, 1B, 4A1, 4A2, 4B), possibly, can be attributed to the leaching of the elements from the ash pile and subsequent mixing of the ash-pond leachate with groundwater. In the present study, it is seen that the ashes contain high concentrations of Al, Ni, Fe, As, Zn, Mo, Mn, Ba and Pb, which exceed their crustal abundance by a factor of 2 3. Hence there are chances of these elements being leached out from the ash pile by the percolating rainwater and other sources of groundwater. Higher concentrations of these elements in tube-well waters of the ash ponds could be attributed to their input from the wind-blown ash particles. It is postulated that the elevated concentration of these elements in the well waters is due to leaching from the ash pile, contributing to the dissolved element load of groundwater. The analyses presented above show that combustion of coal in KTPS causes significant amount of environmental pollution, with the generation of ashes that are high in radioactive and other toxic trace elements compared to other thermal power stations in India (Table 8). Earlier workers 17,18 have measured the activity of 226 Ra and 228 Ac (daughter product of 232 Th) in the ash samples. Since coal contains naturally occurring radioactive materials in radioactive equilibrium, measuring the activity concentration of one member in a closed system provides information on the presence of all other members 9. Hence in the present study the activity concentrations of 238 U and 232 Th were compared with those of their daughter isotopes ( 226 Ra and 228 Ac). Uranium, being relatively more soluble than thorium in water, can be leached out by the percolating water. On the other hand, thorium migrates as a mineral fragment. The dose, emitted from the ash pond to the surrounding 15 is about three times higher than the world average of 51 ngy 1. The natural radiation dose to population from thermal power plants is slightly higher than that from nuclear power plants of the same capacity, due to CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY 2005

113 RESEARCH COMMUNICATIONS high ash content and higher population density around such plants 1. The target organs of the radionuclides 238 U and 232 Th are bones and lungs. Increased incidence of leukaemia, bone sarcomas and chromosomal aberrations are due to the radiotoxicity of thorium. Study of 29 thermal power stations has shown 18 the collective effective dose equivalent commitments from doses to bones, lungs and thyroid is 206 man Sv.y 1 and from doses to the whole body of 73 man Sv. y 1. Lalit et al. 18 have taken into account, the installed capacity of the power stations, the population density per km 2 of the power stations and the concentration of radionuclides in the ash. The power stations studied 18 have installed capacity in the range MWe. On the other hand, the KTPS has an installed capacity of 1260 MWe. The ash released per year is 44%. Moreover, the population density around the power plant is 767 per km 2. Concentrations of radionuclides ( 238 U = 111 Bq/kg and 232 Th = 140 Bq/kg) in the ashes from KTPS are much higher than those of radionuclides in the ashes of other thermal power stations (Table 8). Considering the high population density, ash content and radioactivity in ash, disposal of ash from the power station results in similar high effective doses to the population near the ash ponds and the surrounding villages. Moreover, considering the high absorbed dose rates (150 ngy h 1 ) from the ash pond, disposal of ash may cause radiation hazard to the population in the neighbouring villages of the KTPS. Prolonged exposure to the high dose rates may lead to risk in lung and bone cancer. The bricks made of ash from KTPS showed high activity concentrations of 232 Th ( Bq/kg) and 238 U (87.33 Bq/kg) compared to those of ordinary bricks ( 232 Th, 83.5 Bq/kg and 238 U, Bq/kg). Hence if we take into account the 4π geometry for dwellings made of fly-ash bricks, the total radiation effect will get enhanced and could lead to serious health hazards. Nuclear power plants emit radioactive and negligible amounts of gaseous or particulate pollutants. The risk associated with nuclear radiation is probabilistic. On the other hand, Table 7. Enrichment of elements in tube-well waters with respect to those according to WHO guidelines S. No. Al Cr Mn Fe Ni As Ba Pb 4B A A A A B B B R R R R M M M thermal power plants using fossil fuels produce particulates, oxides of sulphur, nitrogen, carbon and toxic metals like arsenic, mercury, etc. in trace concentrations. The health risk of all these is deterministic 1. Organ dose commitments are greater in the case of thermal power stations compared to nuclear power plants, mainly because the α-emitting bone seekers such as 226 Ra, 228 Th and 228 Ra (through its α-emitting daughter products) released from thermal power plants result in higher organ doses, while β-emitting noble gases and fission products give more external exposure to the whole body due to their high emission rates. Thus, overall dose commitments due to coal production are comparable with those from heavy-water reactor-type nuclear power plants. The doses computed are mainly due to inhalation during the passage of the cloud containing radioactive emissions and from external radiation due to the activity deposited. There is no universally acceptable method for assessing the health risks from conventional pollutants and nuclear radiations. However, if all the gaseous and particulate emissions are considered, coal-based plants have a higher health risk to the population. Carbon dioxide emission from coal giving rise to global warming is almost absent in nuclear power plants 1. Geochemical study of the water samples shows that the tube-well waters near the ash pond and in the surrounding villages are contaminated by toxins released from the ash pile. The enhancement in concentration of toxins (Al, Ni, Fe, As, Zn, Mo, Mn, Ba, Pb) in the tube-well waters near the ash pond denotes significant input from the ash pond. Mn and Pb shows significant enrichment in the waters, exceeding the WHO guideline values for drinking water. Al and As Table 8. Radioactivity in pond ash from different thermal power stations in India Thermal power station Average activity Concentration (Bq/kg) 226 Ra 228 Ac 40 K Reference Allahabad (Uttar Pradesh) Angul (Orissa) Badarpur (Delhi) Bakreshwar (West Bengal) Chandrapur (Madhya Pradesh) Farakka (West Bengal) Raichur (Karnataka) Talchir (Orissa) Bokaro (Bihar) Ramagundam (Andhra Pradesh) Neyvelli (Tamil Nadu) Amarkantak (Madhya Pradesh) Bandel (West Bengal) Indraprastha (Delhi) Durgapur (West Bengal) Korba (Madhya Pradesh) Nasik (Maharashtra) CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY

114 RESEARCH COMMUNICATIONS show enrichment above the WHO guideline value in some places, mainly in the tube wells located near the ash ponds 4A, 4B and 1A (Table 7). Excess amount of arsenic and manganese affects the cardiovascular system, gastrointestinal tract, kidney, liver, skin and blood and prostrate. Excess amount of lead could lead to dysfunctioning of the kidney, gastrointestinal tract and respiratory systems due to inhalation of fine ash particles rich in lead. The present study shows many potential environmental hazards related to coal combustion. Hence proper measures should be taken to check the release of toxins from the ash pond and subsequent mixing with the groundwater. A remedy is to have underground lining in the ash ponds to prevent direct contact of the ash pile with the top soil and the local drainage bodies. Use of bricks from fly ash meant for dwellings should be curtailed in the absence of detailed studies on the indoor radiation dose and its effect on the inhabitants for prolonged exposure. The radiation dose from thermal power plants can be reduced by reducing the ash content of coal and establishing thermal power plants away from populated area. 15. UNSCEAR, Sources and effects of ionizing radiation, United Nations Scientific Committee on the Effects of Atomic Radiation, United Nations, New York, WHO, Water, sanitation and health: Guidelines for drinking water quality, Vijayan, V. and Behera, S. N., Studies on natural radioactivity in coal ash. In Environmental Management in Coal Mining and Thermal Power Plants (eds Mishra, P. C. and Naik, A.), Technoscience, Jaipur, 1999, pp Lalit, B. Y., Ramachandran, T. V. and Mishra, U. C., Radiation exposures due to coal-fired power stations in India. Radiat. Protect. Dosim., 15, ACKNOWLEDGEMENTS. We thank the authorities at KTPS for permission to carry out field work and providing us with water and ash samples; Drs R. Srinivasan and G. K. Reddy for radiometric assaying on rock samples using gamma-ray spectrometry at NGRI and Dr V. Balaram for the elemental analysis of water samples by Inductively Coupled Plasma Spectrometry at NGRI. Received 6 July 2004; revised accepted 22 November Mishra, U., Environmental impact of coal industry and thermal power plants in India. J. Environ. Radiact., 2004, 72, Gulec, N., Gunal (Calci), B. and Erler, A., Assessment of soil and water contamination around an ash-disposal site: A case study from the Seyitomer coal-fired power plant in western Turkey. Environ. Geol., 2001, 40, Mishra, U. C. and Ramachandran, T. V., Environmental impact of coal utilization for electricity generation: In Environmental Impact of Coal Utilization from Raw Materials to Waste Resources (ed. Sahoo, K. C.), Proc. Int. Conf, IIT Bombay, 1991, pp Mandal, A. and Sengupta, D., Radioelemental study of Kolaghat Thermal Power Station, West Bengal, India Possible enviromental hazards. Environ. Geol., 44, Theis, T. L., Westrick, J. D., Hsu, C. L. and Marley, J. J., Field investigation of trace metals in groundwater from fly ash disposal. J. Water Pollut. Control Fed., 1978, 50, Theis, T. L. and Gardner, H. K., Environmental assessment of ash disposal. Crit. Rev. Environ. Control, 1990, 20, Carlson, C. L. and Adriano, D. C., Environmental impacts of coal combustion residues. J. Environ. Qual., 1993, 22, Evans, R. D., The Atomic Nucleus, McGraw Hill, New York, 1969, p Macdonald, W. G., Rozendaal, A. and de Meijer, R. J., Radiometric characteristics of heavy mineral deposits along the west coast of South Africa. Miner. Deposita, 1997, 32, Rao, R. U., Gamma-ray spectrometric set-up at NGRI for analysis of U, Th, and K in rocks. Geophys. Res. Bull., 1974, 12, Mohanty, A. K., Sengupta, D., Das, S. K., Saha, S. K. and Van, K. V., Natural radioactivity and radiation exposure in the high background area at Chhatrapur beach placer deposit of Orissa, India. J. Environ. Radioact., 2004, 75, Zoller, W. H., Gladney, E. S., Gordon, G. E. and Bors, J. J., Trace Substances in Environmental Health (ed. Hemphill, D. D.), University of Missouri, Columbia, 1974, USA, vol. VII. 13. Smith, R. D., The trace element chemistry of coal during combustion and emissions from coal-fired stations. Prog. Energy Combust. Sci., 1980, 6, Kumar, V., Ramachandran, T. V. and Prasad, R., Natural radioactivity of Indian building materials and by-products. Appl. Radiat. Isot., 1999, 51, Identification of elicitor-induced PR5 gene homologue in Piper colubrinum Link by suppression subtractive hybridization J. Dicto and S. Manjula* Plant Molecular Biology Lab, Rajiv Gandhi Centre for Biotechnology, Poojappura, Thiruvananthapuram , India Piper colubrinum Link., the exotic wild Piper shows high degree of resistance to fungal pathogens and is a potential source of resistance genes. A PCR-based suppression subtractive hybridization (SSH) was used to identify P. colubrinum genes that are differentially expressed in response to the signalling molecule, salicylic acid (SA). A subtracted library of SA-induced genes was synthesized and one of the clones showed sequence homology to osmotin, a member of class-v group of pathogenesis-related (PR) gene family. The 315 bp gene fragment was used to probe total RNA prepared from untreated and SA-treated leaf tissues. Osmotin fragment cloned from the subtracted library was also used to probe RNA from ethylene-treated leaf tissues. Northern blot analysis revealed that osmotin is dominantly expressed in SA/ethylene-treated tissue. This indicates that SSH can be used to identify and clone PR genes in P. colubrinum. PIPER colubrinum Link. is an exotic wild species of Piper, which shows high degree of resistance to many devastating *For correspondence. ( [email protected]) CURRENT SCIENCE, VOL. 88, NO. 4, 25 FEBRUARY 2005

115 Using Dynamic Reaction Cell ICP-MS Technology to Determine the Full Suite of Elements in Rainwater Samples This paper describes how scientists at an environmental and geochemical laboratory utilize dynamic reaction cell ICP-MS technology to study the transport and deposition of the full suite of important trace metals from rainwater. Previously, the presence of spectral interferences necessitated the use of additional techniques to achieve the desired detection limits, but this study shows that the lab can now use a single technique, which has increased sample throughput and dramatically improved their lab productivity. Hakan Gürleyük, Robert C. Brunette, Crystal R. Howard, Charles Schneider, and Robert Thomas It is well-documented that natural and anthropogenic sources can significantly increase trace metal concentrations in the atmosphere. This is compounded by acidic precipitation which potentially can make some trace metals more bioavailable than others. At its inception in 1977, the National Atmospheric Deposition Program (NADP) (1) recognized the need for measuring trace metals in acid rain. However, due to cost issues and the difficulty of measuring such low levels, the only toxic metal that was monitored on a regular basis was mercury as a part of the Mercury Deposition Network (MDN) (2). Recent interest has been fueled by the potential environmental effect of trace metals on aquatic and terrestrial ecosystems and potential human health affects through the consumption of drinking water. The publication of the US EPA s critical maximum concentration water quality criteria, designed to protect environmental and human health, has further demonstrated the need for the accurate measurements of trace metals in wet deposition samples (3). As a result in 1998, Frontier GeoSciences (Seattle, WA), which serves as the Hg Analytical Lab (HAL) for the MDN program, in conjunction with the NADP coordinator for air toxics, began to develop and refine the tools needed to support a trace metal Auxilliary chimney for optional For Client Review Only. All Rights Reserved. metals wet-deposition Advanstar collection Communications Inc Primary chimney for mercury wet-deposition collection funnel Warm air path Glass capillary thistle tube Mercury sample bottle Filtered air inlet vent Bottle jack Retractable lid and sealing pad Lid motor box Dry side lid rest Heater Rain sensor Cooling fan Thermostats Insulated enclosure Figure 1. A modified version of the Mercury Deposition Network station for sampling wet deposition samples for trace metal analysis. ac power 24 Spectroscopy 20(1) January

116 Table I. Typical ranges in ng/l (ppt) for trace metals in wet deposition samples. Analyte Range As Cd Cr Cu Pb Ni Se Zn network. This involved taking separate samples from almost 20 of the 90 current sampling sites across the US, using a modified version of MDN sampling station. The focus of the HAL initiative was to identify and develop both sampling and analytical methodology to accurately measure a group of critical elements including Ag, As, Be, Cd, Co, Cr, Cu, Fe, Mg, Mn, Ni, Pb, Se, V and Zn in addition to total and methyl Hg. The first step was to design a modified version of the MDN sampling station to take wet deposition samples for trace metal analysis. This modified sampling station is shown in Figure 1. Initial Analytical Issues One of the major reasons why Frontier GeoSciences began to develop a trace metal wet deposition program in 1998 was that the company had recently invested in inductively coupled plasma mass spectrometry (ICP-MS) technology. This technique s extremely low detection capability and high sample throughput enabled most of the important elements to be determined in one sample run. With many of the trace metal contaminants below 50 ng/l (ppt), method detection limits for flame atomic absorption (AA) and ICP atomic emission spectrometry (AES) wouldn t have been low enough, while electrothermal atomization would have struggled to keep up with the sample workload generated from the multitude of sampling sites. Table I shows typical concentration ranges for eight of the trace metals found in MDN wet deposition samples. Even though many of the analytes were at low ppt levels, the detection capability of ICP-MS enabled a majority of the elements to be determined with relative ease. However, normal levels of two of the most critical elements, arsenic and selenium, were lower than the detection limits for ICP-MS due to polyatomic spectral interferences derived from the plasma gas and the sample matrix. The major isotope of selenium at mass 78 and 80 were overlapped by a massive argon dimer peaks ( 40 Ar 40 Ar +, 40 Ar 38 Ar + ) generated by the plasma gas, while arsenic, which is mono isotopic ( 75 As + ) suffered a major 40Ar 35 Cl + interference from the chloride matrix at mass 75. To get around this problem, hydride For Client Review Only. All Rights generation Reserved. atomic fluorescence Advanstar spec- Communications Inc trometry (HG-AFS) was used to determine both selenium and arsenic. Similar in principle to hydride generation AA, this extremely sensitive technique involves the reduction of the metal to its Table II. Method detection limits (MDL) and reporting limits (RL) generated by Frontier GeoSciences in both 2% HNO 3 and a composite (n = 15) rainwater sample (RW), in a HNO 3 /HF matrix. Frontier GeoSciences Industry Standard (IS) DCR-ICP MS RL Improvement over IS Element DRC-ICP-MS in DRC-ICP-MS in RW/ DRC-ICP-MS RL in Conventional 2% HNO 3 (ng/l) HNO 3 /HF(ng/L) RW/HNO 3 /HF(ng/L) ICP-MS RL (ng/l) Cd Cr Cu Fe Mn Ni Pb Se V Zn As Circle 23

117 ICP-MS (a) Concentration ( g/l) Pulse intensity Figure ppt calibration plots for ( 75 As + ) (a) and ( 78 Se + ) (b) using DRC technology. gaseous hydride, but instead of detection by absorption it uses atomic fluorescence (4). Although this technique solved the detection capability issue, it significantly affected lab productivity, because three separate sample preparation methods and three separate analytical runs had to be carried out: one for For Client Review Only. All ICP-MS Rights to determine Reserved. all Advanstar the analytescommunications Inc the ICP-MS multielement suite and another two for the determination of selenium and arsenic by HG-AFS. (b) Pulse intensity Concentration ( g/l) In order to develop a major trace metals deposition network, especially with the projected sample workload, a critical question had to be answered: How could lab productivity be improved without negatively affecting detection capability and analytical accuracy? It became clear that the only realistic solution was to use including selenium and arsenic. So in the Spring of 2001 it was decided to carry out an evaluation of collision/reaction cell technology. It was well-documented in the literature that this exciting new development could reduce argonbased interferences like 40 Ar 16 O +, 40 Ar + and 38 ArH + in the determination of elements like iron ( 56 Fe + ), calcium ( 40 Ca + ) and potassium ( 39 K + ) (5). But it was unclear whether it was as efficient in reducing the argon-dimer and argonchloride based interferences for the determination of Se and As. Instrument Investigation After a careful evaluation, the instrument chosen for the investigation was an ELAN DRC (PerkinElmer SCIEX, Concord, Ontario), which utilized the principles of dynamic reaction cell technology. There are many references in the public domain describing this approach, which uses a highly reactive gas such as ammonia (NH 3 ) to react with the interference but not the analyte (6, 7, 8). Through a number of different ion-molecule reaction mechanisms, the gaseous molecules react with the interfering ions Circle 24

118 (a) (b) Conventional ICP-MS (ng/l) y = x 400 R 2 = ICP-DRC-MS (ng/l) 300 y = x 250 R 2 = HG-AFS (ng/l) HG-AFS (ng/l) Figure 3. Determination of arsenic in rainwater samples by HG-AFS and compared with both conventional ICP-MS (a) and DRC-ICP-MS (b). to convert them into species that will not interfere with the analyte. The analyte mass then emerges from the dynamic reaction cell free of its interference and enters the analyzer quadrupole for conventional mass analysis. By careful optimization of the DRC quadrupole electrical fields, unwanted reaction by-products, which potentially could lead to new interferences, are eliminated. This cleansing process, known as chemical resolution, means that every time an analyte and interfering ions enter the dynamic reaction cell, the bandpass of the quadrupole can be optimized for that specific problem and then changed on-the-fly for the next analyte The added benefit of this approach for multielement analysis is that the instrument can be operated in both DRC and normal modes. By incorporating stabilization times in the method, different gases (and gas flows) can be used for the DRC elements, while the normal mode can be used for the other elements in the same multielement run. Method Development and Validation After reviewing the literature and carrying out optimization studies, the decision was made to use oxygen gas to chemically-resolve argon chloride and argon dimers from the arsenic and selenium ions. In the case of arsenic, advantage was taken of the DRC s ability to measure the arsenic oxide ionic species ( For Client Review Only. All 75 As Rights 16 O + ) at mass 91, where the argon- Reserved. Advanstar Communications Inc chloride interference at mass 75 does not pose a spectral problem. This has been well-documented and has proven to be a clever way of determining arsenic in a chloride matrix (9). In the case of selenium, the oxygen gas is used to react with the argon dimers to produce non-interfering species so selenium can be determined at its major isotope of 80 Se + or Table III. 100 ng/l spike recovery and precision data on a composite rainwater sample. Analyte Unspiked Conc. in RW (ng/l)determined Conc. (ng/l)spike (ng/l) % Spike Recovery %RSD (n=9) As Cd Co Cr Cu Fe Mn Ni Pb Se V Zn Circle 25

119 ICP-MS 78Se +. It s also worth emphasizing that in addition to oxygen, ammonia also was used as a reaction gas to minimize the formation of other polyatomic spectral interferences such as 40 Ar 16 O +, 40Ar 12 C + and 37 C 18 O + for the determination of elements like 56 Fe + and 52 Cr + and 55 Mn + respectively. Once the gas flows and ion-molecule chemistry of the DRC were optimized Circle 26 THEFOCUSOFTHEHG ANALYTICAL LAB INITIATIVE WAS TO IDENTIFY AND DEVELOP BOTH SAMPLING AND ANALYTICAL METHODOLOGY TO ACCURATELY MEASURE A GROUP OF CRITICAL ELEMENTS. for the full suite of elements, it was decided to validate the performance of the analytical method by comparing HG-AFS data with both standard ICP- MS and DRC-ICP-MS data for arsenic and selenium in rainwater samples. For the other suite of trace metals, results from conventional ICP-MS were compared with DRC-ICP-MS results. Unfortunately no standard reference materials were available for these types of samples, so method detection limits and spike recovery studies were carried out on various archived rainwater samples. For Client Review Only. All Rights Reserved. Advanstar Communications Inc Analysis and Results The calibration graphs for arsenic ( 75 As + ) and selenium ( 78 As + ) using DRC technology are shown in Figure 2. Good linearity has been achieved for both arsenic (Figure 2A) and selenium (Figure 2B) between 0 and 500 ppt. The data also shows that both calibration plots pass through zero, which indicates that the DRC technology success- Circle 49

120 (a) (b) Conventional ICP-MS (ng/l) y = x R 2 = ICP-DRC-MS (ng/l) Conventional ICP-MS (ng/l) y = x R 2 = ICP-DRC-MS (ng/l) Figure 4. Comparison between conventional ICP-MS and DRC-ICP-MS for the determination of cadmium (a) and lead (b) in a group of rainwater samples. fully has removed or avoided the interfering species. It should be noted that the method detection limits obtained from these calibration plots are in the order of 10 ppt for both elements, which is about 3 times lower than achievable by HG-AFS. To further validate the method, a number of rainwater samples were analyzed by HG-AFS and compared with both conventional ICP-MS and DRC- ICP-MS. The results for arsenic are shown in Figure 3. It can be seen in For Client Review Only. All limits Rights (RL) in Reserved. ng/l for a Advanstar group of thecommunications Inc Figure 3A that the HG-AFS/conventional ICP MS comparison shows poor correlation (y = x), primarily because of the impact of the 40 Ar 35 Cl polyatomic interference on the analysis. Whereas in Figure 3B, which shows the HG- AFS/DRC-ICP-MS comparison, the correlation of the data is much better because the impact of the interference has essentially been removed (y=1.0755x). Although it is not shown in this paper, the HG-AFS/DRC data for selenium exhibits even better correlation than arsenic (slope = for Se compared to for As). To confirm that the other elements in the multielement suite also exhibited good correlation, a comparison was made between conventional ICP-MS and DRC-ICP-MS for the same set of rainwater samples. These elements were determined in the same multielement suite as arsenic and selenium, but using different dynamic reaction cell conditions. Data for two of the elements, cadmium and lead, are shown in Figure 4. It can be seen that there is extremely good correlation between ICP-MS and DRC- ICP-MS for both elements. The lead results are especially encouraging because they were generated by two different analysts, running the same set of samples on different days, using two different instruments. Table II shows some typical method detection limits (MDL) and reporting most important trace metals in a composite rainwater sample digested in HNO 3 /HF. It s worth emphasizing that MDLs are calculated in a similar manner to instrument detection limit (IDL) except that the blank solution is taken through the entire sample preparation procedure before the analyte concentration is measured multiple times. For this particular methodology, sample preparation closely followed the procedure set out in EPA Method 1638 (10). The MDL values in Table II represent DRC method detection limits generated by Frontier s laboratory in both 2% HNO 3 and composite rainwater (RW) sample, while the RL values represent real-world reporting limits. On average the RLs are 2 5 times worse than the MDL values and are based primarily upon what an experienced analyst is comfortable with reporting. Of particular interest is the comparison between RL values from this study and typical Circle 28

121 ICP-MS industry standard (IS) RL values (average taken from a population of conventional ICP-MS instruments used by environmental labs). It can be seen that these industry standard values would not be low enough to measure many of the analyte concentrations shown in Table I. The DRC RLs reported in this study are times lower than the industry standard values and in most cases are all within the concentration range requirements. It also ON AVERAGE THE REPORTING LIMITS IN THE STUDY ARE 2 5 TIMES WORSE THAN THE MDL VALUES AND ARE BASED PRIMARILY UPON WHAT AN EXPERIENCED ANALYST IS COMFORTABLE WITH REPORTING. should be noted that the analyte levels in the composite rainwater sample were much higher than in 2% HNO 3.As a result, it can be seen that the MDLs in 2% HNO 3, generally are much lower For Client Review Only. All 7. J. Rights M. Collard, Reserved. K. Kawabata, Advanstar Y. Kishi, R. Communications Inc than in the RW sample. Table III shows 100 ng/l spike recovery and precision data on the composite rainwater (RW) sample. It can be seen that just about all values are within the recovery limits of % as defined in EPA Method (11). The % RSD values are based upon the analysis of nine separate samples of the original composite sample. Summary The study has shown that all the environmentally significant elements, including arsenic and selenium, can be determined successfully in one analytical method using dynamic reaction cell ICP-MS. Because each rainwater sample is not being split into three to determine arsenic and selenium by HG-AFS, only one simple digestion is needed for the entire suite of elements. The findings of this investigation together with comprehensive field data have convinced our laboratory to implement this sampling and analytical methodology for the routine monitoring of trace metals in precipitation samples generated by the Mercury Deposition Network. As a result, this more efficient and economic approach to determine the full suite of elements has driven down costs approximately 2 3 fold, improved lab productivity and increased the feasibility of a long-term, nationwide trace metals monitoring network. References 1. Plan of Research for NC141 North Central Regional Project on Atmospheric Deposition. U.S. Department of Agriculture, Forest Service, North Central Forest Experimental Station, USDA, C. W. Sweet, E. M Prestbo, Measurement Of Trace Metals In The National Atmospheric Deposition Program. Illinois State Water Survey, Frontier Geosciences Inc. White Paper, April National Recommended Water Quality Criteria-Correction. EPA 822-Z , Office of Water, U.S. Environmental Protection Agency, Washington, DC., L. Rahman, W.T. Corns, D.W. Bryce and P.B Stockwell, Talanta, 52, , (2000). 5. P. Turner, T. Merren, J. Speakman and C Haines, Plasma Source Mass Spectrometry: Developments and Applications ISBN , (1996). 6. S. D. Tanner, V. I. Baranov, Atomic Spectroscopy, 20, 2, 45 52, (1999). Thomas, Micro, January, K. Kawabata Y. Kishi, and R. Thomas, Analytical Chemistry, 75(9), 423A, (2003). 9. D. S. Bollinger and A J. Schleisman, Atomic Spectroscopy, 20, 2, 60-63, (1999). 10. US EPA Method 1638: Determination of trace elements in ambient waters by ICP-MS EPA 821-R , April US EPA Method 200.8: Determination of trace elements in waters and waste waters by ICP-MS (Federal Register -Vol. 59 [232] p ), December 5, Hakan Gürleyük, Robert C. Brunette, Crystal R. Howard, Frontier GeoSciences, Seattle, WA Charles Schneider, PerkinElmer Life and Analytical Sciences, Shelton, CT Robert Thomas, Scientific Solutions, Gaithersburg, MD Circle 29

122 Experience Using Filter Paper Techniques for Whole Blood Lead Screening in a Large Pediatric Population J.A. Collins, S.E. Puskas, MEDTOX Laboratories, Inc., Saint Paul, MN Abstract Although the average blood lead levels in the U.S. population has decreased over the last 10 years, environmental lead exposure is still considered a major health risk to children. In 1997, the CDC issued guidance to state public health officials to assist in development of policies that ensure targeted testing and surveillance of those children most at risk. While current CMS policies require blood lead screening for all children enrolled in Medicaid, compliance is an issue. Alternative processes for sample collection, stability and shipping have been investigated to address these barriers to compliance. Dried blood spot sampling techniques are currently utilized for a variety of clinical tests. Use of a capillary sample collected on a filter paper provides a simplified collection procedure, sample stability and reduced costs for shipping and sample storage. Although there is still discussion regarding the applicability of the procedure to pediatric lead screening, data in the literature demonstrates good correlation between the results of filter paper lead testing and venous/capillary whole blood samples. This study describes our experience using capillary blood spotted onto filter paper as the primary specimen for whole blood lead screening in a large pediatric population. We have validated a method for filter paper lead screening by ICPMS with excellent sensitivity (1.0 µg/dl), linearity (50 µg/dl) and precision (cv s 10%). Comparisons to results derived from venous collections were excellent (r 2 = 0.990). Samples were stable for at least 6 months after collection when stored at room temperature. The validation procedure also addressed pre-analytical issues such as differing drop size, drying time, and sample application uniformity. None of these variables demonstrated significant impact on the accuracy of the analysis. Filter paper samples were also evaluated for contamination and background variability. Assay performance meets specifications in the CDC guidance documents and participation in a monthly proficiency program has demonstrated good correlation between target values and reported values (r 2 = 0.983). The specimens are collected and shipped to the laboratory via U.S. Mail. Statistics were calculated from results of samples received between June 1 and August 31, During this period, 51,307 filter paper samples were tested for whole blood lead by ICPMS. The mean blood lead level measured for the population was 3.4 µg/dl. 1872, or 3.65%, of the samples tested yielded results 10 µg/dl. These statistics are comparable to those determined for capillary tube/microtainer samples processed during the same time frame, i.e. out of 434 samples, the mean blood lead level was 3.4 µg/dl and 18, or 4.14%, of the sample results were 10 µg/dl. This is consistent with data from the NHANES III survey which estimates that 4.4% of all children have elevated levels of blood lead. Approximately, 1.1% of the samples were rejected for testing due to inadequate sample (QNS), improper specimen labeling, and improper collection media. In summary, this population data supports the notion that filter paper lead analysis in the context described provides a simple, accurate method for whole blood lead screening in children. Copyright 2003, MEDTOX Laboratories, Inc

123 Introduction Environmental lead exposure continues to present a public health issue in the United States. While overall levels in the population have declined over the last years, lead toxicity remains a significant, preventable risk, particularly for young children. Even though current CMS policies require that all young children enrolled in Medicaid are screened for lead exposure at 12 and 24 months, compliance has not been fully achieved. In addition, while this targeted population likely carries the highest risk, children in general are more impacted by exposure and cognitive deficits have been noted even at very low blood lead levels (1). Use of capillary blood spots dried on filter paper for lead screening has been reported by several investigators (2-4). The use of this methodology in pediatric screening programs is advantageous because the finger-stick collection is simple and non-invasive, relatively small sample volumes are required, and the specimen is stable and easily transported to the laboratory for analysis. Simplification of the process will likely result in higher compliance, earlier detection and better outcomes. While investigators have demonstrated excellent correlation between venous levels and results of filter paper analysis, there are still concerns expressed about pre-analytical variables that contribute to invalid results. These concerns include variability in spotting techniques and external contamination of both the filter paper and the finger surface. In our experience, while these concerns are not without merit, their impact appears to be minimal when evaluated in the context of an ongoing large scale screening program. Methodology: Collection sites are provided with the MEDTOX filter paper collection device which incorporates S&S 903 TM filter paper into a matchbook format. Providers are instructed to wash the patient s hands with soap and water and then scrub the fingertip with an alcohol prep pad and allow to air dry. The skin of the prepped finger is pierced with a lancet and the first drop of blood is wiped off with sterile gauze. Subsequent drops are collected on the sample card by holding the finger so that one drop of blood falls in each of the circles on the card. After allowing the spots to dry for 2 5 minutes, the collector is closed like a matchbook, placed in a ziplock bag and sent to the laboratory for testing. Standards and quality control samples are prepared and spotted onto filter paper in controlled lots. Standards and QC samples are stable for at least 6 months after preparation. A standard curve and QC samples are analyzed with each batch. For the analysis, two 3/16 punches of each standard, QC and patient specimen are punched directly into 15 ml disposable tubes. Blanks are punched in-between each sample to control for carryover. Tubes are vortexed after addition of bismuth internal standard and 3 ml 0.5% nitric acid, allowed to sit for 10 minutes, and vortexed again for 1 minute. Prepped samples are then placed on the autosampler and analyzed by ICP-MS (Perkin-Elmer Elan 6000, 6100 DRC or 6100 DRC Plus). Analytical conditions are optimized for each instrument model. Total run time is <2.0 minutes. Any samples with elevated results ( 10 µg/dl) are repeated on a subsequent batch prior to reporting the result. This re-verification includes analysis of a filter paper blank prepared from the patient s collection card and resampling of the collected capillary blood. Filter paper blanks that exceed 1 µg/dl lead concentration are not acceptable and a recollection is requested. Results that meet required reproducibility criteria are reported in accordance with CDC guidelines. Copyright 2003, MEDTOX Laboratories, Inc

124 The method validation summary demonstrates the following performance characteristics: Linear Range: 1.0 to 50.0 µg/dl Limit of Quantitation: 1.0 µg/dl Precision: CV s 10% Accuracy: 10 µg/dl 40 µg/dl Predicate Method Correlation: r 2 = Results are presented in Figures 1-3 and Table 1 Filter Paper Lead Linearity Filter Paper Lead Proficiency Correlation y = x R 2 = Measured Value (ug/dl) Reported Values (ug/dl) Target Concentration (ug/dl) Figure 1: Results of Method Linearity Determination. N = 6 points at each of the following concentrations: 1, 5, 10, 35 and 50 µg/dl. Filter Paper Lead Concentration (ug/dl) Patient Correlation: Venous vs Filter Paper y = 1.041x R 2 = ICP-MS Venous Concentration (ug/dl) Elan 6000 Elan DRC 6100 Regression Line Figure 2: Analytical Method Correlation: Venous samples analyzed by the predicate method were spotted on filter paper and analyzed on two different ICP-MS instruments Correlation: Target Values (ug/dl) Figure 3: Method Accuracy: Reported results plotted against target values for proficiency samples received in 2002 from the WSLH Filter Paper Proficiency Program Table 1: Inter-Run Precision (CV) and Accuracy: Control Target Concentrations (µg/dl) CT 10 CT 40 Results, Day Results, Day Results, Day Mean S.D. 1 4 CV 10.0% 9.3% Accuracy 100.0% 107.5% Correlation to the current ICP-MS method based on venous data was determined by spotting venous samples onto filter paper, allowing to dry, and processing those samples for testing by the filter paper method. Method correlation was excellent, with a correlation coefficient of Copyright 2003, MEDTOX Laboratories, Inc

125 The analytical method has also demonstrated excellent accuracy when proficiency results are compared to target. Data in Figure 3 represent data generated by participation in the Wisconsin State Laboratory of Hygiene (WSLH) Filter Paper Lead Proficiency Program in 2002 Evaluation of Pre-Analytical Variables: Lead present in the environment can contaminate the skin surface and/or the collection vessel and may contribute to the measured lead concentrations. This is true of all collection techniques, however it is commonly believed that venipuncture most limits the opportunity for contamination. Capillary collections into either a capillary tube or onto filter paper appear more vulnerable. With regard to filter paper, one must consider both skin surface contamination and external contamination of the paper collection device. Other pre-analytical variables include inconsistent spotting density, drying time, extraction recovery and insufficient specimen collection. Evaluation of blank filter paper for contamination: The MEDTOX pediatric lead sample cards containing S & S 903 TM filter paper were evaluated for background lead contamination. Ten filter paper sample cards were punched from five different areas, A-E on each card, and analyzed for lead content (Figures 4 and 5). Only one area out of 50 total had a lead level greater than the acceptable limit of 1 µg/dl. Blank Filter Paper No. of Values Lead (mcg/dl) # of values Figure 4. Sample Location on Filter Paper Sample Card Figure 5. Distribution of results of analysis of blank filter paper sample cards Evaluation of contamination on sample cards received from the field: In addition to the data presented above, contamination of filter paper collectors in the field was evaluated by review of testing statistics between 01/01/2003 and 02/28/2003. During that time interval, 26,029 capillary filter paper specimens were received, 25,876 were tested and 633 had initial results 10 µg/dl. In accordance with laboratory standard operating procedure (SOP), elevated samples are re-analyzed with sample card blanks to verify the result and measure potential contamination prior to reporting. Of the 633 elevated samples, 12 or 1.9% had blanks that exceeded the acceptance criteria and were reported as Invalid due to potential background contamination. This value reproduces the data generated from random sampling of unused cards reported above and indicates that transportation, storage and card handling do not appear to contribute to significant contamination. Spotting Homogeneity, Specimen Insufficiency: Homogeneity and relative volumes of collected spots may also impact overall results. To minimize the impact of collection variation, each sample is examined upon receipt against a set of acceptance criteria to ensure that only appropriate specimens are tested. Specimens are not tested if collected on filter paper other than Schleicher and Schuell #903 (S&S 903), blood spots do not soak through to the back side of filter paper or the area of acceptable spotting is less than that needed for two punches (insufficient specimen volume). Other spotting irregularities are visually identified by use of a chart. Copyright 2003, MEDTOX Laboratories, Inc

126 Rejection statistics from two separate time periods were reviewed to evaluate routine specimen collection procedures in the field. A primary variable between the two data sets was the introduction of the matchbook format collector before the second data interval; the original collector was an unlabeled circle of filter paper. The new collector includes features such as printed collection instructions on the inside plastic cover, printed circles on the filter paper to indicate approximate specimen volume and areas for specimen identification. Data is presented below: Rejected Samples 07/01/ /31/2002 Total # received: 32,135 QNS (inadequate sample) 205 Unsuitable (wrong filter paper) 7 Invalid (no sample ID) 136 Total number of samples rejected prior to analysis 348 (1.1%) Rejected Samples 01/01/ /28/2003 Total # received: 26,029 QNS (inadequate sample) 109 Unsuitable (wrong filter paper) 1 Invalid (no sample ID) 29 Total number of samples rejected prior to analysis: 153 (0.6%) This data demonstrates the effectiveness of providing a more controlled, directed collection procedure in minimizing the number of specimens failing to meet acceptance criteria. A specimen rejection rate of <1% indicates that nonhomogeneous sample collection is not a significant issue. Drying Time: The effect of drying time on filter paper lead results was evaluated by use of three controls spotted onto filter paper and dried for varying time periods prior to analysis. Based on this data, drying time for these collections is minimal and no variance was observed over the time intervals evaluated. Results are summarized in Table 2. Table 2. Effect of Drying Time on Results Filter Paper Dry Time Study *T values are QC target concentrations Results (µg/dl) Results (µg/dl) Results (µg/dl) Time in minutes T = 6* T = 25* T = 42* Population Statistical Data: Initial results represent a population of 51,307 filter paper samples received and tested between June 1, 2002 and August 31, All patients are <18 years of age. Data from capillary tube samples received during the same time frame are presented for comparison. Capillary Microtainer vs. Filter-Paper: 06/01/ /31/2002 Sample Type Capillary Tube Filter Paper Number of Samples ,307 Mean Lead Level (µg/dl) Standard Deviation Number 10 µg/dl 18 (4.14%) 1872 (3.65%) Copyright 2003, MEDTOX Laboratories, Inc

127 Figures 6 and 7 are histograms representing the distribution of lead levels measured by the filter paper technique (Fig 6) as compared to capillary microtainer tube collections (Fig. 7). As indicated by the above data, the average values and % of elevated samples are comparable. LeadFP SAMPLES 6/1/2002-8/31/2002 Lead WB under 18 years old - Capillary SAMPLES 6/1/2002-8/31/2002 Frequency Figure 6. Distribution of Filter Paper Blood Lead Levels in a Pediatric Population collected between 06/01/2002 and 08/31/2002. Figure 7. Distribution of Capillary Microtainer Blood Lead Levels in a Pediatric Population collected between 06/01/2002 and 08/31/2002 A second set of data was evaluated for comparison. This additional data represents tests performed between 12/01/2002 and 03/10/2003. Results are presented below: Capillary Microtainer vs. Filter-Paper: 12/01/ /10/2003 Sample Type Capillary Tube Filter Paper Number of Samples ,775 Mean Lead Level (µg/dl) Standard Deviation Number 10 µg/dl 12 (2.2%) 1081 (2.4%) Figures 8 and 9 are histograms representing the distribution of lead levels measured by the filter paper technique (Fig 8) as compared to capillary microtainer tube collections (Fig. 9). As indicated by the above data, the average values and % of elevated samples are comparable. LeadFP SAMPLES 12/1/2002-3/10/2003 Lead WB under 18 years old - Capillary SAMPLES 12/1/2002-3/10/ Frequency AVERAGE MEDIAN 2 STDEV Frequency More More AVERAGE MEDIAN 3 STDEV Frequency Frequency AVERAGE MEDIAN 2 STDEV Frequency More More ug/dl ug/dl Frequency AVERAGE MEDIAN 2 STDEV Frequency ug/dl ug/dl Figure 8. Distribution of Lead Levels Measured from Filter Paper Samples Collected Between 12/01/ /10/2003 Figure 9. Distribution of Lead Levels Measured from Capillary Microtainer Tubes Collected Between 12/01/ /10/2003 Copyright 2003, MEDTOX Laboratories, Inc

128 Discussion and Summary: According to estimates from the most recent National Health and Nutrition Examination Survey (NHANES III, Phase 2, ) (5), 2.2% of the US population between ages 1 74 years has blood lead levels 10 µg/dl. Among those aged 1-5 years, approximately 4.4% had levels 10 µg/dl. That represents approximately 930,000 US children. While the survey data demonstrated a disproportional risk for certain ethnic groups including children of lower income families living in older housing in large urban areas, risk associated with the age of housing category persisted across race/ethnicity, income and urban status (5). This supports the notion that screening for lead exposure is important for all children, regardless of ethnicity and economic status. While current federal law requires states to screen children enrolled in Medicaid for elevated blood lead levels, estimates from a 1999 GAO report indicated that more than 75% of the children between ages 1 5 had not been screened (6). This emphasizes the need for programs/methodology that will increase compliance with screening requirements. One approach is to facilitate the collection and screening process. Several authors have proposed use of capillary blood collected on filter paper as a simplified, reliable procedure to collect and analyze lead levels in this population (2-4). The data we have presented further supports the validity of this methodology. In addition, our data demonstrates that pre-analytical issues can be controlled to a large extent by provision and use of a well-designed sample collection device. Method performance: The validation and method performance data for analysis of whole blood lead collected on filter paper by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) indicates that the method provides the sensitivity, linearity, precision and accuracy required to effectively monitor lead levels in the population. Successful participation in external proficiency testing programs is further evidence of method robustness and performance. Pre-analytical variables: The data presented addresses concerns that pre-analytical variables such as spotting consistency, drying time and contamination of the skin surface and/or the collector may lead to an increased level of false positives. The inclusion of standard specimen acceptance criteria in the laboratory SOP minimizes the potential that inappropriate samples are tested. In addition, drying time did not appear to impact the results. These variables were clearly improved by the modification to the sample collection device. Evaluation of the filter paper itself for random environmental contamination demonstrated only minimal occurrence (2%). This frequency was confirmed by data from the field. When considered in the context of a large screening program where filter paper positives are reanalyzed prior to reporting and subsequently followed up with venipuncture to confirm the elevated result, the value far outweighs the concern. In addition, a recent report from the Minnesota Department of Health indicates that more than half of the 66,000 tests reported to the department in 2002 were capillary (7). One can argue that filter paper testing provides a mechanism to evaluate environmental contamination by use of paper blanks and separate sampling areas. Capillary microtainer samples provide no means of identifying contamination. Population Statistics: The data from each of the time intervals monitored demonstrate excellent correlation between filter paper and capillary microtainer samples received and tested. The frequency of elevated blood lead levels (BLLs) is also consistent within the interval. These results are similar to the data reported from Phase 2 of NHANES III in which age-group mean blood lead levels (1 5 and 6-11 yrs) were 1.9 and 2.7 µg/dl, respectively. The percent of elevated BLLs in the same age groups were 4.4 and 2.2% (5). In summary, the methodology has proven to be robust and demonstrates the requisite sensitivity and precision to measure blood lead at low concentrations. In addition, the data presented for filter paper lead testing on samples collected from children <18 years of age show excellent correlation to published results of a large population based on venous sampling. The results support the notion that filter paper methodology can simplify sample collection and transportation procedures, thereby facilitating screening of children for lead exposure. Copyright 2003, MEDTOX Laboratories, Inc

129 References: 1. Canfield, R.L., et.al., New Eng. J. Med. 348:16, , Cernik, A.A. and M.H.P Sayers. Brit. J. Industr. Med. 23: , Vereby, K., et.al., J. Anal. Toxicol. 15, , Yee, H. Y. and T.G. Holtrop, J. Anal. Toxicol. 21, , CDC. Morbidity and Mortality Weekly Report 46:7, , US General Accounting Office, Lead Poisoning: Federal Health Care Programs are not Effectively Reaching At-Risk Children, January Minnesota Department of Health Lead Program, Minnesota Blood Lead Surveillance Data, 2002 Copyright 2003, MEDTOX Laboratories, Inc

130 THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 279, No. 18, Issue of April 30, pp , by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. As(III) and Sb(III) Uptake by GlpF and Efflux by ArsB in Escherichia coli* Received for publication, January 5, 2004, and in revised form, February 13, 2004 Published, JBC Papers in Press, February 16, 2004, DOI /jbc.M Yu-Ling Meng, Zijuan Liu, and Barry P. Rosen From the Department of Biochemistry and Molecular Biology, Wayne State University, School of Medicine, Detroit, Michigan The toxicity of the metalloids arsenic and antimony is related to uptake, whereas detoxification requires efflux. In this report we show that uptake of the trivalent inorganic forms of arsenic and antimony into cells of Escherichia coli is facilitated by the aquaglyceroporin channel GlpF and that transport of Sb(III) is catalyzed by the ArsB carrier protein; everted membrane vesicles accumulated Sb(III) with energy supplied by NADH oxidation, reflecting efflux from intact cells. Dissipation of either the membrane potential or the ph gradient did not prevent Sb(III) uptake, whereas dissipation of both completely uncoupled the carrier protein, suggesting that transport is coupled to either the electrical or the chemical component of the electrochemical proton gradient. Reciprocally, Sb(III) transport via ArsB dissipated both the ph gradient and the membrane potential. These results strongly indicate that ArsB is an antiporter that catalyzes metalloid-proton exchange. Unexpectedly, As(III) inhibited ArsB-mediated Sb(III) uptake, whereas Sb(III) stimulated ArsB-mediated As(III) transport. We propose that the actual substrate of ArsB is a polymer of (AsO) n, (SbO) n, or a co-polymer of the two metalloids. Arsenic, one of the most prevalent toxic metals in the environment, derives primarily from geochemical origins but also from man-made sources. Consequently, nearly every organism has intrinsic or acquired mechanisms for arsenic detoxification (1). The arsenical resistance operon (arsrdabc) of the conjugative R-factor R773 confers resistance to inorganic As(III) and Sb(III) in Escherichia coli. The arsenic transport system exhibits a dual mode of energy coupling depending on the subunit composition (2). When both ArsA and ArsB are present, they form the As(III)-translocating ArsAB ATPase, which is independent of the electrochemical proton gradient (3). In contrast, in the absence of ArsA, ArsB catalyzes As(III) extrusion coupled to electrochemical energy, which suggests that ArsB is a uniporter that extrudes the arsenite anion in response to the positive exterior membrane potential (4). This dual mode of energy coupling led us to propose that the ArsAB pump evolved by association of a secondary carrier with a soluble ATPase (5). * This research was supported by National Institute of General Medical Sciences Grant GM52216, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. To whom correspondence should be addressed: Dept. of Biochemistry and Molecular Biology, Wayne State University, School of Medicine, 540 E. Canfield Ave., Detroit, MI Tel.: ; Fax: ; [email protected] Over a decade ago, we proposed that other primary ATP-coupled pumps such as ATP-binding cassette transporters evolved in similar ways (5, 6). ArsB is the most widespread determinant of arsenic resistance in bacteria and archaea, yet its transport properties are not well characterized. It is a member of the ion transporter superfamily (7), with 12 membrane-spanning segments and a membrane topology that is similar to many carrier proteins (8). To date, it has been shown to transport only As(III) (4). Here we report for the first time that ArsB transports inorganic Sb(III) in E. coli, and we describe the relationship with As(III) transport. Considering that the pk a of Sb(III) is 11.8 and As(III) is 9.2, at cytosolic ph the concentration of the oxyanion of either metalloid is negligible. Thus, it is unlikely that ArsB could be an electrophoretic anion uniporter. Instead, we demonstrate here that ArsB is a trivalent metalloid/h antiporter. ArsBcatalyzed uptake of Sb(III) into everted membrane vesicles coupled to either the ph gradient or membrane potential components of the electrochemical proton gradient. Sb(III)/H exchange was monitored using a fluorescent reporter, acridine orange, for the ph gradient and oxonal V for the membrane potential. The nature of the substrate of ArsB was explored by competition and co-transport experiments with Sb(III) and As(III). Given their pk a values, the physiologically relevant species of Sb(III) and As(III) in a solution of neutral ph are hydroxides, but the nature of the true substrate of ArsB is not clear. As(III) inhibits ArsB-mediated Sb(III) transport, yet Sb(III) stimulates ArsB-mediated As(III) transport. From these results and the results of As(III) and Sb(III) co-transport experiments, we propose that the true substrate of ArsB is a polymer of As(III) or Sb(III), or a co-polymer of As(III) and Sb(III). EXPERIMENTAL PROCEDURES Materials Restriction enzymes and nucleic acid-modifying enzymes were purchased from Invitrogen and New England Biolabs, Inc. Carrier-free 73 AsO 3 4 was obtained from Los Alamos National Laboratories. 125 SbCl 5 was produced by PerkinElmer Life Sciences. The pentavalent forms of the isotopes were reduced to 73 As(III) and 125 Sb(III) by the method of Reay and Asher (9). All other chemicals were obtained from commercial sources. Strains, Plasmids, and Media E. coli strains and plasmids used in this study are listed in Table I. E. coli strains harboring the indicated plasmids were grown in Luria-Bertani medium (10) at 37 C with 100 g/ml ampicillin, 35 g/ml chloramphenicol, or 50 g/ml kanamycin as required. Protein expression was induced by addition of 0.1 mm isopropyl-1-thio- -D-galactopyranoside. Resistance Assays For assays of resistance to arsenite and antimonite, cultures were grown overnight at 37 C with shaking. The cells were diluted 100-fold into fresh, prewarmed medium with the indicated concentrations of As(III) in the form of sodium arsenite or Sb(III) in the form of potassium antimonyl tartrate and incubated at 37 C with shaking for an additional 6 h. Growth was estimated from the absorbance at 600 nm. This paper is available on line at

131 Metalloid Transport in E. coli TABLE I Strains and plasmids Genotype/description Ref. or source E. coli strains AW3110 ars::cam F IN(rrnD-rrnE) 14 OSBR1 AW3110 glpf::tnphoa, Km r 13 AW10 JM110 ars::cam dam dcm supe44 hsdr17 thi leu rpsl lacy galk galt ara tona thr tsx 4 (lac-proab), Cm r JM109 reca1 supe44 enda1 hsdr17 gyra96 rela1 thi (lac-proab F [trad36 proab lacl q 10 lacz M15]) Plasmids pkk223 3 Cloning and expression vector with tac promote, Ap r Amersham Biosciences parsa3 2.4-kb HindIII fragment containing arsa inserted into HindIII digested pacyc184, Cm r 23 gene replaced by Km r pkmb1 1.3-kb fragment of arsb inserted into EcoRI-HindIII digested pkk223 3, Ap r 4 Transport Assays For uptake assays in intact cells, cultures were grown to A 600 nm of1at37 C with aeration in Luria-Bertani medium. The cells were harvested, washed, and suspended in a buffer consisting of 75 mm HEPES-KOH, ph 7.5, containing 0.15 M KCl and 1 mm MgSO 4, brought to a concentration of 80 mg of wet cells/ml at room temperature. To initial the transport assay, 0.1 ml of cells was diluted in 1 ml of the same buffer at room temperature containing 10 M sodium arsenite and 0.4 Ci of 73 As(III) or 10 M potassium antimonyl tartrate and 0.4 Ci of 125 Sb(III). Samples (0.1 ml) were withdrawn at the indicated times, filtered through 0.2- m pore diameter nitrocellulose filters (Whatman), and washed with 15 ml of the same buffer, all at room temperature. The filters were dried and quantified by liquid scintillation counting. Transport assays using everted membrane vesicles were performed as described (4). Everted membrane vesicles were prepared essentially as described previously (4). Unless otherwise noted, the reaction mixture contained 0.3 mg of membrane protein in a final volume of 0.6 ml of a buffer consisting of 75 mm HEPES-KOH, ph 7.5, containing 0.1 M K 2 SO 4, 0.25 M sucrose, and 1.25 Ci of either 73 As(III) or 125 Sb(III). Assays were initiated by the addition of 5 mm NADH, final concentration. The concentration of As(III) was adjusted by the addition of buffered sodium arsenite, and buffered potassium antimonyl tartrate was used to adjust the total Sb(III) concentration. Transport assays using inductively coupled mass spectrometry (ICP-MS) 1 were performed by the same procedure with a PerkinElmer ELAN 9000 except that nonradioactive metalloids were used, and the total volume was increased to 1.8 ml. At the indicated times, 0.3-ml samples were withdrawn and filtered through 0.2- m pore size nitrocellulose filters (Whatman). After filtration, the filters were washed with 10 ml of the same buffer and air-dried. For radioactive assays, the radioactivity was quantified by liquid scintillation counting. For ICP-MS measurements, the filters were digested with 0.3 ml of concentrated HNO 3 (69 70%) (EM Science) overnight at room temperature. The dissolved filters were incubated for 10 min at 70 C, allowed to cool to room temperature, and diluted with 5.25 ml of high pressure liquid chromatography grade water (Sigma) to produce a final concentration of HNO 3 of approximately 4%. Standard solutions were made in the range of ppb in 4% HNO 3 using arsenic and antimony standards (Ultra Scientific). Initial rates were determined from a linear regression of time points at 0.5, 1, 2, 3, and 4 min. Kinetic data were analyzed by nonlinear regression using SigmaPlot version6.1. Protein content was determined with a BCA protein assay kit (Pierce), using bovine serum albumin as a standard. Measurement of ph Gradient Formation by Acridine Orange Fluorescence Formation of ph was estimated from the quenching of acridine orange fluorescence (11). The reaction mixture consisted of 20 mm HEPES-KOH, ph7.2, containing 0.1 M KCl, 2.5 mm MgSO 4, 2 M acridine orange, and 0.13 mg of membrane protein in a volume of 2 ml. Quenching was initiated at room temperature by addition of 5 mm NADH (final concentration). Fluorescence was measured in stirred cuvettes with an Aminco AB2 spectrofluorometer, with excitation at 492 nm and emission at 527 nm. Measurement of Membrane Potential Formation by Oxonol V Fluorescence Formation of was estimated from the quenching of oxonol V fluorescence as described (12). The reaction mixture consisted of 20 mm HEPES-KOH, ph7.2, 2.5 mm MgSO 4,5 M oxonol 595 (Aldrich), 1 The abbreviations used are: ICP-MS, inductively coupled plasmamass spectrometry; FCCP, carbonyl cyanide p-trifluoromethoxyphenylhydrazone; EXAFS, x-ray absorption fine structure spectroscopy. and 0.13 mg of membrane protein in a volume of 2 ml. Quenching was initiated at room temperature by addition of 5 mm NADH (final concentration). After 5 6 min, potassium tartrate or potassium antimonyl tartrate was added at the indicated concentrations. Fluorescence was measured in stirred cuvettes with an Aminco AB2 spectrofluorometer with excitation at 589 nm and emission at 616 nm. RESULTS Uptake of As(III) and Sb(III) in E. coli We have previously shown that disruption of the aquaglyceroporin GlpF confers resistance to Sb(III) (13). This disruption was interpreted as a loss of the uptake pathway for Sb(III); however, transport of Sb(III) has never been directly demonstrated in E. coli. Moreover, the GlpF disruption did not gain resistance to As(III), so it was unclear whether GlpF is an uptake pathway for As(III). Uptake of either 73 As(III) or 125 Sb(III) was measured in the E. coli strains AW3110 or AW10, in which the chromosomal arsrbc operon was deleted (4, 14), and OSBR1, which was created from AW3110 by inserting TnphoA into glpf (13). Because AW3110 lacks arsb, it was unable to extrude metalloids and accumulated 125 Sb(III) (Fig. 1A) or 73 As(III) (Fig. 1B). Disruption of glpf greatly reduced the level of uptake of both metalloids. This clearly demonstrates that GlpF is the major uptake pathway for both As(III) and Sb(III). Aquaglyceroporins are channels that facilitate the movement of neutral substrates such as glycerol and other polyols but not ions (15). Considering that the pk a value of trivalent arsenic and antimony is 9.2 and 11.8, respectively, there is essentially no arsenite or antimonite anion at physiological ph levels. To be substrates of aquaglyceroporins, the metalloids would be expected to be the neutral hydroxides As(OH) 3 or Sb(OH) 3, which are the inorganic equivalents of polyols (13, 16). However, as discussed in more detail below, other possibilities exist, so, for the purposes of this study they are designated As(III) and Sb(III). ArsB Confers Sb(III) and As(III) Resistance Intracellular As(III) and Sb(III) are toxic to most cells, including E. coli (1). E. coli has a chromosomal arsb gene in the three-gene arsrbc operon that confers moderate levels of resistance to these metalloids (14). Plasmids such as R773 have five-gene arsrdabc operons that confer high levels of resistance (17). The R773 and chromosomal ArsBs share 90% identity at the amino acid level, and the R773 arsb gene complemented both Sb(III) (Fig. 2A) and As(III) (Fig. 2B) hypersensitivity resulting from deletion of the chromosomal arsb gene. Expression of arsa in trans increased resistance. These results demonstrate that ArsB can function either alone or as a complex with ArsA to confer resistance to Sb(III), and importantly, they demonstrate the interchangeability of the chromosomal and plasmid ArsB carriers. ArsB Catalyzes 125 Sb(III) Uptake in Everted Membrane Vesicles ArsB has been shown to transport As(III) (4), but Sb(III)

132 18336 Metalloid Transport in E. coli FIG As(III) and 125 Sb(III) uptake into cells of E. coli is facilitated by the aquaglyceroporin channel GlpF. Transport of 125 Sb(III) (A) and 73 As(III) (B) were assayed as described under Experimental Procedures. The strains used were AW3110 ( ars) ( ) and OSBR1 ( ars glpf) (E). FIG. 2.Contributions of arsa and arsb to resistance to Sb(III) and As(III). Cells of E. coli strain AW10 ( ars) were grown in the presence of various concentrations of potassium antimonyl tartrate (A) and sodium arsenite (B) at 37 C for 6 h in the absence of isopropyl-1- thio- -D-galactopyranoside. Plasmids contained in cells were: pkk223-3 (vector) and parsa3 (arsa) ( ); pkmb1 (arsb) (E); pkmb1 and parsa3 ( ). transport has not been demonstrated. Uptake into everted membrane vesicles prepared from cells expressing the R773 ArsB is the equivalent of efflux from cells. Everted membrane vesicles from strain AW10 ( ars::cam) expressing arsb from plasmid pkmb1 accumulated 125 Sb(III) when NADH was used as a respiratory substrate (Fig. 3). No uptake was observed without NADH or in vesicles from cells with vector only. Addition of the uncoupler carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) produced a rapid loss of accumulated Sb(III). These results demonstrate that Sb(III) uptake is coupled to the electrochemical proton gradient. Similar results were obtained if 125 Sb(III) was prepared with unlabeled potassium antimonyl tartrate or SbCl 3 (data not shown), showing that only the oxidation state of antimony is relevant. The effects of oxyanions on ArsB-catalyzed 125 Sb(III) uptake into everted membrane vesicles were examined (Fig. 4). Among the oxyanions tested, the sodium or potassium salts of AsO 4 3, Sb(OH) 6,PO 4 3,NO 3,NO 2,SO 3 2, SeO 3 2, and BO 3, or tartrate had little effect on 125 Sb(III) transport. Only As(III) (added as sodium arsenite) inhibited transport; therefore ArsB appears to be specific for Sb(III) and As(III). ArsB Catalyzes Sb(III)/H Exchange We previously proposed that ArsB is a uniporter for the arsenite anion, AsO 2, which was thought to be the form of As(III) in solution (4). This idea is consistent with the uncoupling effect of FCCP on NADH-driven As(III) uptake in everted membrane vesicles that shows that ArsB-catalyzed As(III) uptake is driven by the electrochemical proton gradient, which is acid and positive interior in these vesicles (Fig. 3). However, as discussed above, the metalloid substrates of ArsB are not anions at physiological ph levels. An alternative for coupling transport of a neutral solute into the acid and positive interior of the everted vesicles is by exchange with a cation or proton, but dependence on a specific cation such as Na,K,Mg 2,orCa 2 was not observed (data not shown). To examine the possibility that ArsB is an electrophoretic Sb(III)/H antiporter, the effects of a permeant anion (SCN ) and weak base (NH 4 ) on Sb(III) uptake into everted membrane vesicles were measured (Fig. 5). In an attempt to limit the use of exotic isotopes, the amount of antimony taken up in this and other assays, as noted, was determined by ICP-MS. When SCN and NH 4 were added together, they uncoupled Sb(III) transport from NADH oxidation as well as did FCCP. However, neither SCN, which dissipates the positive interior but does not dissipate ph, nor NH 4, which dissipates the acid interior ph but does not dissipate, by themselves had any effect on NADH-driven accumulation of Sb(III). These results indi-

133 Metalloid Transport in E. coli FIG. 3. Energy-dependent accumulation of 125 Sb(III) in everted membrane vesicles. Accumulation of 125 Sb(III) in everted membrane vesicles prepared from strain AW10, harboring either plasmid pkk223-3 or pkmb1 with 100 M 125 Sb(III), was measured as described under Experimental Procedures., pkk223-3 with 5 mm NADH; E, pkmb1 without NADH;, pkmb1 with 5 mm NADH; ƒ, pkmb1 with 5 mm NADH to which 10 M FCCP was added after 2 min (arrow). FIG. 5.Energetics of ArsB-mediated 125 Sb(III) transport. Accumulation of 125 Sb(III) was assayed in everted membrane vesicles prepared from strain AW10 harboring pkmb1 with 5 mm NADH and 100 M 125 Sb(III). At 1.5 min (arrow) the following additions were made: no addition ( ); 10 mm NH 4 Cl (ƒ); 10 mm KSCN (E); 10 mm (NH 4 ) 2 SO 4 ( ); 10 mm NH 4 Cl 10 mm KSCN ( ); 10 mm (NH 4 ) 2 SO 4 10 mm KSCN ( ); 10 M FCCP (f). FIG. 4.Specificity of 125 Sb(III) transport by ArsB 125 Sb(III) uptake was assayed in everted membrane vesicles with 5 mm NADH as an energy source and 125 Sb(III) adjusted to 10 M with potassium antimonyl tartrate. Sodium arsenite was used at 100 M and other salts at 1.0 mm (final concentration). First column, no addition; second, NaAsO 2 (AsIII); third, Na 2 HAsO 4 (AsV); fourth, KSb(OH) 3 (SbV); fifth, NaK tartrate; sixth, Na 2 SeO 3 ; seventh, Na 2 SO 3 ; eighth, KNO 3 ; ninth, KNO 2 ; tenth, K 2 HPO 4 ; eleventh, H 3 BO 3. All salts were adjusted to ph 7.4. The error bars represent the standard deviations. cate that either or ph alone is sufficient to energize Sb(III) transport. Because dissipation of only does not inhibit transport, ArsB cannot be an electrophoretic anion uniporter. Instead, these results are consistent with metalloid-proton exchange. The effect of Sb(III) on ph was examined. Solute/proton exchange can be assayed by the effect of the co-transported species on ph using a fluorescent, weak base (18, 19). The fluorescence of the weak base acridine orange is quenched on formation of a ph gradient in everted membrane vesicles (Fig. 6) (11). Addition of 10 mm (NH 4 ) 2 SO 4 completely reversed fluorescence quenching, consistent with dissipation of ph by a weak base (data not shown). Addition of Sb(III) in the form of potassium antimonyl tartrate reversed quenching in a concentration-dependent manner; potassium tartrate alone had no FIG. 6.Effect of Sb(III) on ph. The transmembrane ph gradient was estimated from the quenching of acridine orange fluorescence by everted membrane vesicles of strain AW10 pkmb1 (curves 1 8) or AW10 pkk223-3 (curve 9) as described under Experimental Procedures. Formation of ph was initiated by addition of 5 mm NADH (first arrow). Potassium tartrate or potassium antimonyl tartrate was added at the indicated times (second and third arrows). Additions: curve 1, 10 M potassium antimonyl tartrate; curve 2, 50 M potassium antimonyl tartrate; curve 3, 100 M potassium antimonyl tartrate; curve 4, 200 M potassium antimonyl tartrate; curve 5, 500 M potassium antimonyl tartrate; curve 6, 1000 M potassium antimonyl tartrate; curve 7, 2000 M potassium antimonyl tartrate; curve 8, 500 M potassium tartrate; curve 9, vector 500 M potassium antimonyl tartrate. effect. Membranes from cells lacking an arsb gene showed no fluorescence enhancement, demonstrating that this assay measures ArsB activity. The effect of Sb(III) on was examined. Membrane potential formation in everted membrane vesicles can be visualized from the quenching of the permeant dye oxonol (12). Addition of NADH quenched oxonol fluorescence (Fig. 7). Sb(III) rapidly reversed quenching in a concentration-dependent manner. There was no effect of Sb(III) on oxonol fluorescence in membranes from cells without arsb. Potassium tartrate slowly depolarized the membrane, but the same effect was observed in membranes from cells with or without arsb. Addition of 10 mm KSCN rapidly and com-

134 18338 Metalloid Transport in E. coli FIG. 7.Effect of Sb(III) on. The membrane potential was estimated from the quenching of oxonol fluorescence by everted membrane vesicles of strain AW10 pkmb1 (curves 1 3 and 5) or AW10 pkk223-3 (curves 4 and 6) as described under Experimental Procedures. Formation of was initiated by addition of 5 mm NADH (first arrow). Potassium tartrate or potassium antimonyl tartrate was added at the indicated times (second arrow). Additions: curve 1, ArsB 500 M potassium antimonyl tartrate; curve 2, ArsB 1000 M potassium antimonyl tartrate; curve 3, ArsB 2000 M potassium antimonyl tartrate; curve 4, vector 2000 M potassium antimonyl tartrate; curve 5, ArsB 2000 M potassium tartrate; curve 6, vector 100 M potassium antimonyl tartrate. FIG. 9.Co-transport of Sb(III) and As(III). Co-transport of As(III) ( ) and Sb(III) (f) was measured by ICP-MS as described under Experimental Procedures. Each panel shows two assays, each of which contained a mixture of sodium arsenite and potassium antimonyl tartrate. In each assay the concentration of one of two metalloids was held constant at 10 M (A), 50 M (B), 100 M (C), 500 M (D), or 1000 M (E). In each assay the concentration of the other metalloid was varied between 0 and 1000 M. Samples were withdrawn at 0.5, 1, 2, 3, and 4 min, and the amounts of both As(III) and Sb(III) in each sample were determined by ICP-MS. The rate of uptake was determined from linear regression of the resulting time course using SigmaPlot 6.1. FIG. 8. Sb(III) stimulates ArsB-catalyzed 73 As(III) transport. Uptake of 73 As(III) by everted membrane vesicles of strain AW10 pkmb1 (arsb) or AW10 pkk223-3 (Vector) was assayed with 100 M 73 As(III) as described under Experimental Procedures., no ArsB 5mM NADH; E, ArsB without NADH;, ArsB 5mM NADH; ƒ, ArsB 5mM NADH 100 M potassium antimonyl tartrate. pletely reversed fluorescence quenching, showing that this permeant anion dissipates (data not shown). ArsB Catalyzes Co-transport of As(III) and Sb(III) Everted membrane vesicles prepared from cells of E. coli strain AW10 expressing ArsB from plasmid pkmb1 exhibited NADH-coupled 73 As(III) uptake compared with cells with vector plasmid pkk223-3 (Fig. 8). Comparing the rates of uptake at a metalloid concentration of 0.1 mm, the rate of 125 Sb(III) uptake (31.3 nmol/min/mg protein) (Fig. 3) was approximately 35-fold higher than that of 73 As(III) (0.9 nmol/min/mg protein) (Fig. 8). When Sb(III) was added together with 73 As(III), the rate of uptake was stimulated 10-fold (9.1 nmol/min/mg protein) (Fig. 8). In contrast, As(III) inhibited the uptake of 125 Sb(III) (Fig. 4). To examine this apparently paradoxical result in more detail, the rates of As(III) and Sb(III) uptake were determined simultaneously using ICP-MS (Fig. 9). In this experiment, the concentration of one metalloid was fixed at five different concentrations: 10, 50, 100, 500, and 1000 M. At each of these concentrations of one metalloid, the concentration of the other metalloid was varied between 0 and 1000 M. At every concentration of As(III), Sb(III) stimulated uptake of As(III), and at every concentration of Sb(III), As(III) inhibited uptake of Sb(III). When the two metalloids were present at approximately equal concentrations, the rates of uptake of the two were approximately the same. The most parsimonious explanation for these results is that ArsB catalyzes co-transport of the two metalloids. However, co-transport does not explain the converse effect of one metalloid on the rate of uptake of the other. If it were simply co-transport, with separate sites on ArsB for As(III) and Sb(III), then As(III) would be expected to stimulate Sb(III) transport as Sb(III) stimulates As(III) transport. If As(III) and Sb(III) were simply alternate substrates for the same site on ArsB, then Sb(III) would be expected to compete with the uptake of As(III) as As(III) competes for Sb(III) uptake. Moreover, the effect of As(III) on Sb(III) uptake does not appear to be simple competitive inhibition. When the concentration dependence of Sb(III) uptake was analyzed as a function of As(III), increasing sigmoidicity was observed (Fig. 10). In the absence of As(III), the data from two separate experiments could be reasonably fitted to the Michaelis- Menten relationship, generating apparent K m and V max values of 43 M and 182 nmol/mg protein/min, respectively. When the data were analyzed using the Hill relationship, the apparent K m value was 44 M, with a V max of 172 nmol/mg protein/min and a Hill coefficient of 1.6 (Table II). As the concentration of As(III) increased, the apparent K m increased, the V max decreased, and the Hill coefficient in-

135 Metalloid Transport in E. coli FIG. 10. Effect of As(III) on the kinetics of Sb(III) transport. Sb(III) accumulation in everted membrane vesicles prepared from strain AW10 pkmb1 (arsb) was measured by ICP-MS as described under Experimental Procedures. Each assay contained 5 mm NADH and the indicated concentrations of Sb(III) in the form of potassium antimonyl tartrate. Each assay was performed in the presence of the following concentrations of sodium arsenite: none ( ); 50 M (ƒ); 100 M (f); 500 M ( ), and 1000 M (Œ). The lines represent best fit of the data to the Hill equation using SigmaPlot 6.1. creased to a value of 2.5 at 1000 M As(III). These results strongly suggest some sort of interaction of As(III) and Sb(III) associated with transport by ArsB. DISCUSSION Reflecting the pervasiveness of environmental arsenic (20), ArsB is a ubiquitous transport protein found in the genomes and plasmids of most bacteria and archaea (1). ArsB is unusual in that it is either a secondary carrier coupled to the electrochemical proton gradient or the translocation subunit of the As(III)/Sb(III)-translocating ArsAB ATPase (2, 3). Based on this novel dual mode of energy coupling, we had proposed that not only the ArsAB pump but other solute-translocating ATPases such as the F 0 F 1 and ATP-binding cassette transporters evolved from the association of carriers or channels with soluble ATPases (5, 6). Yet, both the mechanism and substrate of ArsB are probably different from those that were previously conceived (4). Based on a dependence on the electrochemical proton gradient, we had proposed that ArsB is a uniporter that catalyzes electrophoretic efflux of the arsenite anion out of cells in response to the outside positive membrane potential. In this report we demonstrate for the first time translocation of Sb(III) by ArsB. Because the pk a value of inorganic trivalent antimony is 11.8, the concentration of the antimonite anion at a cytosolic ph level of 7.5 (21) is 4 orders of magnitude lower than the total Sb(III) concentration. The paucity of the intracellular antimonite anion would make a uniport mechanism improbable; rather, As(III) and Sb(III) are protonated neutral molecules at cytosolic ph level, and mechanisms that couple efflux of a neutral substrate to the electrochemical proton gradient must be considered. Extrusion of a neutral molecule from cells into the acid and positive exterior could be accomplished by exchange with a cation. Because dependence on an inorganic cation was not observed, protons are a reasonable alternative. This hypothesis was tested in two ways. First, the electrochemical proton gradient was applied as only a membrane potential or only a ph gradient. Either was capable of supporting Sb(III) uptake into everted membrane vesicles. These results are inconsistent with either a uniporter for a neutral solute, which would catalyze only facilitated diffusion, or a uniporter for an anion, which would be able to couple only to the membrane potential and not the ph gradient. Second, Sb(III)/H exchange can be inferred by the dissipation of ph concomitant with the addition of Sb(III). That exchange is electrophoretic is shown by the ability of Sb(III) to dissipate. The most reasonable explanation for these results is that ArsB is a metalloid-proton antiporter. From the reported topological determination of ArsB (8), two glutamate and four aspartate residues can be predicted to be located in transmembrane domains of ArsB, some of which may be involved in H translocation. What is the nature of the trivalent metalloid substrate of ArsB? We have recently shown by extended x-ray absorption fine structure spectroscopy (EXAFS) that in solution at a neutral ph the predominant arsenic species is As(OH) 3, 2 and by analogy the antimony species would be Sb(OH) 3. Indeed, we postulate that the trihydroxides are the forms of the metalloids that are translocated by GlpF and the eukaryotic aquaglyceroporin channels (16). On the other hand, As(OH) 3 and Sb(OH) 3 are not likely to be the substrates of ArsB because these forms cannot explain the interactions observed between As(III) and Sb(III): 1) As(III) inhibits uptake of Sb(III); 2) Sb(III) stimulates As(III) uptake; 3) the rates of uptake of the two metalloids are approximately equal when the two are present at roughly equivocal concentrations; and 4) the kinetics of Sb(III) uptake become increasingly sigmoidal in the presence of As(III). One possibility is that ArsB oligomerizes, with subunit-subunit interaction; another possibility is that ArsB has separate binding sites for As(III) and Sb(III). Neither possibility easily explains the stimulation of uptake of one substrate by the other and yet reciprocal inhibition of the first by the second. A third possibility is that the substrate polymerizes analogously to phosphate, pyrophosphate, and polyphosphate. In fact, trivalent As(III) is known to readily form oxo-bridged polymers. The crystal structure of arsenious oxide, As 4 O 6,isa six-membered (As-O) 3 ring with the fourth As(III) coordinated to the three axial oxygens (22). In addition, a search of the Cambridge Structural Database identifies 109 oxo-bridged As- O-As compounds, nearly all of which are cyclic, including 10 with hexose-like six-membered (AsO) 3 rings. (Note that polymerization of (AsO) n creates even-numbered rings, and six- and eight-membered rings are the most common in the data base, whereas pentose-like five membered rings are not formed.) If these are physiologically relevant forms, why are they not visible in the EXAFS spectra? First, EXAFS is remarkably accurate in determining bond length, but it is quite insensitive to the presence of minor species. If the As(OH) 3 equilibrium with polymeric forms favors the monomeric form, the polymeric forms would probably not be observed. Second, an EXAFS assay was not performed because Sb(III) requires higher energy x-rays than were available, so it is possible that oxobridged Sb(III) species would be observed. Admittedly, this proposition relies on the existence of solution structures that have not yet been identified. However, we have recently found that As(III) is transported by most of the hexose transporters in Saccharomyces cerevisiae and have proposed that the substrate is a hexose-like six-membered (As-O) 3 ring (24). In yeast transport of As(III) via hexose, permeases are inhibited by hexoses, and glucose transport is inhibited by As(III). In contrast, ArsBmediated transport of either As(III) or Sb(III) is not inhibited by a 1000-fold excess of glucose, mannose, galactose, or fructose 2 A. Ramírez-Solís, R. Mukopadhyay, B. P. Rosen, and T. L. Stemmler, unpublished observations.

136 18340 Metalloid Transport in E. coli TABLE II Effect of sodium arsenite on the kinetics of Sb(III) uptake a As(III) Sb(III) uptake 0 M 50 M 100 M 500 M 1000 M K m ( M) V max (nmol/min/mg of protein) Hill coefficient (n) a Average of two ICP-MS measurements. Data were fit to the Hill equation using SigmaPlot version 6.1. FIG. 11. Model of metalloid transport by GlpF and ArsB. A, in cells of E. coli, As(OH) 3 (or Sb(OH) 3 ) uptake is facilitated by the GlpF channel. The ArsB antiporter exchanges a six-membered, oxo-bridged metalloid ring in exchange for a proton. Exchange of the neutral metalloid with positively charged H couples efflux to the electrochemical proton gradient. B, the true substrate of ArsB is generated by the equilibrium of three As(OH) 3 with the six-membered (AsO) 3 ring. This oxo-bridged ring is identical to the six-membered (AsO) 3 ring found in arsenious oxide (As 4 O 6 ) (22). Similarly, other substrates of ArsB are proposed to be generated in solution by the equilibrium of three Sb(OH) 3 with a six-membered (SbO) 3 ring, and mixtures of As(OH) 3 and Sb(OH) 3 in various ratios are in equilibrium with the two mixed rings shown. (data not shown). Thus, ArsB is a specific metalloid carrier and not a sugar carrier. A proposal for the substrate of ArsB that explains the present results similarly involves six-membered rings composed of As(III), Sb(III), and co-polymers of the two metalloids (Fig. 11). If the equilibrium favors rings containing Sb(III) over those containing As(III), the apparent preference for Sb(III) in ArsB catalysis may in actuality be simply the higher concentration of the (SbO) 3 substrate compared with the (AsO) 3 substrate at the same total amount of metalloid. The apparent stimulation of As(III) by Sb(III) is simply the mass action effect of Sb(III) in formation of the true substrate of ArsB. Similarly, the apparent inhibition of Sb(III) uptake by As(III) is also the result of mass action. The results of the co-transport experiment (Fig. 9) appear to require 1:1 co-transport, but these results could also be explained by formation of the two co-polymer forms (Fig. 11B) that would appear as an average of 1:1 co-transport. The most parsimonious explanation for the apparent cooperativity that As(III) imposes on Sb(III) uptake is that the rate-limiting step in uptake is formation of the substrate. As the concentration of As(III) increases, formation of the preferred (SbO) 3 is impeded. The increase in the Hill coefficient with increasing concentration of As(III) to values approaching 3 is consistent with a progression of six-membered rings from (SbO) 3 to (AsO)(SbO) 2 to (AsO) 2 (SbO) to (AsO) 3. Thus, rather than interactions of sites within the ArsB carrier, we propose that cooperativity is imparted by interaction of substrates. REFERENCES 1. Rosen, B. P. (2002) FEBS Lett. 529, Dey, S., and Rosen, B. P. (1995) J. Bacteriol. 177, Dey, S., Dou, D., and Rosen, B. P. (1994) J. Biol. Chem. 269, Kuroda, M., Dey, S., Sanders, O. I., and Rosen, B. P. (1997) J. Biol. Chem. 272, Rosen, B. P., Dey, S., Dou, D., Ji, G., Kaur, P., Ksenzenko, M., Silver, S., and Wu, J. (1992) Ann. N. Y. Acad. Sci. 671, Driessen, A. J., Rosen, B. P., and Konings, W. N. (2000) Trends Biochem. Sci. 25, Prakash, S., Cooper, G., Singhi, S., and Saier, M. H., Jr. (2003) Biochim. Biophys. Acta 1618, Wu, J., Tisa, L. S., and Rosen, B. P. (1992) J. Biol. Chem. 267, Reay, P. F., and Asher, C. J. (1977) Anal. Biochem. 78, Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11. Perlin, D. S., San Francisco, M. J., Slayman, C. W., and Rosen, B. P. (1986) Arch. Biochem. Biophys. 248, Perlin, D. S., Kasamo, K., Brooker, R. J., and Slayman, C. W. (1984) J. Biol. Chem. 259, Sanders, O. I., Rensing, C., Kuroda, M., Mitra, B., and Rosen, B. P. (1997) J. Bacteriol. 179, Carlin, A., Shi, W., Dey, S., and Rosen, B. P. (1995) J. Bacteriol. 177, Fu, D., Libson, A., Miercke, L. J., Weitzman, C., Nollert, P., Krucinski, J., and

137 Metalloid Transport in E. coli Stroud, R. M. (2000) Science 290, Liu, Z., Shen, J., Carbrey, J. M., Mukhopadhyay, R., Agre, P., and Rosen, B. P. (2002) Proc. Natl. Acad. Sci. U. S. A. 99, Rosen, B. P. (1999) Trends Microbiol. 7, Beck, J. C., and Rosen, B. P. (1979) Arch. Biochem. Biophys. 194, Brey, R. N., Beck, J. C., and Rosen, B. P. (1978) Biochem. Biophys. Res. Commun. 83, Mukhopadhyay, R., Rosen, B. P., Phung, L. T., and Silver, S. (2002) FEMS Microbiol. Rev. 26, Slonczewski, J. L., Rosen, B. P., Alger, J. R., and Macnab, R. M. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, Hamson, G. C., and Stosick, A. J. (1938) J. Am. Chem. Soc. 60, Chen, C. M., Misra, T. K., Silver, S., and Rosen, B. P. (1986) J. Biol. Chem. 261, Liu, Z., Eckhard, B., and Rosen, B.P. (February 14, 2004) J. Biol. Chem /jbc.M

138 FIELD APPLICATION REPORT ICP-OPTICAL EMISSION Continuous Flow Hydride Generation Using the Optima ICP Authors: Cynthia P. Bosnak, Leon Davidowski PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT Introduction Hydride generation techniques have been combined with ICP optical emission spectrometry for the routine determination of trace concentrations of As, Sb, Bi, Ge, Se and Te since More recently, several statements of work (SOW) for the U.S. EPA's contract lab program (CLP) have been published which call for "Hydride Generation Inductively Coupled Argon Plasma Optical Emission Spectroscopic Analysis" (HY-ICP) using a continuous flow system 2. In a continuous flow system, the acidified sample, blank or standard is continuously pumped and mixed with a pumped stream of reductant, usually sodium borohydride, to produce the gaseous hydrides. At the point of reaction, hydrogen gas is produced as a by-product. A flow of argon is added to this mixture and the hydrides are "stripped" into the gas phase. A gas/liquid separator allows the gaseous, hydride-containing phase to enter the ICP for analysis, and allows the remaining liquids to be pumped to waste. Limits of detection can generally be improved by about two orders of magnitude over simple solution nebulization using hydride generation. Work was done in an ICP Application Study 4, which detailed the construction and application of a simple and inexpensive continuous flow hydride generator for ICP-OES analyses using the Optima 3000 ICP system. In this study, experimental conditions were further optimized for the determination of As and Sb using the Optima 4300 DV ICP system. Experimental Instrumentation All emission data were acquired on an Optima 4300 DV ICP spectrometer. Instrumental operating conditions and parameters are listed in Table 1. The built-in three-channel peristaltic pump used for the hydride generator is a standard, software-controlled part of the ICP spectrometer system. The ICP spectrometer's mass-flow controlled "nebulizer argon" is used as the source of the hydride stripping argon.

139 Table 1. Instrumental Operating Conditions and Parameters ICP RF Power 1450 W Plasma Gas 17 L/min Auxiliary Gas 0.2 L/min Nebulizer Gas 0.50 L/min Viewing Type Axial Integration 2 / 20 seconds Read Delay 30 seconds Replicates 3 Continuous Flow Apparatus The continuous flow hydride generator (Figure 1) consisted of inexpensive, readily available components. A parts list is given in Table 2 (including the PerkinElmer part numbers). Two mixing T's (Chemifold Assembly) were used to combine the sample and reductant streams and to add the stripping argon (Figure 2). The chemically resistant plastic gas/liquid separator provided a very small volume, low sample dispersion and very fast rinse-out times. The gaseous hydrides were transported by the stripping argon flow directly into the base of the Optima 4300 ICP torch. The standard 2 mm i.d. alumina sample injector tube was used. To ICP Reductant Argon Waste Sample Figure 1. Chemifold apparatus Figure 2. Tubing diagram Reagent Preparation for As and Sb Determination 5% KI/Ascorbic Acid Solution Weigh 5 g potassium iodide and 5 g ascorbic acid into a 100 ml polypropylene bottle. Dilute to 100 ml with deionized water. 0.5% Sodium Borohydride in 0.05% Sodium Hydroxide Weigh 0.25 g of NaOH into a 500 ml polypropylene bottle. Add approximately 100 ml deionized water and swirl to dissolve the NaOH. Weigh 2.5 g NaBH 4 and add to the NaOH/deionized water mixture. Dilute to 500 ml volume with deionized water. Prepare fresh daily. Procedures For this work, sodium borohydride (NaBH 4 ) was used as the reducing agent to generate the gas phase hydride. An example of the hydride generation reaction is given below: NaBH 4 + 3H 2 O + HCl H 3 BO 4 + NaCl + 8H + E EH n + H 2 where E is the analyte element. Sodium borohydride was used at a concentration of 0.5% (w/v). This solution was stabilized with the addition of sodium hydroxide to a final concentration of 0.05%. The flow rate used was 2.2 ml/min. 2

140 Table 2. Continuous Flow System Components Quantity Item Part Number 1 Chemifold Assembly B Gas line with quick connect N mm PTFE Tubing B mm PTFE Tubing B mm PTFE Tubing and Sinker B Injector Adapter N PTFE Membrane (pkg. of 50) B Adapter C B Adapter E B Adapter A B Adapter K B Connector 1B B Silicone Tubing B PVC Tubing B mm id peristaltic pump tubing, PVC, red/red N mm id peristaltic pump tubing, PVC, blue/yellow B * mm id peristaltic pump tubing, PVC, black/white B * * Remove the center tab of peristaltic pump tubing The efficiency of the hydride generation reduction reaction is dependent on the oxidation state of the analyte; usually the reaction proceeds best from the lower oxidation states. To ensure that the analyte is in the lower oxidation state, a prereduction reaction is carried out before the hydride generation step. Table 3 lists the typical oxidation states likely to be found in samples and the reagents and procedures utilized for prereduction. The KI/ascorbic acid mixture used for a reducing agent for As and Sb cannot be used on samples in which Se or Te are to be determined because Se and Te will be reduced all the way to their elemental forms and will therefore not be available for the hydride formation reaction. It is therefore necessary to carry-out two separate pre-reduction procedures on samples for which As or Sb and Se or Te are to be determined. Bismuth could be pre-reduced by either method. Standards and blanks were treated in the same manner as the samples, and a flow rate of 3.6 ml/min was used for the sample flow. Table 3. Analyte Oxidation States and Pre-reduction Element Oxidation State Pre-Reduction Conditions Sb V III 0.2% KI/Ascorbic Acid + wait > 20 minutes As V III 0.2% KI/Ascorbic Acid + wait > 20 minutes Bi V III 0.2% KI/Ascorbic Acid + wait > 20 minutes // 7M HCl + 90 C Heat Se VI IV 7M HCl + 90 C Heat Te VI IV 7M HCl + 90 C Heat Pre-Reduction Step For the pre-reduction of Sb and As, 20 ml of sample was placed in a 50 ml polypropylene autosampler tube. To this, 2 ml of a 5% solution of KI and ascorbic acid was added. Six ml of concentrated HCl was also added, and the mixture was allowed to sit for at least 20 minutes. The tube was brought to the 50 ml mark with deionized water and the sample was ready to run. 3

141 For the pre-reduction of Se and Te, 20 ml of sample was placed in a clean 125 ml beaker and 20 ml of concentrated HCl was slowly added. The mixture was then heated on a hot plate just to a boil and then allowed to cool. The solution was then transferred to a 50 ml polypropylene autosampler tube which was diluted to the 50 ml mark with deionized water. Results Method detection limits were calculated by the three sigma method from the standard deviations of replicate measurements in the multi-element determinations of Sb, As, Bi, Se, and Te using blank solutions 4. Two-point background correction was used for the analyses. The results are listed in Table IV. The detection limits listed are on average at least 100x lower than those attainable by normal pneumatic nebulization 3 and well within environmental requirements 2. Very high sensitivities are attainable with this vapor generation technique and the hydride generation reactions and the sample transfer step are very efficient. In addition, there are no losses or carry-over problems of the type associated with spray-chamber based sample introduction systems. Table 4. Method Detection Limits ( µ g/l) Element Wavelength (nm) Detection Limit (3*SD) Sb As Bi Se Te Conclusion This report shows that a very simple continuous flow hydride generator can be constructed for use with standard ICP-OES instrument and applied to the determination of hydride forming elements at very low concentration levels. In addition, multiple elements can be determined simultaneously using the Optima 4300 DV ICP. Pre-reduction can be used to lower the oxidation state of the hydride-forming elements to insure more efficient vapor generation. Improvements in sensitivity of approximately 2 orders of magnitude can be obtained by hydride generation over standard typical pneumatic nebulization. References 1. M. Thompson, B. Pahlavanpour, S. J. Walton, G. F. Kirkbright, Analyst, 103, 703 (1978). 2. U.S. EPA CLP Statement of Work for Inorganic Analyses (HCIN), October "Guide to Techniques and Applications of Atomic Spectroscopy", The Perkin-Elmer Corporation (1991). 4. PerkinElmer ICP Application Study Number 67, L. Davidowski, March Perkin Elmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA Phone: (800) or (+1) For a complete listing of our global offices, visit PerkinElmer, Inc. All rights reserved. The PerkinElmer logo and design are registered trademarks of PerkinElmer, Inc. Optima is a trademark of PerkinElmer, Inc. or its subsidiaries, in the United States and other countries. All other trademarks not owned by PerkinElmer, Inc. or its subsidiaries that are depicted herein are the property of their respective owners. PerkinElmer reserves the right to change this document at any time and disclaims liability for editorial, pictorial or typographical errors. The data presented in this Field Application Report are not guaranteed. Actual performance and results are dependent upon the exact methodology used and laboratory conditions. This data should only be used to demonstrate the applicability of an instrument for a particular analysis and is not intended to serve as a guarantee of performance _01

142 FIELD APPLICATION REPORT ICP-OPTICAL EMISSION Trace Level Analysis of Calcium, Magnesium, Potassium and Sodium Using the Optima ICP Authors: Cynthia P. Bosnak Perkin Elmer Life and Analytical Sciences 2000 York Road / Suite 132 Oak Brook, IL Introduction The U. S. Geological Survey (USGS) is the lead federal agency for the monitoring of wet atmospheric deposition in the United States. The USGS atmospheric-deposition program provides continuous assessment of the chemical constituents in precipitation throughout the United States. They conduct scientific research to evaluate the effects of atmospheric deposition on aquatic and terrestrial ecosystems. 1 The National Atmospheric Deposition Program (NADP) monitors wet atmospheric deposition at 250 sites throughout the United States. 2 The precipitation at each site is collected weekly and analyzed for hydrogen, sulfate, nitrate, ammonium, chloride and base cations (such as calcium, magnesium, potassium, and sodium). 3 Many times, the volume of sample collected is just a few milliliters with the need to perform many analytical tests. As a result, there is a need to perform the analysis using the least amount of sample possible. Also, the levels of Ca, Mg, K and Na can be at very low concentrations in these types of samples. The Optima 4300 DV ICP system was evaluated to determine if it could simultaneously determine Ca, Mg, K and Na in precipitation samples at a detection limit equal or lower than 1.0 µg/l while consuming no more than 2 ml of sample. Experimental Instrumentation The Optima 4300 DV ICP system equipped with an AS-93plus autosampler (with an integrated rinse pump) was used for these analyses. The demountable torch cassette (Figure 1) included a 2.0 mm alumina injector and a single slot, quartz torch. Attached to the torch cassette was a cyclonic spray chamber with a concentric glass nebulizer (Type C) as shown in Figure 2.

143 Figure 1. Demountable torch cassette Figure 2. Cyclonic spray chamber and concentric nebulizer attached to torch cassette Although the concentric nebulizer provides the lowest detection limits, this nebulizer may clog if the samples are not filtered prior to analysis. To overcome this limitation, a GemCone nebulizer, optimized for low nebulizer gas flows, may be used with the cyclonic spray chamber. The GemCone nebulizer (Figures 3-4) has a large sample orifice that resists clogging and is made from corrosion resistant PEEK material which tolerates high acid concentrations. Figure 3. The GemCone nebulizer Figure 4. GemCone/cyclonic sample introduction system The use of the SCD (Segmented Charge-coupled Device) solid state detector provides very low levels of detection. The SCD has excellent quantum efficiency, particularly compared to the use of photomultiplier tubes (PMT). This improvement is most notable at the longer wavelengths used for Na and K (Figure 5). % Quantum Efficiency vs Wavelength SCD PMT(R955) % Q Wavelength (nm) Figure 5. Plot of Quantum Efficiency vs. Wavelength Axial viewing of the plasma (Figure 6) provides a longer path length which typically improves detection limits by about a factor of five to ten. However, with axial viewing there can be a higher occurrence of matrix effects. Often, these effects can be corrected with the use of an internal standard. An internal standard should be selected which is not present in the sample and does not contain the analytes of interest. It is very difficult to find an internal standard that does not contain trace levels of Ca, Mg, K or Na. For these reasons, some internal standard elements rejected in this work were Y, Ga, and Sc. Platinum was chosen because it contained negligible amounts of the metals of interest. The internal standard solution was added to all solutions automatically using an on-line addition kit (Figure 7). The instrumental conditions and sample parameters are shown in Table 1. 2

144 Figure 6. Axial viewing of the plasma Figure 7. On-line internal standard addition Table 1. Instrumental and Method Parameters Elements Ca Mg Na K Pt Auto Integration Wavelength nm nm nm nm nm 1-5 seconds Plasma Parameters Plasma Gas Auxiliary Gas Nebulizer Gas Power Plasma View Sample Flow Rate Sample Flush Time Sample Flush Rate Wash Time 17 L/min 0.2 L/min 0.8 L/min 1200 W Axial 2 ml/min 20 seconds 3 ml/min 10 seconds (after every sample) Peak Processing Peak Algorithm Spectral Correction Internal Standard Calibration Equation Standard/Sample Units Peak Area, 3 points/peak 2 point background correction Platinum (all analytes) Linear, calculated intercept mg/l Sample Analysis As expected, one of the most difficult parts of this analysis is getting the sample introduction system clean enough to determine low levels of Ca, Mg, K and Na. With the demountable torch cassette design, it is very easy to switch to a different sample introduction system. In addition, this design incorporates a short transfer line between the spray chamber and torch to minimize sample carry-over and contamination. 3

145 Prior to analysis, the nebulizer, spray chamber, injector adapter, injector, and torch were soaked for 30 minutes in warm, diluted (~10%) nitric acid. This action provided a clean sample introduction system to measure these low-level analytes. Standards A 10 mg/l multielement standard was diluted to provide instrument calibration standards. The diluted concentrations of the standards were 0.1, 1.0 and 2.0 mg/l. This range was chosen because the precipitation samples can contain a wide range of concentrations, from less than the limit of detection up to 2 mg/l. Results In this study it was important to achieve detection limits of 1 µg/l or lower. The average standard deviation of seven different analyses of the precipitation samples and the estimated detection limits are shown in Table 2. Table 2. Detection Limits Element Wavelength (nm) Standard Deviation (µg/l) Estimated Detection Limit (3σ, µg/l) Ca Mg Na K Although more sensitive wavelengths for Ca exist at nm and nm, they were not used in this study because it would not be possible to measure up to 2.0 mg/l Ca at these wavelengths. However, should lower Ca detection limits be necessary, these more sensitive Ca wavelengths could be used for analysis. Conclusion The Optima 4300 DV system is capable of performing trace level analysis of Ca, Mg, K and Na at levels less than 1 µg/l. The use of the Optima 4300 DV provides the simultaneous ICP analysis of the four required analytes for precipitation samples: Na, Mg, K and Ca. The sample volume required for this analysis is approximately 2 ml which enables even small sample quantities of precipitation samples to be analyzed by this technique. References 1. USGS Fact Sheet FS , U.S. Geological Survey, U. S. Department of the Interior, December, Acid rain data and reports, USGS, 3. National Atmospheric Deposition Program History and Overview, Perkin Elmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA Phone: (800) or (+1) For a complete listing of our global offices, visit PerkinElmer, Inc. All rights reserved. The PerkinElmer logo and design are registered trademarks of PerkinElmer, Inc. Optima and GemCone are trademarks of PerkinElmer, Inc. or its subsidiaries, in the United States and other countries. All other trademarks not owned by PerkinElmer, Inc. or its subsidiaries that are depicted herein are the property of their respective owners. PerkinElmer reserves the right to change this document at any time and disclaims liability for editorial, pictorial or typographical errors. The data presented in this Field Application Report are not guaranteed. Actual performance and results are dependent upon the exact methodology and laboratory conditions used. This data should only be used to demonstrate the applicability of an instrument for a particular analysis and is not intended to serve as a guarantee of performance _01

146 FIELD APPLICATION REPORT ICP OPTICAL EMISSION Determination of Impurities in High-purity Gold Authors: Kapil Dev Khullar, Ph.D. PerkinElmer Technical Centre 108 Gateway Plaza Hiranandani Gardens Powai, Mumbai India Introduction A method has been developed for the determination of known impurities in high-purity gold. The gold concentration/purity is calculated by the difference from the sum of the determined concentrations of all other impurities. The reliability of the method, in comparison with fire assay for gold analysis, is verified. The impurities analyzed are Ag, Cu, Fe, Ni, Pb, Pd, Pt and Zn. Sc is added to all sample and standard solutions as the internal standard. Experimental Calibration Standards A multielement standard containing 1 mg/l of Ag, Cu, Fe, Ni, Pb, Pd, Pt and Zn was prepared from 1000 mg/l single element standards in 2% v/v nitric acid. A 2% v/v nitric acid was used as the blank. Sc was added to the multielement standard as the internal standard at a final concentration of 1 mg/l. Sample Preparation The gold sample (0.1 to 0.15 g) was weighed accurately to ± 0.01 mg and then dissolved in 20 ml aqua regia (HCL : HNO 3 ; 3 :1) with gentle heating on a hot plate. After complete dissolution, Sc stock solution was added to obtain the final concentration of 1 mg/l. Each solution was made up to 100 ml with deionized water. Instrumental A PerkinElmer ICP model Optima 2000 DV, equipped with an AS-91 autosampler using ICP Winlab32 software was used for the analysis. The Optima 2000 DV was configured with a Ryton Scott-type spray chamber, a standard torch and a GemTip crossflow nebulizer. The operating parameters are listed in Table 1. The wavelengths and the viewing mode for each element are shown in Table 2.

147 Table 1. Instrument Conditions for Optima 2000 DV Parameter Setting RF Power 1300 W Nebulizer Flow 0.80 L/min Auxiliary Flow 0.20 L/min Plasma Flow 15 L/min Sample Flow 1.5 ml/min Source Equilibration Time 15 seconds Background Correction Manual selection of points Measurement Processing Mode Peak Area Auto Integration 1-10 seconds (min-max) Read Delay 60 seconds Rinse Delay 20 Replicates 2 Table 2. Wavelengths and Viewing Modes for Each Element Element Wavelength Survey Lower Survey Upper Points/Peak Viewing Mode Ag Axial Cu Axial Fe Axial Ni Axial Pb Axial Pd Axial Pt Axial Sc Axial Zn Axial Results The purity of the gold was established up to or 99.9 % indirectly by analyzing total known impurities. Two customer samples and a reference standard (Aldrich 99.99%) were analyzed and results are presented in Table 3. The known impurities checked were Ag, Cu, Fe, Ni, Pt, Pd, Pb and Zn. Initially 20 elements were checked, all of which can find their way into the gold either from the source or during the refining process. However, V, Cr, Mn, Co, Al, Ca, Mg, Na, K, Cd, Bi and Hg were present at very low concentrations which do not affect the purity of the gold. The results were also compared to the results of cupellation, which is a high-temperature refining process for non-oxidizing metals. A second procedure was also used to analyze the impurities in gold. In this case, instead of an internal standard, a matrixmatched standard containing all the known impurities was prepared in 0.1% solution of gold (99.99% purity). Since the concentration of impurities is insignificant in the gold used as the matrix, the results were found to be comparable to results obtained with the normalization procedure using the internal standard. 2

148 Table 3. Experimental Data Obtained for Samples Element Experimental Data (wt. %) Aldrich G.K. Exim TT Ag Cu Fe Ni Pb Pd ND* ND* Pt ND* Sc Zn * None detected Conclusion In this work, the purity of gold was calculated by the difference from the sum of the determined concentration of all other alloying elements. The analysis of gold purity by ICP-OES has advantages compared to the fire assay technique, since all alloying elements can be simultaneously determined including non-precious metals that cannot be determined by cupellation. The method presents a fast, simple and accurate way of determining gold purity using PerkinElmer Optima 2000 DV ICP- OES. Acknowledgement This work was carried out at Agee Gold Refinery at Shirpur, Maharashtra, India. PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA Phone: (800) or (+1) For a complete listing of our global offices, visit PerkinElmer, Inc. All rights reserved. The PerkinElmer logo and design are registered trademarks of PerkinElmer, Inc. Optima, WinLab and GemTip are trademarks and PerkinElmer is a registered trademark of PerkinElmer, Inc. or its subsidiaries, in the United States and other countries. All other trademarks not owned by PerkinElmer, Inc. or its subsidiaries that are depicted herein are the property of their respective owners. PerkinElmer reserves the right to change this document at any time without notice and disclaims liability for editorial, pictorial or typographical errors. The data presented in this Field Application Report are not guaranteed. Actual performance and results are dependent upon the exact methodology used and laboratory conditions. This data should only be used to demonstrate the applicability of an instrument for a particular analysis and is not intended to serve as a guarantee of performance _01

149 FIELD APPLICATION REPORT ICP OPTICAL EMISSION Determination of Impurities in Purified Terephthalic Acid (PTA) Authors: Kapil Dev Khullar, Ph.D. PerkinElmer Technical Centre 108 Gateway Plaza Hiranandani Gardens Poway, Mumbai India Introduction Polyethylene terephthalate (PET) is one of the largest volume synthetic fibers produced globally. The feed stock for PET is highly purified terephthalic acid (PTA). Metallic impurities, particularly cobalt and manganese, present at trace concentrations, can cap the polyester chain during the process and result in an inferior quality polymer product. A method has been developed for the determination of impurities in PTA by ICP-OES. Impurities determined are Co, Mn, Fe, Ni, Cr, Cu, Ca, Mg, Zn, Al and Na. Experimental Calibration Standards A multielement standard containing 5 mg/l of Co, Mn, Fe, Ni, Cr, Cu, Ca, Mg, Zn, Al and Na was prepared from 1000 mg/l single-element standards (Emerck, Germany) in 2% v/v nitric acid. A 2% v/v nitric acid was used as the blank. Sample Preparation The PTA sample (100g) was ashed in a muffle furnace at 500 ºC. The ash was dissolved in hydrochloric acid and the volume was made up to 25 ml with deionized water. A reagent blank having the same acid concentration was used. Instrumental A PerkinElmer ICP model Optima 2000 DV, equipped with an AS-91 autosampler using ICP Winlab32 software was used for the analysis. The Optima 2000 DV was configured with a Ryton Scott-type spray chamber, a standard torch and a GemTip crossflow nebulizer. The operating parameters are listed in Table 1. The wavelengths and the viewing mode for each element are shown in Table 2.

150 Table 1. Instrument Conditions for Optima 2000 DV Parameter Setting RF Power 1300 W Nebulizer Flow 0.80 L/min Auxiliary Flow 0.20 L/min Plasma Flow 15 L/min Sample Flow 1.5 ml/min Source Equilibration Time 15 seconds Background Correction Manual selection of points Measurement Processing Mode Peak Area Auto Integration 5-10 seconds (min-max) Read Delay 60 seconds Rinse Delay 30 Replicates 2 Table 2. Wavelengths and Viewing Modes for Each Element Element Wavelength Survey Lower Survey Upper Points/Peak Viewing Mode Co Axial Mn Axial Fe Axial Cr Axial Ni Axial Cu Axial Ca Axial Mg Axial Zn Axial Al Axial Na Axial Results Five PTA samples were analyzed and the results are presented in Table 3. The total impurities are listed at the bottom of each column and are a sum of all analyzed impurities in the sample. The total impurities found were 0.3 to 2.39 µg/g. The sale value of the product depends on the level of total impurities in the sample (less than 5 µg/g requirement for this customer). Since no reference sample was available, the results were confirmed by the customer. 2

151 Table 3. Experimental Data Obtained for Samples Element Experimental Data (µg/g) Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Al Ca Co Cr Cu Fe Mg Mn Na Ni Zn Total Impurities Conclusion This method demonstrates the simultaneous analysis of PTA impurities at very low levels using the Optima 2000 DV. Producing a high-quality end product of polyethylene terephthalate requires that the highly purified terephthalic acid feed stock be free of metallic contaminants, particularly cobalt and manganese. The method described here using the PerkinElmer Optima 2000 DV was shown to be effective for rapid determination of multiple elements at ppb levels. Acknowledgement PTA samples were received from Reliance Industries Ltd., Hazira, Gujarat, India. PerkinElmer Life and Analytical Sciences 710 Bridgeport Avenue Shelton, CT USA Phone: (800) or (+1) For a complete listing of our global offices, visit PerkinElmer, Inc. All rights reserved. The PerkinElmer logo and design are registered trademarks of PerkinElmer, Inc. Optima, WinLab and GemTip are trademarks and PerkinElmer is a registered trademark of PerkinElmer, Inc. or its subsidiaries, in the United States and other countries. All other trademarks not owned by PerkinElmer, Inc. or its subsidiaries that are depicted herein are the property of their respective owners. PerkinElmer reserves the right to change this document at any time without notice and disclaims liability for editorial, pictorial or typographical errors. The data presented in this Field Application Report are not guaranteed. Actual performance and results are dependent upon the exact methodology used and laboratory conditions. This data should only be used to demonstrate the applicability of an instrument for a particular analysis and is not intended to serve as a guarantee of performance _01

152 Spectrochimica Acta Part B (1998) Correction for non-spectroscopic matrix effects in inductively coupled plasma-atomic emission spectroscopy by internal standardization using spectral lines of the same analyte Assad S. Al-Ammar, Ramon M. Barnes* Department of Chemistry, Lederle Graduate Research Center, University of Massachusetts, Box 34510, Amherst, MA , USA Received 1 May 1998; accepted 30 June 1998 Abstract The common analyte internal standardization (CAIS) technique was extended to correct for non-spectroscopic matrix effects in inductively coupled plasma-atomic emission spectroscopy (ICP-AES) measurements. The technique is based on simultaneous measurement of two different spectral lines of the same analyte. A matrix correction factor is then estimated from its linear correlation with the ratio of intensities of these two measured lines. Experimental tests with four elements (Ba, La, Mg, and Mn) in three matrices (NaCl, H 2 SO 4, HNO 3 ) demonstrate a significant decrease (from 3 to 22 times) of the matrix effect after correction Elsevier Science B.V. All rights reserved Keywords: Matrix effect correction; Inductively coupled plasma-atomic emission spectroscopy; Common analyte internal standardization (CAIS) 1. Introduction The existence of non-spectroscopic matrix effects in inductively coupled plasma-atomic emission spectroscopy (ICP-AES) leading mostly to suppression and in some cases to enhancement of the analyte signal is firmly established [1]. Matrix effects are either nebulizer-related phenomena resulting from gross changes in the physical properties (density, viscosity and surface tension) of the nebulized solution [2] or plasma-related effects [3 8]. These plasmarelated matrix phenomena are attributed to several factors: (a) ionization suppression by an excess of electrons from the matrix; (b) increased collisional excitation as a result of the increased number of high-energy * Corresponding author. Fax: electrons; or (c) analyte signal suppression resulting from ambipolar diffusion. Unfortunately, owing to the existence of several interrelated factors, describing the matrix effect by a quantitative theory is not practical [7]. Accordingly, all the methods used to correct for matrix effects are empirical. The existence of a substantial matrix effect has important ramifications in practical analysis. For example, standard and sample solutions should be closely matrix-matched to ensure that the sensitivity for any analyte line is consistent throughout. However, sometimes matrix-matching is not possible, either because the matrix is inherently variable from sample to sample or because catch weights of the sample are analyzed. When matrix-matching is not practical, several techniques could be used to correct for matrix effects /98/$ Elsevier Science B.V. All rights reserved PII S (98)

153 1584 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) These include standard addition and internal reference approaches. However, the former increases the time required for an analysis by several hundred per cent for a single analyte [1,6]. Internal standardization is not always recommended, because of the considerable number of complications that may arise. In complex matrices the results can be distorted, and several internal reference elements may be required for samples in which many analytes are determined [1]. Other approaches could also be used. These include matrix stripping, the generalized standard addition method, interactive matrix matching, the parameterrelated internal standard method, mathematical correction by curve fitting to an empirical function, and matrix-swamping and excitation-buffering techniques. However, all have limitations as discussed by Thompson and Ramsey [6]. de Boer and co-workers developed an empirical procedure for the reduction of mixed-matrix effects in ICP-AES with an internal standard and proportional correction [9,10]. The aim of the present study is to develop a practical and efficient method to correct matrix effects. The common analyte internal standardization (CAIS) technique recently developed by the authors [11] seems to offer a promising, novel approach to improve matrix effects. The CAIS technique was originally developed to correct for drift in ICP-AES measurements and has not been applied to matrix effects. The approach has a flexible theoretical framework that allows it to be applied to correct for the change in several ICP-AES parameters. Accordingly, the CAIS technique is extended here to reduce the matrix effect. 2. Theory and derivation The CAIS technique is based on simultaneous measurement of two different spectral lines of the same analyte. One of the lines is used as a measure for the concentration (analyte line) and the other line as internal reference for matrix effect correction. Accordingly, in this technique there is no need to add an internal reference element to the sample solution as is typically required by the conventional internal reference method. The CAIS theory requires that the analyte line intensity, I a, should be affected by the change in the parameter P (i.e. drift, matrix, etc.) with a rate different from that of the internal reference line intensity, I b. Mathematically, this could be stated as follows: (I a I a )/I a is not equal to I b I b /I b. These two terms represent the relative intensity changes of the analyte, a, and internal reference, b, when the intensities of the respective lines are changed from I to I by the effect of changing the parameter P (i.e. the matrix) from P to P (i.e. matrix concentration changes from C to C). Consequently, the larger the difference between the relative intensity changes of the analyte and the internal reference, the more desirable a pair they become when using the CAIS technique. This approach is converse to the conventional internal reference technique in which the more closely the changes in relative intensities of the analyte and internal reference are matched, the more desirable a pair they become. Therefore, CAIS is a more general and more powerful approach than the conventional internal reference approach. In CAIS, an analyte internal reference pair that consists of an atomic line ionic line pair is a perfect combination. However, atomic atomic or ionic ionic line pairs could also be used provided that the difference in excitation energies of the two lines in the pair is large. In the CAIS technique the second step that follows the selection of the proper analyte internal reference lines pair is the development of a linear mathematical formula that relates the correction factor I a /I a to a function, f, that contains I a and I b as independent variables: I a =I a = L + Mf(I a, I b ) (1) where L and M are constants. Eq. (1) need not be a linear relationship. However, linear relations are preferred for simple and accurate correction of matrix effects Example of use of Eq. (1) The use of Eq. (1) is demonstrated in the following example: assume that an analyte concentration is to be determined in a series of sample solutions containing unknown and varying concentrations, C, of the matrix using a standard solution containing concentration C of the matrix. (Note that it is a general practice when measuring samples containing varying concentrations

154 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) of the matrix to use a standard solution containing a matrix concentration that lies midway between the upper and lower values of matrix concentrations expected in the measured samples.) The difference between the concentrations of the matrix in the samples and the standard makes the determined analyte concentrations inaccurate. To correct for this matrix effect using Eq. (1) assumes that the numerical values of the constants M, L and the exact algebraic form of the function f are already known. Then from the measured values of I a and I b of the sample, the numerical value of f could be calculated and subsequently used to calculate the correction factor I a /I a. The calculated correction factor is then multiplied by the measured intensity I a of the analyte in the sample to transform it to I a that corresponds to the predicted intensity of the analyte in the sample if the latter contains a matrix concentration equal to that in the standard solution. The corrected intensity I a is then used to calculate the correct concentration. The values of the constants L and M can be determined by assuming the algebraic form of f is known by using the following procedure: a series of three or four control solutions are prepared in such a way that exactly the same amount of the analyte is added to each member of this series. However, varying amounts of the interfering matrix should be added, so that the first and the last members of this series contain matrix concentrations that are lower and higher than the matrix concentrations expected in all the measured samples. These control solutions should be measured to determine I a and I b. From the results of this measurement, Eq. (1) is used to plot the corrections factors I a /I a against the values of f for all the measured control solutions. From the straight line thus obtained, the values of M and L are determined as the slope and the intercept, respectively. The value of I a in the correction factor I a /I a is taken to be the measured intensity at the control solution that contains a matrix concentration equal to that in the standard Algebraic form of function f The exact algebraic form of the function f in Eq. (1) can be determined with the aid of the following three-step procedure: 1. An exact mathematical representation of the analyte intensity I versus matrix concentration curve is established first. These curves represent the enhancement or suppression effect of the matrix on the intensity of the analyte and internal reference lines. 2. It is difficult to obtain the exact algebraic form of f by performing rigorous mathematical analysis on the mathematical representation of the intensity matrix concentration curves obtained in step 1. Therefore, a tentative mathematical treatment based on simplified condition is performed first. The result of this tentative treatment is expected to produce an approximate or at least a hint about the exact algebraic form of f. 3. The approximate algebraic form of f obtained in step 2 is then used as a base for more rigorous mathematical analysis that involves conditions that simulate exactly the experimental conditions typically used in routine chemical analysis. This rigorous mathematical analysis is expected to produce the exact algebraic form of f Applications The application of steps 1 through 3 to establish the exact algebraic form of f for a matrix effect is illustrated in the following. The matrix effect on the analyte intensity in relation to matrix concentration resembles an exponential decay curve [6,12] given in Eqs. (2) and (3). I = I e K(C C) (2) I = I e K(C C ) (3) I and I are the intensities of the spectral line in the presence of matrix concentrations C and C, respectively. Here C represents the matrix concentration in the standard solution. K is a constant. Eq. (2) is used when the matrix suppresses the analyte line, while Eq. (3) is applied for matrix enhancement. We tested this exponential relation under our experimental and instrumental arrangements. These operating conditions are typical of those widely applied for routine analysis and are discussed in detail in the experimental section (Table 1). The effects of 0 to 1 M HNO 3, 0 to 1 M H 2 SO 4, and 0 to 2 M NaCl

155 1586 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) Table 1 Operating conditions ICP system Optima 3000 Rf power 1.1 kw Frequency (free running) 40 MHz ICP torch Type 2 quartz slotted extension Torch injector Ceramic alumina Outer argon flow rate 15 l min 1 Intermediate argon flow rate 1.0 l min 1 Central gas flow rate 1.0 l min 1 Nebulizer Gem Tip cross flow Sample pump rate 1.0 l min 1 Spray chamber Ryton Scott Double Pass Drain Pumped matrices were investigated on several Mg, Mn, Ba, and La atomic and ionic lines. These measurements were followed by evaluating the applicability of Eqs. (2) and (3) by plotting ln I against the matrix concentration C. Straight lines were obtained in all the cases studied in excellent agreement with the logarithmic forms of Eqs. (2) and (3): ln I = ln I + KC KC (2 ) ln I = ln I KC + KC (3 ) To obtain an indication of the form of f or, if possible, to establish its approximate form, one must convert the exponential version of Eqs. (2) and (3) to a simpler algebraic form. The latter is easier to analyze mathematically than the exponential form. Therefore, the exponential Eqs. (2) and (3) were expanded using Taylor Series to a simpler version. The Taylor expansion for e X is given as e X = e a +(x a) e a +[(x a) 2 e a ]=2! + +[(x a) n e a ]=n! 4 where X = K C and a = KC. Since this is only a tentative mathematical analysis with the aim of establishing the approximate form of f, then performing this mathematical analysis is made easier by using simplified approximate conditions. Therefore, the following simplifications were made: (a) Expanding Taylor Series from zero concentration, C = 0, (i.e. a = 0 in Eq. (4)). (b) Assuming that only a very small difference in matrix concentration exists between the analyzed samples and the standard. (c) Assuming the value of the constant K in Eqs. (2) and (3) is much smaller than 1.0. As a result of the assumption in (b), only small suppression or enhancement occurs, so that it is valid to state that: (I I)=I (I I)=I (5) The approximation in (c) deserves some explanation. The value of K in Eqs. (2) and (3) is always less than one when using molar concentration units [6,12]. However, the assumption that the value of K is much less than one is employed in (c) to make the sum of the terms with powers higher than one in Eq. (4) much less than the sum of the first two terms: e a + (x a)e a. Accordingly, Eq. (4) can be approximated by Eq. (6): e x = e a +(x a) e a (6) Furthermore, considering the simplified condition in (a), Eq. (4) could be converted to: e x = 1 + X (7) By using Eq. (7), Eqs. (2) and (3) can be written as Eqs. (8) and (9), respectively. (I I)=I = KC (8) and (I I )=I = KC (9) By simple mathematical manipulations using the assumption stated in (b) and its implication stated by Eq. (5), Eqs. (8) and (9) can be converted to Eq. (10). I a =I a = L + M(I a =I b ) (10) where the subscripts a and b represent the analyte and internal reference lines, respectively. Eq. (10) is valid when both the analyte and the internal reference lines are enhanced or suppressed at the same time, or when one of them is suppressed while the other is enhanced. It could be inferred from Eq. (10) that the approximate form of the function f is I a /I b. This algebraic form will subsequently be used as a base for more rigorous mathematical analysis with the aim of revealing the true form of f. In this mathematical analysis Eqs. (2) and (3) are used without expansion and without any simplified assumptions to produce the algebraic form of f under conditions that simulate

156 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) exactly the experimental conditions typically used for routine chemical analysis. Accordingly, starting from Eq. (2) (i.e. both the analyte and internal reference lines are suppressed by the matrix): For the analyte line: I a =I a = e K a(c C) (11) and for the internal reference line: I b =I b = e K b(c C) (12) Dividing Eq. (11) by Eq. (12) yields Eq. (13): (I b =I a )(I a =I b )=e (K a K b )(C C) (13) From Eq. (11): I a =I a = e K a(c C) (14) After subtracting Eq. (14) from Eq. (13), I a /I a becomes I a =I a =(I b =I a )(I a =I b ) e (K a K b )(C C) + e K a(c C) (15) By applying a similar mathematical treatment we found that when the analyte line is enhanced while the internal reference line is suppressed, the following equation could be obtained: I a =I a =(I b =I a )(I a =I b ) e (K a + K b )(C C ) + e K a(c C ) (16) 2.4. Numerical testing The behavior of Eqs. (15) and (16) was numerically tested under conditions that simulate typical operations used in routine analysis. In this numerical test C (the matrix concentration in the standard solution) was taken to be 0.3, 0.5 and 1.0 M, while the matrix concentrations in the samples were allowed to change by 0.3 M around the specified values of the matrix concentrations in the standards. The values of K a and K b used in this simulation are 0.5 and 0.8, respectively. These values are selected to be higher than those generally found in true situations to make the result of the simulation more reliable. The results of this simulation indicate that under the chosen conditions the variation of the exponential terms in Eqs. (15) and (16) are much slower than that of I a /I a and I a /I b. Therefore, Eqs. (15) and (16) behave, under the selected conditions, as if they represent a straight line equation similar to Eq. (1) with I b /I a as the slope and the exponential terms as the intercept. The plots of I a /I a against I a /I b for Eqs. (15) and (16) under some of the selected conditions are illustrated in Figs. 1 and 2, respectively. Fig. 1. Plot of the correction factor I a /I a against line ratio I a /I b using Eq. (15) showing a matrix suppression of both the analyte and internal reference lines. Slope 3.34, intercept 0.667, correlation coefficient

157 1588 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) Fig. 2. Plot of the correction factor I a /I a against analyte line ratio I a /I b using Eq. (16) showing a matrix enhancement of the analyte and suppression of internal reference lines. Slope 0.734, intercept 1.385, correlation coefficient Experimental 3.1. Instrumentation A commercial ICP-AES system (Perkin-Elmer Optima 3000) was used for all the experiments. The experimental operating parameters (Table 1) are selected based on the optimization of Hellingess and Krampitz [13], and no additional optimization was undertaken. Experimental wavelengths are listed in Table Reagents and standard solutions Commercial atomic emission standards (SPEX Plasma standards 1000 mgml 1, Metuchen, NJ, USA) were used to prepare three types of test solutions: sample, standard, and control solutions. Each of these test solutions contained a mixture of La, Mg, and Mn. The control solutions contained 10.0 mgml 1 of each element prepared in 0, 0.25, 0.5, 0.75, and 1.0 M H 2 SO 4 or HNO 3 matrix solutions and in 0, 0.5, 1.0, 1.5, and 2.0 M NaCl matrix solutions. The sample solutions contained 5.00 mg ml 1 of each element in matrix-free and in 1.0 M HNO 3 or H 2 SO 4 and 2.0 M NaCl matrix solutions. The standard solutions contained 10.0 mg ml 1 of each element in 0.5 M HNO 3 or H 2 SO 4 and 1.0 M NaCl matrix solution. Barium control solutions contained 5.00 mg ml 1 Ba in 0, 0.125, 0.25, and 0.5 M HNO 3 and 0, 0.25, 0.5, and 0.75 M NaCl as matrix solutions. The sample solutions contained 2.50 mg ml 1 Ba in matrix-free Table 2 Analyte and emission wavelengths Element Ionization potential (ev) Analyte wavelength (nm) Excitation potential (ev) Internal reference wavelength (nm) Excitation potential (ev) Ba II* II 5.70 La II II 3.94 Mg I II 4.36 Mn I II 4.8 *II Ionic line, I Atomic line.

158 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) solution and in 0.5 M HNO 3 and 0.75 M NaCl matrix solutions. The standard solutions contain 5.00 mg ml 1 Ba in 0.25 M HNO 3 and in 0.5 M NaCl matrix solutions. Distilled deionized water was used as a blank and to prepare the solutions. High purity HNO 3,H 2 SO 4 and NaCl (Fisher Chem Alert) were used to prepare the matrices. The purity of these materials was checked experimentally by ICP-AES for the presence of Ba, Mn, Mg and La and found to contain undetectable amounts Method The instrument was first stabilized for drift by allowing it to run for at least 1 h after which the samples, standards, blanks, and control solutions were measured. The procedure for matrix correction was then followed in exactly the same way as discussed in the theory section. 4. Results and discussion The proposed CAIS procedure was tested by applying it to correct for the effects of three different matrices, HNO 3,H 2 SO 4 and NaCl, on four different elements, Mn, Mg, La, and Ba (HNO 3 and NaCl only). In this study the difference in matrix concentration between the standards and the samples for the elements Mn, Mg and La is 0.5 M (HNO 3,H 2 SO 4 ) and 1.0 M (NaCl). This difference is larger than expected in practical applications. Accordingly, the results of this study will validate the practical applicability of CAIS. However, in the case of Ba the difference in matrix concentration between the standards and samples is 0.25 M (HNO 3 ) and 0.5 M (NaCl), which is comparable to that found in most applications. The reason for selecting a low matrix concentration in Ba solutions is because Ba was found to be more sensitive than the other elements to matrix effects. Its exponential intensity matrix concentration curves reach the saturation region at a lower matrix concentration than observed for the other three elements. The elements in Table 2 were selected because they are resistant to hydrolysis in neutral aqueous solutions. Thus the acid matrix concentration can be varied over a wide range. The analyte internal reference couples selected for Mn and Mg represent atomic ionic line combinations. This is a perfect combination for the matrix effect correction, because of the large difference in behavior between atomic and ionic lines under the influence of the matrix. This difference in behavior makes the function I a /I b a rapidly varying function to changes in matrix concentration. Thus, the function provides a means for an accurate matrix effect correction. A perfect atomic ionic line combination is not available for some elements, such as Ba and La. Using only ionic ionic lines, combination is possible owing to the absence of sensitive Ba and La atomic lines. In this situation two ionic lines that are as different as possible in their excitation potential should be selected to ensure the largest possible variation in the function I a /I b with the matrix concentration. This criterion was applied when selecting the Ba and La ionic ionic line combinations. The results obtained from the measurement of the control solutions were used to plot the correction factor I a /I a against I a /I b. Straight line plots were obtained for all combinations of four elements and three matrices studied. Some of these plots are illustrated in Figs. 3, 4, 5 and 6. These results validate the theoretical base of the CAIS technique and all the equations that were derived from this theoretical base. Although the regression fit of data in some of the figures could be improved with a second-order polynomial function (r 2 = vs for Fig. 2., vs for Fig. 3, vs for Fig. 4), the error in the correction factor introduced by using the linear regression is less than 2%. Furthermore, employing a secondorder polynomial-derived correction factor would unnecessarily complicate the experimental requirements, since at least one additional control solution would be required. Thus, to maintain a simple matrix correction, only linear regression fits were applied in this investigation. Results obtained when analyzing samples containing matrix concentrations that differ from that in the standards are summarized in Table 3 for CAISmatrix-effect corrected and uncorrected data. The large differences between the corrected and the uncorrected data sets indicate the efficiency of the CAIS technique for matrix correction. However,

159 1590 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) Fig. 3. Correction factor curve for Ba nm analyte line using Ba nm internal reference line in HNO 3 matrix. Slope , intercept 0.741, correlation coefficient the differences between the expected and the corrected values in Table 3 are higher than those caused by only statistical variations. The experimentally determined statistical variations are very small and equal to 0.02, 0.04, 0.05, and 0.04 for 2.5 ppm Ba, and 5 ppm La, Mg and Mn, respectively. This error could be explained by assuming that the determined algebraic form I a /I b of the function f needs some refinement. Also the error could arise from the large difference in the concentrations of the matrix between the samples and the standard. This difference is much larger than is expected in practical situations and represents an extreme case. Fig. 4. Correction factor curve for Mg nm analyte line using Mg nm internal reference line in H 2 SO 4 matrix. Slope 6.71, intercept 3.63, correlation coefficient

160 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) Fig. 5. Correction factor curve for La nm analyte line using La nm internal reference line in NaCl matrix. Slope , intercept 3.64, correlation coefficient Conclusion Spectral lines of the same element can be used as internal references for accurate and efficient matrix effect correction. The CAIS technique is simple to apply and does not require sophisticated mathematical calculations. The approach is not time-consuming and can be adopted for routine applications. This technique, however, can be applied only when two spectral lines of the same element are found with different excitation potentials (if the two lines are ionic) or one of the lines is an atomic line while the Fig. 6. Correction factor curve for Mn nm analyte line using Mn nm internal reference line in NaCl matrix. Slope , intercept 1.31, correlation coefficient

161 1592 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) Table 3 Matrix effect correction using the lines of the same analyte as internal reference Element Matrix Matrix concentration (M) Analyte concentration (ppm, mgml 1 ) Sample Standard Expected Found uncorrected Found corrected Mn H 2 SO NaCl HNO Mg H 2 SO NaCl La H 2 SO NaCl HNO Ba NaCl HNO other is an ionic line. Nevertheless, the possibility that two lines cannot be identified is expected to be rare owing to the multielement capability of the spectrometer system and should not be considered a fundamental limitation. Acknowledgements The authors thank the Perkin-Elmer Corporation (Norwalk, CT, USA) for providing the Optima 3000 prototype system. This investigation was supported by the ICP Information Newsletter. References [1] M. Thompson, R.M. Barnes, Analytical performance of inductively coupled plasma atomic emission spectrometry in inductively coupled plasma, in: A. Montaser, D.W. Golightly (Eds.), Analytical Atomic Spectrometry, 2nd ed., Wiley, New York, 1992, p [2] R.F. Browner, Fundamental aspects of aerosol generation and transport, in: P.W. Boumans (Ed.), Inductively Coupled Plasma Emission Spectrometry, Part II: Applications and Fundamentals, Wiley-Interscience, New York, 1987, pp [3] L.M. Faires, C.T. Apel, T.M. Niemczyk, Intra-alkali matrix effects in the inductively coupled plasma, Appl. Spectrosc. 37 (1983) [4] M. Blades, G. Horlick, Interference from easily ionizable element matrices in inductively coupled plasma emission spectrometry, Spectrochim. Acta, Part B 36 (1981) 881. [5] L. Paama, P. Peramaki, Matrix effects due to calcium in argon plasma: analysis of calcitic mortars by ICP-OES, J. At. Spectrosc. 18 (1997) 119. [6] M. Thompson, M. Ramsey, Matrix effects due to calcium in inductively coupled plasma-atomic emission spectrometry, Analyst 110 (1985) [7] X. Romero, E. Poussel, J.M. Mermet, The effect of sodium on analyte ionic line intensities in inductively coupled plasma atomic emission spectrometry: influence of operating conditions, Spectrochim. Acta, Part B 52 (1997) 495. [8] A. Fernandez, J.M. Mermet, Influence of operating conditions on the effects of acid in inductively coupled plasma atomic emission spectrometry, J. Anal. At. Spectrom. 9 (1994) 217. [9] J. de Boer, M. Velterop, Empirical procedure for the reduction of mixed-matrix effects in ICP-AES using an internal standard and proportional correction, Fresenius J. Anal. Chem. 356 (1996) 362. [10] J. de Boer, W. Van Leeuweem, U. Kohlmeyer, P. Brengem, The determination of chromium, copper and nickel in ground water using axial plasma ICP-AES and proportional correction

162 A.S. Al-Ammar, R.M. Barnes/Spectrochimica Acta Part B (1998) matrix effect reduction, Fresenius J. Anal. Chem. 360 (1998) 213. [11] A. Al-Ammar, R. Barnes, Correction for drift in ICP-OES measurements by internal standardization using spectral lines of the same analyte as internal reference, At. Spectrosc. 19 (1998) 18. [12] F. Maessen, H. Balke, J. De Boer, On matrix and acid effects in the analytical practice of ICAP-AES, ICP Inf. Newsl. 8 (1982) 21. [13] D. Hellingess, P. Krampitz, ICP-OES analysis of complex alloys containing Ni, Cr, Cu and Al, At. Spectrosc. 15 (1994) 254.

163 Anal. Chem. 1998, 70, A Drift Correction Procedure Marc L. Salit* and Gregory C. Turk Chemical Science and Technology Laboratory, Analytical Chemistry Division, National Institute of Standards and Technology, Gaithersburg, Maryland A procedure is introduced that can mitigate the deleterious effect of low-frequency noisesoften termed driftson the precision of an analytical experiment. This procedure offers several performance benefits over traditional designs based on the periodic measurement of standards to diagnose and correct for variation in instrument response. Using repeated measurements of every sample as a drift diagnostic, as opposed to requiring the periodic measurement of any given sample or standard, the analyst can better budget the measurement time to be devoted to each sample, distributing it to optimize the uncertainty of the analytical result. The drift is diagnosed from the repeated measurements, a model of the instrument response drift is constructed, and the data are corrected to a drift-free condition. This drift-free condition allows data to be accumulated over long periods of time with little or no loss in precision due to drift. More than 10-fold precision enhancements of analytical atomic emission results have been observed, with no statistically significant effects on the means. The procedure is described, performance data are presented, and matters regarding the procedure are discussed. Analytical instruments transduce chemical properties to a signal, which is accompanied by noise. This noise will have a power spectrum encompassing many frequencies, all of which obscure the analytical signal. The frequency distribution of the noise will have different effects on a practical analysis: for our purposes, we can define high-frequency noise as noise that results in a poor signal-to-noise ratio for the measurement of a given sample, and low-frequency noise as noise that results in a poor signal-to-noise ratio for repeated measurements of a given sample. High-frequency noise is commonly treated in several fashions, including time-correlated internal standardization. 1 Low-frequency noise is often called driftsa slowly varying change in instrument background or sensitivity. Various filtering schemes (analog or digital) have been applied to both high 2 and low-frequency noise, 3-5 with varying degrees of general applicability. Treatment of drift is most often performed by periodic reestablishment of the relationship between measured signal and chemical compositionsrecalibration. 6 Recalibration approaches demand extra time spent measuring the standard, time that would be better spent measuring samples. Presented here is a more efficient, simple, high-performance drift correction approach that should have general utility for precision chemical metrology. It is useful here to represent the measured signal as a true signal perturbed by high- and low-frequency noise, as defined above: S measured ) S truth + ɛ drift + ɛ noise (1) where S measured is the measured signal, S truth is the true signal, ɛ noise is the high-frequency noise, and ɛ drift is the low-frequency noise, or drift. In a drifting system (one dominated by low-frequency noise, ɛ drift > ɛ noise ), time is also an enemy of the measurement; depending on the rate of drift, a get-in-and-get-out approach (where the experiment is designed to take as short a time as possible) may yield more robust results, since the magnitude of the drift is limited by the experiment duration. For precision measurements with well-characterized uncertainties, long integration times (a simple method for low-pass filtering) and replicate measurements are often required, demanding long experiment times. In such scenarios, system drift often puts a lower limit on the measurement precision that can be attained, either by determining the maximum permissible experiment time or by its contribution to measurement variability. Consider an example where five samples are to be analyzed, with five replicate measurements, using an external standard approach for calibration. In a drifting system, the analyst might choose the traditional experimental design that alternates measurement of the standard and the samples, illustrated in the upper section of Figure 1. This design requires as many as 50 measurements to analyze these samples when adjacent standards are used to calibrate a given sample. Such a design compensates for drift simply by frequent recalibration. A more sophisticated use of this type of design exploits the change in response of the measurements of the standard to infer instrument response at the time the sample was measured, through some sort of interpolation. Some efficiency may be gained, with some sacrifice in responsiveness to drift, by measuring more samples between repeated measurements of the standard. We introduce here a more efficient and effective procedure designed to be optimally responsive to system drift, which places (1) Myers, S. A.; Tracy, D. H. Spectrochim. Acta 1983, 38B, (2) Collins, J. B.; Ivaldi, J. C.; Salit, M. L.; Slavin, W At. Spectrosc. 1990, 11, (3) Rutan, S. C.; Bouveresse, E.; Andrew, K. N.; Worsfold, P. J.; Massart, D. L. Chemom. Intell. Lab. Syst. 1996, 35, (4) Wienke, D.; Vijn, T.; Buydens, L Anal. Chem. 1994, 66, (5) Hartley, R. W. Lab. Pract. 1979, 28, (6) Svehla, G.; Dickson, E. L. Anal. Chim. Acta 1982, 136, Analytical Chemistry, Vol. 70, No. 15, August 1, 1998 S (98)00095-X CCC: $ American Chemical Society Published on Web 06/09/1998

164 Figure 1. Experiment designs for the traditional drift correction through recalibration approach and the proposed, efficient drift correction approach. fewer demands on the run order of samples and standards and which permits efficient distribution of measurement time for the samples and standards. A simple analysis of this procedure relies on the premise that ɛ drift is some relatively smooth function of time that is independent of what samples are being measured. Equation 1 is used to estimate a time trend representing ɛ drift by separating the noise sources from an estimate of S truth. This estimate of ɛ drift as a function of time, ɛ drift (t), is then used to correct the measured data to a nominally drift-free condition. For each sample, the grand mean of the individual signals S measured (t), Ŝ measured, is used as the estimate of S truth, and the deviations of the individual S measured values from Ŝ measured are used as an estimate of ɛ drift + ɛ noise. The separation of these noise terms is performed by fitting a smooth curve to the deviations, where the curve is the estimate of ɛ drift (t), and the residuals of the fit are an estimate of ɛ noise. The drift-corrected signals are calculated by adding the estimated drift, ɛ drift (t), to the measured signals S measured (t), yielding 0 a corrected series of signals, S measured (t). This procedure requires that replicate measurements are performed on all samples and standards in order to obtain Ŝ measured, an estimate of S truth. A already noted, this is a common practice when precise analytical results are desired. The fitted model of ɛ drift is based on all the measurements, regardless of what is being measured (any standard or sample with sufficient a signal-to-noise ratio that the condition ɛ drift > ɛ noise is met). This approach uses both the standards and the samples as diagnostics of the system drift, while the traditional approach uses only the periodic measurements of the calibration standard as the diagnostic. The lower part of Figure 1 depicts the experiment design for our example of five samples using this more efficient drift correction approach. Not only is the experiment shorter in elapsed time, which allows less time for drift to occur, but this is a more efficient way to distribute measurement time between the samples and the standards. The flexibility permitted by this drift correction approach allows the analyst to budget more efficiently the time spent determining the calibration relationship (i.e., measuring standards) and the time spent measuring samples. Multiplicative Drift. The discussion to this point assumes that the drift is additivesthat the magnitude of the perturbation is independent of the signal level. This is typical of a system with a drifting baseline or offset. Equation 1 describes this type of drift. Equation 2 models multiplicative drift, which is dependent on signal level. This is typical of a system with drifting sensitivity, or calibration curve slope. Because different samples (hence different signal levels) are being used to characterize the drift, this method requires that the signal drift be characterized as either additive or multiplicative. When the drift is additive, as presented in eq 1, it is modeled directly from the deviations. When the drift is multiplicative, as in eq 2, the deviations are mean scaled (made into relative deviations), and the drift is modeled from these relative deviations. For multiplicative drift, we again use Ŝ measured as an estimate of S truth, and the mean scaled deviations, S measured /Ŝ measured, are fitted to a smooth function, ɛ drift (t), permitting the series S measured (t) to be corrected as in eq 3. 0 S measured S measured ) S truth (1 + ɛ drift + ɛ noise ) (2) ) [Ŝ measured (1 - ɛ drift (t))] + S measured (3) Implementation. Development of this drift correction procedure resulted from the need for a precise method for the value assignment of analyte mass fraction in single-element spectrometric solution Standard Reference Materials (SRMs). These materials are intended to be used as primary standards for calibrant preparation for determination of elemental composition. Inductively coupled plasma-optical emission spectroscopy (ICP- OES) was selected for its broad elemental coverage, relative simplicity, and performance capabilities. Our laboratory had previous, successful, experience with ICP-OES in a similar application, using a dual internal standard approach to help compensate for drift. 7 Long experiment times are required to minimize and accurately assess uncertainties, and instrument drift is the limit to precision. The value assignment experiment relies on the comparison of standards prepared from different source materials, under different conditions, with potentially different levels and species of impurities, in different quantities, and by different analysts. The ability to distinguish between solutions that are close in composition is directly related to the precision of the measurement, which can (7) Beck, C. M., II; Salit M. L.; Watters, R. L.; Butler, T. A.; Wood, L. J. Anal. Chem. 1993, 65, Analytical Chemistry, Vol. 70, No. 15, August 1,

165 be considered as the concentration resolution of the experiment. The target relative uncertainty for the value assignment is 0.3%, precision performance more typically expected from classical methods. Though developed for ICP-OES, this method should be applicable to any system where the high-frequency noise is small enough that drift dominates the precision, allowing the drift to be well characterized. A general scheme for application of the procedure is as follows: (1) Design the experiment measurement run order to include repeated measurements of samples and standards that exhibit signal much greater than the detection limit. (2) Measure the set of samples and standards. (3) Select whether the drift to be corrected is additive or multiplicative (use eq 1 or eq 2). (4) Calculate means for each sample and standard (Ŝ measured ). (5) Calculate deviations. These are calculated as S measured - Ŝ measured if the drift is additive, and as S measured /Ŝ measured if the drift is multiplicative. (6) Plot time series of deviations; select and fit model (for example, select order of polynomial, perform least-squares estimation). Model selection and fitting should be performed with good scientific and statistical practice. 8 (7) Calculate corrections for individual samples. For additive drift, these corrections are directly predicted from the fitted function ɛ drift (t), and for multiplicative drift, the correction is Ŝ measured (1 - ɛ drift (t)). (8) Calculate drift-free time series of signals by adding corrections to measured signals, S measured (t). EXPERIMENTAL SECTION We present two examples of this drift correction procedure applied to high precision ICP-OES measurement of single-element solutions. The ICP-OES instrument used in this experiment is a Perkin-Elmer Optima 3000 XL, an axial-view ICP with solid-state array detection and an integrated Perkin-Elmer AS-91 autosampler. 9 Data processing is performed external to the instrument software, in a spreadsheet program. All measurements are performed with time-correlated, or real-time, internal standardization. 1 This is made possible through the explicit selection of the integration times to be used for the measurement of the different spectral regions for the analyte and the internal standard. The concentration ratio of the analyte and internal standard must also be chosen to allow such fixed-time integration, typically selected to yield photoelectron count rates within a factor of 2 for the lines of interest and permitting high signal-to-noise ratio measurements for both signals. The two long-duration experiments each measured six different zirconium solution samples. In each experiment, three samples are aliquots of a single solution (being evaluated for solution homogeneity) and three are different comparison standards. The experiment design employs duplicate preparations (dilution and (8) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes in C, 2nd ed.; Cambridge University Press: Cambridge, 1992; Chapter 15. (9) To adequately describe experimental procedures, it is occasionally necessary to identify commercial products by manufacturer s name or label. In no instance does such identification imply endorsement by the National Institute of Standards and Technology, nor does it imply that the particular products or equipment are necessarily the best available for that purpose. Figure 2. Schematic representation of the experiment design. The different symbol shapes denote different solutions; open and dotted symbols of the same shape are from different sample preparations. Circles, squares, and upward-pointing triangles are aliquots of a single solution, downward-pointing triangles, diamonds, and hexagons are three different solutions. Table 1. ICP-OES Experiment Parameters and Operating Conditions Spectroscopic Parameters analyte Zr analyte wavelength nm analyte excitation energy cm -1 internal standard Y internal standard wavelength internal standard excitation cm -1 energy ICP Parameters plasma flow 15 L min -1 auxiliary flow 0.5 L min -1 nebulizer flow 0.8 L min -1 sample uptake 1 ml min -1 power 1300 W Measurement Parameters signal measurement mode peak integration, high-resolution readout integration time manual, 20 ms measurement time 10 s (sum of ms integrations) replicate measurements 10 addition of internal standard) of each of these six solutions to permit distinction between preparation effects and statistically significant differences among the six different Zr solutions. This experiment design is schematically described in Figure 2. The shapes used here are used throughout the following figures, with the different shapes indicating the different samples and the open or dotted shape indicating the different preparations. The solutions were prepared for analysis gravimetrically, diluting the 10 mg g -1 Zr solutions to 10 µg g -1 in two stages of about 32:1. Yttrium, used as the internal standard element for zirconium, was added at the second stage, mixed with the 2% (v/ v) HNO 3 diluent, to 10 µg g -1. An artifact of the gravimetric dilution procedure is that the internal standard is present at a slightly different level in each solution. However, the relative amounts of internal standard are well-known and the measured signal ratios are corrected for these differences, accounting for the mutual dilution of analyte and internal standard. The experimental parameters and operating conditions are summarized in Table 1, which has three sections describing the spectroscopic, ICP, and measurement parameters. ICP signal quantitation was performed in peak integration mode, using four-pixel summation under the peak, with two-point background correction. Typical signal levels were on the order 3186 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

166 of to counts s -1, and background levels were less than 1% of the signal. Figure 3. Data transformation through the stages of the first experiment. See Figure 2 for description of the symbols. (a) Zr/Y signal ratios as measured. (b) Zr/Y signal ratios corrected for sample dilution. (c) Noise, drift, and drift model in the Zr/Y ratios. (d) Zr/Y signal ratios corrected for sample dilution and drift. RESULTS AND DISCUSSION The results from the two long-duration experiments are presented to illustrate the performance of the drift correction procedure and the ability, after drift correction, to detect very small differences between samples. These data are corrected for multiplicative drift, as the spectral background correction employed eliminates baseline effects. The observed instrumental performance is well suited to this drift correction procedure because the short-term precision of the Zr/Y intensity ratio is excellent, averaging 0.02% relative standard deviation in both experiments. In the first example, the drift in the Zr/Y intensity ratio was (0.4% over a period of more than 9 h. In the second example, the drift in the Zr/Y intensity ratio was (2.3% over a period of more than 13.7 h. The different panels of Figure 3 illustrate the stages of the drift correction procedure and make clear the ability to deduce different information about the samples at each stage. Figure 3a is a time trend of the measured ratios of Zr signal to Y signal, the raw data. See Figure 2 for the symbol descriptions. At this stage of data treatment, the variation in the ratio levels is due to the variation in both the parent sample and the amount of dilution (which effects both the analyte and the internal standard level). Figure 3b depicts the time trend of the Zr/Y ratios, corrected for dilution. This time trend is the S measured data for this experiment. In this panel, the drift behavior is obvious, as is its sample-to-sample correlated nature. This correlation is a clear signature of instrument drift. The different levels of the different solutions are also evident, with the six solutions derived from the samples of the SRM solution clustered together (implying homogeneity) and with the preparation-to-preparation effect also evident. These same data are also summarized in Figure 4a, which shows box plots of the dilution-corrected results for the samples in this experiment, over the duration of the experiment. When assessed in this manner (or with summary statistics), the drift is sufficient to obfuscate the subtle, yet present, preparationto-preparation effect. Figure 3c shows the ratio of these data to their sample means, and a quartic polynomial fitted to these data. This time series is the estimate of the sum of relative perturbations, ɛ drift + ɛ noise, while the fitted polynomial is an estimate of ɛ drift, the multiplicative drift observed in this experiment. Finally, Figure 3d displays the Zr/Y ratios corrected for both dilution and drift, using the multiplicative quartic model to estimate and remove drift as a function of time. These data show more clearly a preparation-to-preparation effect than the data before drift correction. Though some variability in level remains, pairwise comparisons of duplicate preparations of the same solutions exhibit the same level relationship (dotted greater than open, or vice versa). The data in Figure 3b, before drift correction, do not behave this way, rather they seem to track the drift. These data are summarized in the box plots of Figure 4b, where the narrowing of the distributions of the data is apparent in comparison to Figure 4a, as is the ability to detect the preparation effect. As a practical matter, when detectable, this preparation-to-preparation effect is the floor of the precision in our experiment. Further improvement in the precision of instrumental measurements will not improve the precision of the analytical result. In this experiment, the precision before drift correction averages 0.059% relative standard deviation of the mean, and after drift correction, the average precision is 0.014% relative Analytical Chemistry, Vol. 70, No. 15, August 1,

167 Figure 4. Box plots of Zr signal ratios for first experiment. See Figure 2 for description of the symbols. (a) Distribution of signal ratios before drift correction and (b) distribution of signal ratios after drift correction. Figure 5. Data transformation through the stages of the second experiment. See Figure 2 for description of the symbols. (a) Zr/Y signal ratios as measured. (b) Zr/Y signal ratios corrected for sample dilution. (c) Noise, drift, and drift model in the Zr/Y ratios. (d) Zr/Y signal ratios corrected for sample dilution and drift. standard deviation of the mean. This is a more than 4-fold improvement. A second experiment is presented in a manner identical to the first, to illustrate both a more significant drift correction improvement, and system drift of somewhat different behavior. Different preparations of the solutions were analyzed in this experiment, and the results are presented in Figures 5 and 6. Here, the drift was monotonic, and a quadratic polynomial was fitted as the model. There are missing data for several of the solutions because the solutions were consumed before the experiment ran its course. Notable in Figure 5d is the correlated fluctuation remaining in the results after drift correction. Despite this imperfect removal of system drift, significant precision improvement was attained. The precision before drift correction was 0.39% relative standard deviation of the mean, and after drift correction it was 0.031% relative standard deviation of the mean. This is a greater than 12-fold improvement. The results in Figures 4 and 6 dramatically demonstrate the enhanced concentration resolution available when this drift correction procedure is used. More chemical information about the samples is available. Samples that would otherwise be statistically indistinguishable are now resolved into their different concentration levels. Uncertainty Estimates. Upon careful consideration, this approach demanded examination of the uncertainty estimates for the drift-corrected data. It is to be expected that there is a sampleto-sample covariance introduced by the determination of the drift correction from multiple samples. Additionally, application of the drift model uses degrees of freedom for the model parameters. A first approach to estimating uncertainty in the drift-corrected mean 3188 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

168 Figure 6. Box plots of Zr results for second experiment. (a) Distribution of signal ratios before drift correction and (b) distribution of signal ratios after drift correction. will assume the covariance is negligible. This has been our experience, evidenced by the fact that the sample means show no statistically significant shifts from before to after drift correction. A heuristic approach to account for the loss of degrees of freedom is to correct the variance of the drift-corrected means. When the variance of the mean of a sample population is estimated, the divisor used is the degrees of freedom, typically N-1, where N is the number of measurements. N - 1 is used to account for the fact that the mean (used in the numerator) is determined from the data, and not independently. 10 So, to estimate conservatively the standard deviation of the drift-corrected data, the degrees of freedom must be corrected by the number of parameters used in the estimate of the drift-corrected mean, now not merely the average, but in fact, dependent upon the value of a polynomial. The polynomial is determined from a large number of data, in our example cases N ) 120. The correction for the number of parameters using a quartic polynomial drift model is a factor of ( 1 / 120 ) 1/2 /( 1 / 115 ) 1/2, or So long as there are many more data than parameters used to fit the model, the correction is in fact negligible. A more statistically rigorous approach to estimating the uncertainties in the drift-corrected means, including induced uncertainty from covariance, is based on a regression approach which will be described in further work. General Considerations. As the second experiment demonstrates, modeling the drift as a low-order polynomial is not ideal for every experiment. We have observed simple behavior, such as that depicted here, as well as more complex behavior. Functions with many more inflection points may be required to model the observed drift in experiments of longer duration or with less stable instruments. The low-order polynomial modeling used here was selected for its intuitive clarity and its readily understood statistical treatment. Higher-order polynomials or splines are easily computed and applied, but there is the pitfall of overfitting, and as discussed above, the use of excess degrees of freedom limits the precision improvement that can be attained. Nonparametric models are a more general solution to the drift model. Approaches such as a moving average or a more robust method such as LOWESS 11 are excellent candidates for drift modeling. Regardless of the modeling approach, the ability to estimate the number of fitting parameters is a requirement to properly estimate the standard deviation of the drift-corrected data. Two other considerations are worthy of note: the time spacing of the measurements derived from the experiment run order and the effects of outliers on the drift correction. Both of these considerations have the potential to induce small biases in what is already a relatively small correction (not more than a couple of percent in the examples presented here). Because the drift diagnostic is mean-based, the time spacing of the repeated measurements of a sample can effect the drift model. If a sample is measured more times while the instrument signal is climbing, its mean may be biased high relative to a sample that is measured more frequently while the system is drifting down. For this reason, after a run order is established for the samples to be measured (typically randomized), this same order is used for the repeated measurements from which the means are derived. This helps to ensure that sample spacing is distributed evenly over the entire experiment duration and, hopefully, evenly over the drift behavior. Subtle effects will still occur, especially if the system is drifting faster at some times than others and if there are strong slope changes in the drift behavior within the duration of a single measurement of the samples. Outlier effects also affect the drift correction by biasing the sample means, from which the drift diagnostic is calculated. We have found the procedure to be robust with respect to outliers when there are a reasonable number of repeated measurements and samples (on the order of five repeated measurements and five samples). In these cases, the outlier effect on the sample mean and the drift model is minimized. Though we have noted outliers in our data, we have noticed no significant effects in the drift-corrected data as a result of their presence. CONCLUSIONS The drift correction approach presented here is effective in reducing the uncertainty of the results of a comparison of solutions with ICP-OES. The procedure is simple and should prove to be generally useful in any analytical methodology where the precision of the results is detrimentally effected by system drift. We have demonstrated that this procedure permits the use of ICP-OES at a level of precision that was previously expected only from the classical methodsstitrimetry and gravimetrysor from isotope dilution measurements. (10) Bevington, P. R.; Robinson, D. K. Data Reduction and Error Analysis for the Physical Sciences, 2nd ed.; McGraw-Hill: New York, 1992; p 11. (11) Cleveland, W. S. J. Am. Stat. Assoc. 1979, 74, Analytical Chemistry, Vol. 70, No. 15, August 1,

169 This drift correction approach is efficient and effective. It has allowed our laboratory to extend the capabilities of an existing instrumental technique, ICP-OES, to a problem that requires precision heretofore unavailable from a non-isotope-dilution instrumental measurement. Application of this approach to other techniques should yield similar enhancements. Extension of this procedure to experimentally characterize drift as additive or multiplicative, and to adapt it to circumstances where the drift is a mix of these types, will be presented in a subsequent study. This extension will be based on the comparison of the measured results for different dilutions of a given sample. The signal relationships for multiple dilutions will be compared to the dilution relationships, allowing the separation of additive and multiplicative components. Received for review January 29, Accepted May 7, AC980095B 3190 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

170 Anal. Chem. 2001, 73, Single-Element Solution Comparisons with a High-Performance Inductively Coupled Plasma Optical Emission Spectrometric Method Marc L. Salit,* Gregory C. Turk, Abigail P. Lindstrom, Therese A. Butler, Charles M. Beck II, and Bruce Norman Chemical Science and Technology Laboratory, National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg, Maryland A solution-based inductively coupled plasma optical emission spectrometric (ICP-OES) method is described for elemental analysis with relative expanded uncertainties on the order of 0.1% relative. The single-element determinations of 64 different elements are presented, with aggregate performance results for the method and parameters for the determination of each element. The performance observed is superior to that previously reported for ICP-OES, resulting from a suite of technical strategies that exploit the strengths of contemporary spectrometers, address measurement and sample handling noise sources, and permit rugged operation with small uncertainty. Taken together, these strategies constitute high-performance ICP-OES. Since early 1997, inductively coupled plasma optical emission spectroscopy (ICP-OES) has been used to perform comparisons of candidate elemental solution Standard Reference Materials (SRMs) against well-characterized primary materials. These SRMssthe 3100 seriess provide a basis for the accuracy of inorganic elemental analysis. They are prepared in bulk and batch certified for the mass fraction of the constituent element, typically at 10 g/kg levels. The results of these comparisons are now an integral part of the SRM certification process, establishing traceability of the SRM to the primary materials. ICP-OES is being employed here as an element-specific twopan balance. To obtain performance comparable to isotope dilution and classical methods (titration and gravimetric analysis), we have developed a suite of tools which, used together, constitute high-performance ICP-OES (HP-ICP-OES). A key enabler is a drift correction procedure invented to address what was the major source of uncertainty observed in our ICP-OES results. 2 The other essential components of HP-ICP-OES are described and developed in a paper describing the characterization of a LiAlO 2 ceramic material. 3 Using HP-ICP-OES, we typically observe relative expanded uncertainties on the order of 0.1%. This level of uncertainty is * To whom correspondence should be addressed. [email protected]. (1) Salit, M. L.; Turk, G. C. Anal. Chem. 1998, 70, (2) Salit, M. L.; Vocke, R. D., Jr.; Kelly, W. R. Anal. Chem. 2000, 72, (3) Guide to the Expression of Uncertainty in Measurement, 1st ed.; ISO: Switzerland, quantitatively different from that previously reported for ICP-OES results. Different chemical information becomes available with this ability to discriminate between solutions of very similar compositionsan improvement in concentration resolution. Additionally, at this level of uncertainty, uncertainty is dominated by effects other than spectroscopic measurement (e.g., sample handling and preparation). Measurements are performed using unmodified commercially available equipment and are significantly less costly than classical analysis. The opportunity to achieve results of the quality expected of classical analysis with instrumental efficiency enables new analytical applications. In contrast to classical analysis, instrumental automation permits the analysis of multiple samples with little incremental cost per sample. Similar to classical analysis, this method is suitable for determination of major and minor constituents, with analytes typically introduced to the instrument at 10 mg/kg, with no concomitant elements present at levels high enough to cause interferences and, therefore, no expectation of bias. HP-ICP-OES has been used in the certification analyses for 64 different single element SRMs in the 3100 series (some elements have been certified more than once with HP-ICP-OES). Additional analyses include a Key Comparison (K8) of the Comité Consultatif pour la Quantité de Matière of the Comité International des Poids et Mesures (the CCQMsConsultative Committee for Amount of Substance - of the CIPMsInternational Committee for Weights and Measures) 4 and two multielement analyses: the certification of the major constituents of a high-temperature alloy SRM 5 and the characterization of LiAlO 2. 3 This paper will serve to describe how HP-ICP-OES can be used to determine major inorganic constituents in solution with uncertainty on the order of 0.1% relative. A performance overview and a detailed procedural guide for the single-element determination of 64 elements are presented. EXPERIMENTAL SECTION HP-ICP-OES is composed of several technical strategies, as outlined in refs 2 and 3. It is a ratio method, where correlated (4) comparison3 4.asp.Aaccessed 16 April (5) Ceritificate of Analysis, Standard Reference Material 2175, Refractory Alloy MP-35-N. National Institute of Standards and Technology, Gaithersburg, MD, 10 October /ac Not subject to U.S. Copyright. Publ Am. Chem. Soc. Analytical Chemistry, Vol. 73, No. 20, October 15, Published on Web 09/19/2001

171 noise sources cancel through the simultaneous measurement of an internal standard element reference signal with the analyte element signal. 6,7 It is often the case that the dominant short-term noise in the measured ICP emission signal is correlated for different elements. This correlation occurs in the noise regime where signal-carried noise is dominantswhere the signal-to-noise ratios of the emission signals are sufficiently greater than shot or detector read noise. This signal-carried noise typically arises from the common-mode sample input noise carried by all the emission signals. 6,8,9 Correlated noise does not appear in the signal ratio. Accurate and precise ratios are best measured with solid-state array detector-based ICP-OES spectrometers. Such instruments typically allow flexible selection and simultaneous measurement of multiple emission lines and the spectral background adjacent to each. Background correction is essential for calculation of accurate element signal ratios, and simultaneous measurement of signal and background permits background correction with no flicker-noise penalty. Various approaches are taken to address remaining noise sources, notably drift correction to mitigate longer term, sampleto-sample noise that cannot be addressed with internal standardization. Other important sources of uncertainty that cannot be mitigated with the measurement schemese.g., those arising from chemical form or from sample preparation and handlingsare quantified using a suitable experiment design. The signal measurement strategies associated with HP-ICP- OES are predicated on moderately high signal levelssthis is not a method that measures at or near the analyte detection limit. HP-ICP-OES was designed to make traceable measurements of elemental composition with small uncertainty. In such an application, it is typical that nominal analyte levels are known, and method development can be optimized for small uncertainties. Sample Handling. We have developed a robust and practical scheme of sample handling that minimizes sample-handling errors and their impact. Gravimetric solution handling is employed, exploiting the accuracy and ease of use of computer-integrated electronic force balances. Amount fraction is typically determined as mass fraction, eliminating uncertainty associated with density and its temperature dependence. This sample-handling scheme, adopted from isotope dilution mass spectrometry, is of general utility, whether using primary materials for calibration (as we do at NIST) or where any suitable reference material is employed as a calibrant. Sample-handling contributions to analysis uncertainty include the uncertainty in the element amount ratios (amount of analyte to amount of internal standard) in calibration solutions and uncertainty in the sample mass-to-internal standard mass ratios in unknown samples. These amount ratios can be expressed in any set of consistent unitssthe measured signal ratio is proportional to the ratio of the number density of emitters in the observed volume of the plasma, and any proportionality to this number density (e.g., atomic weight) ends up as a multiplicative factor in the calibration relationship. (6) Myers, S. A.; Tracy, D. H. Spectrochim. Acta 1983, 38B, (7) Ivaldi, J. C.; Tyson, J. F. Spectrochim. Acta 1996, 51B, (8) Cicerone, M. T.; Farnsworth, P. B. Spectrochim. Acta 1989, 44B, (9) Olesik, J. W.; Fister, J. C., III. Spectrochim. Acta 1991, 46B, Aliquots of sample and internal standard are weighed into the same vessel to minimize these uncertainties. Once the solutions are added and mixed, the amount ratio is fixed. No quantitative transfers are required before (because the same vessel is used for all weighings) or after spiking with internal standard (because the amount ratio is fixed upon mixing sample and spike), eliminating several sources of variability. All further handling can be performed without regard to quantitation, so long as the amount ratio is not perturbed (e.g., dilution, but not separations). The usually straightforward constraint is that all constituents remain in solution upon mixing. It is impractical and unnecessary for every solution to contain the same amount of internal standard, as is typical when volumetry is used for sample preparation. Though a target analyte-to-internal standard ratio is established to abet precision photometry, the target is a range, usually spanning a factor of 2. Method Development. For single-element analysis, analyteto-internal standard amount ratio and solution mass fractions are selected to optimize signal quantification. Measurements reported in this work were performed on PerkinElmer Instruments Optima 3000XL and 3300DV ICP-OES instruments. 10 The spectrometers in these instruments match the 10 9 dynamic range of the plasma to the dynamic range of the spectrometer through a combination of on- and off-detector signal integration. 11,12 These instruments are capable of on-detector integration on the order of 10 5 photoelectrons, after which signal must be accumulated offdetector. This arrangement permits both weak and strong emissions to be integrated with high precision. Long integration (on the order of 5-10 s for this work), is accomplished through multiple short integrations (typically on the order of 100 ms). Long integration enhances signal-ratio quantification, even for strong emission lines, because the noise-power spectrum of emission is dominated by frequencies higher than a few hertz. 13 Optimal internal standard noise cancellation is achieved when analyte and internal standard emission signals accumulate at similar rates, such that it is possible to simultaneously integrate both signals at high signal-to-noise ratio. The analyte-to-internal standard amount ratio is thus selected to balance photoelectron accumulation rates. Solution mass fraction is typically established such that an ondetector integration time between 50 and 1000 ms can be employed. Such signal accumulation rates are typical (for the instruments used in this work) when measurements are in the signal-carried noise domain. Shorter on-detector integration times can be used (to a minimum of 1 ms), but with a consequent increase in readout overhead and reduction in measurement duty cycle. Longer on-detector integration times are unusual, but are occasionally required due to poor emission behavior (and poor signal-to-noise ratio) of an element in the ICP. Sample-to-sample cross-contamination, or carryover; signal linearity; and signal-to- (10) To adequately describe experimental procedures, it is occasionally necessary to identify commercial products by manufacturer s name or label. In no instance does such identification imply endorsement by the National Institute of Standards and Technology, nor does it imply that the particular products or equipment are necessarily the best available for that purpose. (11) Barnard, T.; Crockett, M.; Ivaldi, J.; Lundberg, P.; Yates, D.; Levine, P.; Sauer, D. Anal. Chem. 1993, 65, (12) Barnard, T.; Crockett, M.; Ivaldi, J.; Lundberg, P. Anal. Chem. 1993, 60, (13) Goudzwaard, M. P.; de Loos-Vollebregt, M. T. C. Spectrochim. Acta 1990, 45B, Analytical Chemistry, Vol. 73, No. 20, October 15, 2001

172 blank ratio also play a role in determining appropriate nominal mass fractions for analysis. Method development typically begins with measurement of a 10 µg/g solution of analyte, permitting the instrument controller to determine the appropriate on-detector integration time for this mass fraction level. The mass fraction of analyte in the solution to be analyzed is then adjusted to permit on-detector integration time on the order of 100 ms (e.g., if the instrument selects a 20- ms integration time for the 10 µg/g solution, the analyst might select 2 µg/g as the nominal mass fraction for the analysis). A similar process is followed for the internal standard selected, targeting the same integration time as selected for the analyte. These nominal mass fractions are then used to establish the sample spiking and dilution scheme. Mass fractions above 20 µg/g are typically avoided to minimize cross contamination and carryoversas noted above, on-detector integration can be increased to several seconds as needed to achieve reasonable total signal counts. If too long an integration time is selected by the analyst, the signal will saturate and an error will be detected. If too short an integration time is selected, no error will be detected, but the signals will likely be noisy, and poor correlation between analyte and internal standard may be observed. The integration times vary from instrument to instrument, or for a given instrument, as conditions vary. We have developed a simple, practical set of criteria for internal standard selection. Selection is based upon the following: the mutual absence of the internal standard from the sample and the analyte from the internal standard solution, chemical compatibility to ensure that the elements remain in solution, the absence of spectral interferences at the wavelengths of interest, low likelihood of contamination of the sample with the internal standard by the laboratory environment, periodic similarity, excitation energy matching, and wavelength matching. The most often used internal standard element is scandium. Scandium has numerous emission lines to select from; is scarce in samples and laboratory environments; and is a strong emitter, permitting it to be used at low mass fraction. In practice, a small number of Sc lines suffice as internal standards for most of the elements. Line selection for the analyte is similarly unconstrained. Strong analytical lines are generally used, affording high signal-to-noise photometry at low mass fraction. Here again, care is taken to avoid spectral interference from nonanalyte sources. Spectra of single-element solutions of the analyte and internal standard are measured in the region of the analyte and internal standard lines, to determine that no cross-contamination or spectral overlaps occur. Also at this point, the background correction parameters are established to ensure accurate estimation of the background under both lines. The instruments we use permit the selection of different entrance slit widths, permitting a balance between spectral resolution and light acceptance. Where the nature of the spectra permit (no nearby interferences), the lowest resolution slit (greatest slit width) is selected to achieve maximum signal. Manufacturer-suggested default plasma conditionsscompromise conditionssare typically used for analysis (Table 1). Throughout the work presented here on single-element solutions, the analyst selected the analyte mass fraction for suitably high signal-to-noise Table 1. ICP Operating Conditions ICP Source Operating Parameters plasma flow 15 L min -1 auxiliary flow 0.5 L min -1 nebulizer flow 0.8 L min -1 power 1300 W sample uptake 1 ml min -1 autosampler probe rinse 15 s in 2% volume fraction HNO 3 Spectrometer Operating Parameters signal measurement mode peak integration, low-resolution readout background correction manually selected, 2-point interpolation measurement time 10 s replicate measurements 7 ratio under default conditions. Exceptions were made in the case of several of the alkali elements, where emission lines of the atomic spectra were observed for quantitation, and a lower power and higher nebulizer flow than the default was used in an attempt to suppress ionization. Axial viewing of the plasma was used for most of the analyses, again an instrument default condition for the equipment in our laboratory. Several of our instruments permit radial viewing, which offers the option to avoid viewing through the cool plasma tail plume and the option to integrate over a shorter path length. Radial viewing was employed where a resonance line (or a line from a low-lying lower energy level) of the atomic spectrum was observed for quantitation, to avoid self-absorption from cool atoms (e.g., Na) or where a very strong line emits so much light that the decreased path length makes it easier to use manageable mass fractions in sample preparation (e.g., Ba). Different configurations of sample input system have been employed for these measurements: a glass concentric nebulizer with a cyclonic spray chamber; a cone-spray -type high-solids nebulizer, also with a cyclonic spray chamber; and a sapphire orifice cross-flow nebulizer with Scott-type spray chamber. Used appropriately, any of these arrangements is satisfactory. The analyte and internal standard wavelengths, nominal analyte and internal standard mass fractions for the solutions introduced to the instrument (the samples being analyzed typically have times higher mass fractions before preparation), and experimental notes are listed for the 64 elements discussed in this study, in Table 2. Calibration. HP-ICP-OES is a relative method that compares the analyte-to-internal standard signal ratio measured in an unknown sample to those ratios measured in mixtures whose amount ratio is well known. A calibration relationship is established to infer composition of unknown samples from their measured signal ratios. Equation 1 is the relationship that permits mass fraction ( ( (IAnalyte mg g ) ) I IntStd )Unknown ( I Analyte I IntStd )Calibrant)( ) (mganalyte g IntStd )Calibrant ( g Sample g IntStd )Unknown (1) the calculation of analyte mass fraction in an unknown sample from the measured signal and mass ratios of the calibrants and the sample. Analytical Chemistry, Vol. 73, No. 20, October 15,

173 Table 2. Method Parameters and Performance Results for 64 Elements analyte internal standard performance wavelength (nm) nominal mass fraction (µg/g) wavelength (nm) nominal mass fraction (µg/g) notes rel std uncertainty of signal ratio (%) rel std uncertainty of replicate preps (%) difference from gravimetry a Ag Sc Al Mn , As Se Au In B Sc matrix was 3% mass fraction mannitol, in 2% volume fraction of nitric acid, to reduce memory effects Ba Sr radial viewing Be Sc Bi Sc Ca Sc Cd Sc Ce Mn Co Sc Cr Mn Cs Sc W, 1.1 L min nebulizer flow Cu Mn Dy Sc Er Sc Eu Sc Fe Sc Ga Sc Gd Y Ge In Hf Sc Hg In Cr 2O 2-7 ion was used to stabilize Hg: 10% K 2Cr 2O 7 in water added to concentrated solutions for spiking; after dilution, K 2Cr 2O 7 was 0.05% mass fraction Ho Sc In Sc K Sc La Sc Li Mn NB: it must be assured that any difference in atomic weight between samples and calibrants is accounted for Lu Sc Mg Mn Mn Sc Mo Sc Na Sr radial viewing Nb Mn Nd Sc Ni Sc P Se Pb Co Pd Sc Pr Sc Pt In Rb Sc Re Sc sample-to-sample memory effects Rh In Sb Sc Sc Co Se Sc Si Mn uncertainty dominated by variability of calibration standards Sm In Sn In Ta Zr diluted Ta solution before addition of IS to prevent precipitation Analytical Chemistry, Vol. 73, No. 20, October 15, 2001

174 Table 2 (Continued) analyte internal standard performance wavelength (nm) nominal mass fraction (µg/g) wavelength (nm) nominal mass fraction (µg/g) notes rel std uncertainty of signal ratio (%) rel std uncertainty of replicate preps (%) difference from gravimetry a Tb Mn Te Sc Ti Mn uncertainty dominated by variability of calibration standards Tl Sc Tm Sc U Sc V Sc W Mo uncertainty dominated by variability of calibration standards Y Sc Yb Sc Zn Sc Zr Y diluted Zr solution before addition of IS to prevent precipitation a HP-ICP-OES, gravimetry. The calibration produces an indirect inferencesthe calibration estimates the analyte-to-internal standard amount ratio in an unknown sample. The amount of analyte in the unknown sample is calculated from knowledge of the amount of internal standard (spike) added. The amount fraction (typically mass fraction) of analyte in the sample is calculated from knowledge of the amount of sample that was spiked. The experiment is simplified by using a single, thoroughly mixed solution of internal standard solution to spike all samples and calibrants in an analysis. This permits the mass of spike solution to be used in lieu of internal standard element amount in all amount ratios. Another source of uncertainty is eliminated because there is no need to accurately know the spike solution mass fraction. Analyte-to-internal standard amount ratio is expressed in somewhat peculiar units: mass of analyte element to mass of spike solution, instead of mass of analyte element to mass of internal standard element. The calibration between signal ratio and amount ratio is a simple, straight-line relationship that passes through zero, so long as the signal ratios are calculated from background-corrected intensities and the blank level is either small relative to the signal level or is measured and corrected for. HP-ICP-OES analyte and internal standard levels are selected to be significantly greater than blank levels; and the multiplex detectors in commercially available ICP-OES spectrometers permit background correction with no flicker-noise penalty, so every line intensity measurement is made with an interpolated estimate of the plasma background under the line center. 14 Calibration Materials. HP-ICP-OES lends itself to the use of calibration materials that can be prepared, stored, and used accurately. We calibrate using weighed aliquots of accurate solutions gravimetrically prepared from well-characterized primary materials. The mass of analyte element in each aliquot is well (14) Ivaldi, J. C.; Barnard, T. W. Spectrochim. Acta 1993, 48B, known. It is expected that the mass of analyte contained in such aliquots, packaged in single-use plastic bottles (typically LDPE), is stable over long periods of time. The analyte mass in solution is not affected by transpiration (of solvent only) from the container. It is anticipated that no significant loss of analyte through interaction with the container materials or formation of volatile compounds occurs. Excellent stability over several months has been established (variance indistinguishable from other noise sources). Long-term (>3 y) stability of analyte mass in the aliquots is being evaluated for a number of different analytes. For use in HP-ICP-OES (or other ratio techniques), internal standard spikes are weighed directly into the calibration aliquot containers, typically at the time of analysis. The solutions are well mixed, fixing the amount ratios, and diluted to the proper level for analysis. Alternatively, the spike can be weighed and mixed into the calibrant aliquots at any time, and aliquots of this same spike solution can be weighed out, storing this set of solutions as a kit. Regardless of whether the spike is added to samples and calibrants at the time of analysis or whether calibrants are prespiked and samples are added to preweighed spikes, no quantitative transfers are required. Our calibration hierarchy is based on high-purity solid materials designated as NIST Primary (NP) materials; solutions made from these NP materials, NIST Primary Solutions (NPS); and weighed aliquots of NPS solutions, NPS Aliquots. NPS materials are prepared in replicate, typically two solutions from each of two analysts. Directly after preparation, NPS aliquots are weighed out and the weights recorded in a database. NPS materials generally contain 1 g of analyte/kg of solution, a mass fraction that is stable for most elements in solution. The usual matrix is dilute HNO 3. Aliquots are typically 30gina60- ml bottle, containing 30 mg of analyte. This leaves adequate volume for spike addition. Analytical Chemistry, Vol. 73, No. 20, October 15,

175 Figure 1. Cause and effect analysis for uncertainty contributions. Buoyancy Correction. The varying densities of materials being weighed may introduce bias from air buoyancy effects on apparent mass, unless accounted for by correction to true mass (mass in vacuo). A relative bias on the order of 1 part-per-thousand is introduced when solutions of density 1gcm -3 in standard air are weighed. Where a ratio of two masses of matter of similar density (e.g., an unknown and a spike solution) is being calculated, buoyancy correction is typically ignored. However, because the analyte and the solution into which it is dissolved typically have different densities, both the mass of analyte in the NPS aliquots we use and the mass fraction of the 3100 series SRMs are always corrected to true mass. When these materials are used, the mass of spike solution added to the calibration materials must be buoyancy-corrected so the ratios are on a consistent scale. In practice, true mass is used for all calculations. RESULTS AND DISCUSSION Essential to the HP-ICP-OES method is an experiment design that permits robust and accurate quantitative evaluation of the uncertainty of the determined amount fraction. As demonstrated in the LiAlO 2 work, 3 uncertainty sources other than measurement dispersion contribute significantly to the uncertainty of the comparison, and care must be taken to quantify them. Uncertainty Budget. Equation 1 describes the calculation of the mass fraction of an unknown solution from the measured quantities. The result is the product of two ratiossan intensity ratio (actually, a ratio of ratios) and a mass ratio (again, a ratio of ratios). Both dispersion and the potential for bias contribute to uncertainty in the measured quantities and, ultimately, the mass fraction. Figure 1 depicts these quantities and their influences. While this cause-and-effect diagram 15,16 appears complex, it is highly symmetric, with most factors appearing in both numerator and denominator of the ratios that appear in eq 1. HP-ICP- OES owes much of its robustness to this - correlated dispersion and biases that cancel. The sole violation of symmetry is the assay component, which is a bias correction for the purity of the primary material used for calibration. Intensities. Potential biases in the intensity measurements include spectral background offset and deviation from a linear relation between signal and solution mass fraction. These are mitigated with simultaneous spectral background correction and measurement in a linear region of mass fraction for both analyte and internal standard. While variation might be observed for these biases, the dispersion from variation in the sensitivity as a function of timesthe noisesdominates intensity variability. Noise in the intensity ratio is quantified through replicate ratio measurements. Masses. Biases in each mass measurement include a sensitivity error in balance calibration and deviation from linearity in (15) Ellison, S. L. R.; Barwick, V. J. Analyst 1998, 123, (16) Eurachem/CITAC Guide: Quantifying Uncertainty in Analytical Measurement, 2nd ed.; Laboratory of the Government Chemist, London, Analytical Chemistry, Vol. 73, No. 20, October 15, 2001

176 balance response. Several considerations mitigate the effects of these biases on the method. First, all weighings for a given solution are typically performed on a single balance within a short time period and are of similar magnitudes (typically on the order of 3 g on a top-loading balance with 1-mg readability). Used in this manner, the biases related to the calibration state of the balance cancel. Second, the magnitudes of these biases can be determined from balance specifications, and experiments performed such that uncertainty associated with the presence of these biases is small relative to the (also specified) dispersion. It is interesting to note that only relative spike weights are requireds the units (and hence, the balance sensitivity) are immaterial but must be linear. This is another robust element of this sample handling routine. Experiment Design. The experiment design permits quantitative evaluation of the uncertainty from sample and calibrant preparation and handling. In most ICP-OES experiments, variation of the measured signal intensities dominates the uncertainty, and quantifying the signal variability as a standard deviation suffices. Because the variation due to handling includes the variation of the measured signal ratios, the ratio measurement standard deviation is used solely for diagnostic purposes and does not appear directly in the uncertainty budget. All steps in the preparation of solutions for measurement must be considered as sources of variability. Because the calibration material must be dissolved before analysis, multiple solutions are prepared and analyzed to assess dispersion from variability in this process. Similarly, to address the possibility of solution heterogeneity, replicate samples of the calibrants and unknowns are prepared and analyzed. We typically calibrate with duplicate aliquots of each of four different solutions, two each prepared by two different analysts. These eight solutions are measured to determine the slope and population distribution of the calibration curve, permitting proper evaluation of the uncertainty of the slope. Before calculating the uncertainty, care is also taken to determine the number of independent data contributing to the mean, so the standard uncertainty can be calculated as the standard deviation of the mean. For instance, where no significant differences are observed among the calibrants, there are eight data; if the four different solutions can be distinguished, there are only four; and if the solutions from each analyst can be distinguished, there may only be two. The importance of duplicate spiking for analysis is clear when one is trying to determine the population of calibrants. When the certification measurements for the 3100 series SRMs are performed, the batch is packaged and three randomly selected containers (either high-density polyethylene bottles or glass ampules) are sampled in duplicate. This helps to ensure that the comparison is made using a representative sample of the population of the solutions being delivered and that the uncertainty estimate includes any sample-to-sample heterogeneity (potentially due to inadequate mixing). Here again, spiking of replicate samples from each container is critical. Container-to-container variability can be distinguished from preparation variability, which permits detection of and distinction between heterogeneity and blunders. Data Evaluation. We employ a graphical tool for simple evaluation of the results on an HP-ICP-OES analysis. This chart Figure 2. HP-ICP-OES analysis performance presented as relative residuals, for Fe in a CCQM key comparison. compares the calibration slopessthe ratio of the signal ratio-tomass ratiosobserved for each solution. The data are normalized such that the mean of the calibrant slopes is 100, so uncertainties and deviations can be read directly in relative percent. These data are the relative residuals from the fitted calibration curve forced through zerosconsistent with the calibration strategy used. Examining the residuals is a sensitive way to observe the population of results for both calibrants and samples at different levels of mass fractionsexamining the calibration curve directly only shows large discrepancies, and the typical population scatter we observe would be dominated by the change in mass ratio. Such a graph is in Figure 2, for the determination of Fe in two single-element solutions in an international comparison. The eight solutions used to calibrate are labeled NPS and the samples (triplicate preparations of a single sample of each of the two different solutions) are labeled K8. The error bars are the standard deviation of the drift-corrected signals. Preparation-topreparation effects may be present in two casessnps 2-A and -B disagree as does K8 2-C disagree with -A and -B. The conservative approach treats these results as part of the population and retains them in the data analysis. The overall scatter is small, and the NPS solutions appear to arise from a single population. The observed range of calibrant slopes is 0.27%, and the standard uncertainty of the slope is 0.04% relative. The standard uncertainties for the two different K8 samples are 0.02% and 0.07%, and the relative uncertainty in the calibration material purity is %. Estimates of the degrees of freedom using the Welch-Satterthwaite formula 17 were 9 and 3, suggesting expansion factors of 2.26 and The combined expanded uncertainty of the first K8 sample is 0.09% ((0.04% % % 2 ) 1/2 2.26) and for the second K8 sample is 0.26% ((0.04% % % 2 ) 1/2 3.18). The results reported for this analysis agreed with the gravimetric reference value for the first sample and with the (17) Taylor, B. N.; Kuyatt, C. E. Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, 2nd ed.; NIST Technical Note 1297; NIST: Gaithersburg, MD, Analytical Chemistry, Vol. 73, No. 20, October 15,

177 Figure 4. Population distribution of mean signal ratio uncertainty for 64 elements. Figure 3. Population distribution of differences between HP-ICP- OES and gravimetry (N ) 30). Solution mass fractions were nominally 10 g/kg for all solutions except B (5 g/kg), the second Al listed, and Cu, Fe, and Mg (all 1 g/kg). consensus value (the only reference value available) for the second. Bias. Differences between HP-ICP-OES and gravimetric results were tabulated for 30 single-element comparisons. These differences were calculated wherever reliable gravimetric results were available and are included in Table 2. Though not completely independentsthe gravimetric preparation results were often based upon the same material as that used to prepare the HP-ICP-OES calibrantssthese gravimetric results represent our best approximation to an unbiased estimate. Equivalence of HP-ICP-OES and gravimetric results demonstrates that we are, to the best of our ability, establishing HP-ICP-OES as an element-specific twopan balance. The frequency distribution of these differences is charted in Figure 3. These data pass a Kolmogorov-Smirnov test for normality 18 at the 0.05 probability level. A fitted Gaussian is overlaid on the frequency histogram. The mean difference between the HP-ICP-OES and gravimetric results was ( mg g -1 (95% confidence interval). There is no evidence of bias from these data. Dispersion. Figure 4 charts the frequency distribution of the dispersion of the measured signal ratios for the 64 elements in our study. These dispersions are the average of, for between 12 and 24 solutions, the relative standard deviations of the means of between 5 and 10 replicate measurements. This tailed distribution has a median of 0.032%, with a maximum of 0.18%. As would be expected, the poorer ICP emitting elements are at the high tail of the distribution (As, Sb, Ce, Cs). These data are included in Table 2 as the relative standard uncertainty of signal ratio. ANOVA was used to test the measured ratios of replicate preparations of homogeneous solutions (the three candidate SRM (18) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes in C, 2nd ed.; Cambridge University Press: Cambridge, U.K., 1992; pp Figure 5. Population distribution of replication standard uncertainty for 64 elements. samples, from which duplicate spikes were prepared) for statistically significant preparation effects. This ANOVA tests whether the difference between preparations was greater than would be expected by chance, given the dispersions of the measured ratios for a given preparation. Of 21 experiments (21 different elements) examined with ANOVA, 15 showed statistically significant preparation effects among the 6 solutions (with at least 99% confidence), and 6 showed that the differences among preparations could be due to the dispersion of the measured signal ratios. This analysis justifies our inclusion of preparation as a component of uncertainty in our experiment and uncertainty budget. In addition to the ANOVA, the distribution of the standard uncertainty of replicate preparation (standard deviation of the mean of the different preparations) was tabulated. This tailed distribution is charted in Figure 5 and has a median of 0.029%. This dispersion is of the same magnitude as the dispersion of the signal ratios, and the two distributions are quite similars suggesting that neither source dominates the uncertainty. These 4828 Analytical Chemistry, Vol. 73, No. 20, October 15, 2001

178 data are included in Table 2 as the relative standard uncertainty of replicate preparations. CONCLUSIONS Evidence is presented demonstrating that the HP-ICP-OES method is generally useful for 64 elements and is an unbiased method of comparison with small uncertainty for solutions containing predominantly a single element. This comparison technique is relevant in many inorganic applications, including traceable reference material dissemination, demonstration of international comparability of measurement systems, material (19) Turk, G. C.; Yu, L. L.; Salit, M. L.; Guthrie, W. F.; Fresenius J. Anal. Chem., in press. assay, bulk compositional analysis of samples, and determination of element ratios. The underlying concepts of HP-ICP-OES are being extended to alternate spectroscopies and trace element analysis, 19 with great promise. ACKNOWLEDGMENT The authors thank Dennis Yates for useful discussions during the preparation of the manuscript. Received for review April 17, Accepted July 29, AC Analytical Chemistry, Vol. 73, No. 20, October 15,

179 An ICP-OES Method with 0.2% Expanded Uncertainties for the Characterization of LiAlO 2 Marc L. Salit,* Robert D. Vocke, and W. Robert Kelly National Institute of Standards and Technology, Chemical Science and Technology Laboratory, Analytical Chemistry Division, 100 Bureau Drive, M/S 8391, Gaithersburg, Maryland An improved inductively coupled plasma-optical emission spectrometry (ICP-OES) method 1 has been applied to the determination of Li and Al mass fractions and the Li/Al amount-of-substance ratio in representative samples of LiAlO 2. This ICP-OES method has uncertainty on the order of 0.2%, 2,3 comparable to the best analytical methods. This method is based on several strategies, which are detailed in this work. The mean measured mass fractions of Li and Al in eight samples were ( ((0.16%) and ( ((0.14%), and the mean Li/Al amount-of-substance ratio was ( ((0.17%). The uncertainty is dominated by sample handling and heterogeneitysabout a factor of 2 larger than the ICP-OES instrumental uncertainties, which were 0.04% for Al and 0.07% for Li. In the spring of 1997, our laboratory took on a project to characterize a ceramic material, LiAlO 2, for its Li isotope ratio, Li and Al mass fractions, and Li/Al amount-of-substance ratio. Pellets of this material are intended for use as a target in commercial light water reactors for the manufacture of 3 H. 4,5 Stringent engineering requirements for the bulk composition of the target material and the mass of 6 Li/unit length required both analytical method development and the generation of a quality control material for process control of production analytical measurements. The 6 Li/unit-length-of-rod specification is a critical parameter in the reactor core design and is the main motivation for these measurements. The 6 Li acts as a burnable poison in the reactor core, capturing neutrons as it is consumed to make 3 H. The presence of 6 Li depletes the neutron flux in the reactor, and its concentration must be managed to avoid poisoning the nuclear chain reaction. The 6 Li/unit-length specification is designed to permit the required neutron flux to be maintained as both the 6 Li (1) Salit, M. L.; Turk, G. C. Anal. Chem. 1998, 70, (2) Unless otherwise stated, all uncertainties reported in this work are expanded uncertainties, calculated according to procedures outlined in the ISO Guide to the Expression of Uncertainty in Measurements GUM. 3 These uncertainties represent a 95% confidence interval and are estimated from both statistically evaluated uncertainty componentsstype Asand uncertainty components evaluated by other meansstype B. All uncertainty components are expressed as standard deviations and are combined as variances. (3) Guide to the Expression of Uncertainty in Measurement, 1st ed.; ISO: Switzerland, (4) Pincus, W. TVA Plant to Supply Nuclear Bomb Material Tritium. Washington Post, December 23, 1998, A07. (5) TVA Approves Plan to Make Weapons Material. Washington Post, December 9, 1999, A13. and the reactor fuel are burned. The Li mass fraction, the 6 Li/ 7 Li ratio, and the material linear density are all required to establish the 6 Li/unit length. While the Li mass fraction is required primarily for the 6 Li/ unit-length specification, it is also needed, along with the Al mass fraction, to verify a mass balance specification (to ensure material purity, sum of all determined species > 99%). The Li/Al amountof-substance ratio is specified to be /-0.06 to ensure chemical and mechanical stability of the ceramic material. The analytical requirements arose from the need for reliable assessment against these specifications and from the need to establish a process-control material. These requirements include small uncertainties and traceability to SI. SI traceability ensures measurement comparability between different laboratories, different techniques, and measurements made at different times. Coupled with small uncertainty, traceability ensures a control material which is useful for control of accuracy over time, not merely variability assessment. Small uncertainty is assured through precision measurement approaches, coupled with management of bias; traceability is assured by calibration with primary materials and a rigorous uncertainty budget. This analysis required precise and accurate sample handling, isotope ratio measurement, and assays. The target expanded uncertaintiesstaking into account potential biasessfor the isotope ratio, assays, and amountof-substance ratio were on the order of 0.2%. ICP-OES Analysis. A Li assay with uncertainty of a few parts per thousand would conventionally be performed with an isotopedilution Thermal Ionization Mass Spectrometry (TIMS) procedure, and a comparable Al assay would be performed with gravimetry. A high-performance ICP-OES method recently developed in our laboratory provides an expedient alternative to these methods for the determination of the Li and Al mass fractions and the Li/Al amount-of-substance ratio. 1 The ICP-OES procedure is an extension of the approach developed for the analysis of single-element spectrometric solutions which are issued as NIST Standard Reference Materials (SRMs) for experiment control and instrument calibration. Potential biases in this measurement are controlled by several strategies, and the measurement yields relative expanded uncertainties at the parts-per-thousand level, as opposed to the more typical 1% level expected from ICP-OES. The multielement ICP-OES procedures for the bulk composition (Li and Al mass fractions) and amount-of-substance ratio measurements were developed for this application. ICP-OES measurements compare the number of emitting species in the /ac CCC: $19.00 xxxx American Chemical Society Analytical Chemistry A Published on Web 00/00/0000 PAGE EST: 7.3

180 Table 1. Li Sample Preparation for TIMS Isotope Ratio Measurement: Ion Chromatography Separation cation separation column AG 50W-X8, mesh, 5-mL resin bed, m i.d. elution scheme columns conditioned 10 ml of 0.5 mol/l HCl in volume fraction of 80% methanol samples evaporated to dryness, redissolved, in 1 ml of 0.5 mol/l HCl in volume fraction of 80% methanol then loaded onto conditioned columns wash 2 ml of H 2O, discarded 16 ml of 0.5 mol/l HCl in volume fraction of 80% methanol added to columns, discarded Li eluted and collected 72 ml of 0.5 mol/l HCl in volume fraction of 80% methanol eluted fraction evaporated to dryness with HClO 4 and H 2O conversion anion column AG 1-X8, 4-mL resin bed conversion to LiOH columns conditioned NH 4OH samples loaded with 2 ml of NH 4OH elute and collect LiOH with 40 ml of NH 4OH dry solution to appropriate mass fraction 100 Li µg/g solution observed volume of the source between the samples and the standards, not the mass fraction of the species. Because the Li atomic weight in the samples is deliberately perturbed, and the atomic weight of Li is variable in nature, unbiased ICP-OES mass fraction determination requires knowledge of both the sample and the calibration standard atomic weight, which were provided by the TIMS measurement of the isotope ratio, also required for characterization of anticipated material performance as a reactant for 3 H production. EXPERIMENTAL SECTION Sample Preparation. All reagents used in this analysis were of high purity. Eight hollow cylinders of LiAlO 2, approximately 2.5 cm in length and 0.8 cm in diameter, with a wall thickness of 0.15 cm and weighing 2.5 g, were each crushed in polypropylene bags and the chips transferred to clean 50-mL Nalgene polycarbonate bottles. LiAlO 2 samples of 0.3g( 5 chips) were taken from each of the eight samples, weighed, and placed into eight individual Carius tubes (internal volume, 25 ml). Ten milliliters of HCl was added to each tube. Two reagent blanks were prepared in the same manner as the samples. The contents of each tube were frozen in a solid CO 2 -CHCl 3 -CCl 4 mixture and the tube sealed with an O 2 natural gas torch. After being warmed to room temperature, the tubes were placed inside steel shells along with 20 g of solid CO 2 for external pressurization, the caps were tightened to effect a gastight seal, and the tubes were heated to 240 C in an oven for approximately 15 h. The safe handling of Carius tubes is discussed in Kelly et al. 6 and references therein. 7,8,9 After the dissolution procedure, no solid material was observed, and no light scattering was observed when the solution was laser illuminated. The tubes were opened and their contents quantitatively transferred to 125-mL polycarbonate Nalgene bottles and diluted with water to a total weight of 100 g. Aliquots of these solutions were used for the isotopic analyses and for the Li and Al assays. Li Isotope Ratio. Lithium isotope ratios ( 6 Li/ 7 Li) were measured on a TIMS of NIST design and construction. Thermal ionization sources fractionate the isotopes in a sample, typically (6) Kelly, W. R.; Paulsen, P. J.; Murphy, K. E.; Vocke, R. D.; Chen, L.-T. Anal. Chem. 1994, 66, (7) Gordon, C. L. J Research NBS 1943, 30, (8) Gordon, C. L.; Schlecht, W. G.; Wichers, E. J. Res. Natl. Bur. Stand. (U.S.) 1944, 33, (9) Gordon, C. L.; Schlecht, W. G.; Wichers, E. J. Res. Natl. Bur. Stand. (U.S.) 1944, 33, enriching the ion signal in the lighter isotope ( 6 Li). Fractionation is managed by loading samples and an isotopic reference material (IRM) in the same chemical form and measuring them in exactly the same manner, permitting absolute lithium isotopic ratios to be determined by calibration with the IRM. The certified IRM used in this study was IRMM 016 (Li 2 CO 3, Institute for Reference Materials and Measurements, European Commission Joint Research Centre, Geel, Belgium). Chemical Form of Li. The standard and samples were prepared using ion exchange chromatography to separate Li from the other cations present, and Li was converted to LiOH, then to Li 3 PO 4, before analysis. Care was taken to ensure complete recovery of Li from the ion exchange column so that no fractionation occurred in the separation. The separation is detailed in Table 1. Solutions were prepared from the eight samples and the calibrant, IRM 016, in identical fashion. Mass Spectrometry Measurements. The isotopic analysis of Li followed a modified positive-ion rhenium triple filament procedure derived from that of Sahoo and Masuda. 10 With this technique, only the central ionizing filament is heated to ablate and ionize the Li ions, while the side filaments held the Li in the form of Li 3 PO 4. Details of the sample loading, instrument description, and operating conditions are in Table 2. After heating and conditioning the sample for 1 h, baseline and sample beam currents were measured. The run order of the samples and standard was randomized and repeated. ICP-OES Measurement. The ICP-OES instrument used in this experiment is a Perkin-Elmer Optima 3000 XL, an axial-view ICP with a free-running 40 MHz RF generator, solid-state array detection, and an integrated autosampler. 11 Relevant measurement parameters are reported in Table 3. The ICP operating parameters are the default conditions, while spectroscopic measurement parameters have been selected for precise measurement of spectral intensities. Aluminum and lithium are determined at commonly used analytical transitions of the atomic spectrum; thus, excitation energy is inversely proportional to the wavelength. Manganese, the internal standard, is also measured with a commonly used (10) Sahoo, S. K.; Masuda, A. Int. J. Mass Spectrom. Ion Processes 1995, 151, (11) To adequately describe experimental procedures, it is occasionally necessary to identify commercial products by manufacturer s name or label. In no instance does such identification imply endorsement by the National Institute of Standards and Technology, nor does it imply that the particular products or equipment are necessarily the best available for that purpose. B Analytical Chemistry

181 line of the atomic spectrum. Mn is a convenient internal standard: the excitation energies of the Al and Mn lines are well matched; Mn was not present in the samples at levels relevant to its use as an internal standard; and Mn is typically well-behaved in the ICP, often used as a performance diagnostic element. No evidence of spectral interference was observed when the wavelengths selected were examined with high-purity single-element solutions. Line selection and sample preparation (including selection of internal standard mass fraction) were done such that the three spectral lines could be measured simultaneously and with high signal to noise. The dynamic range of the spectral lines must be approximately matched to that of the spectrometer to permit simultaneous integration. The approximately 0.3 g/100 g sample preparations were diluted about 30-fold with 50 µg/g Mn to produce the solutions for analysis. A single measurement of a solution required approximately 5 min, with sufficient delay to accommodate complete flushing of the sample-input system with the new solution. The run order of samples, calibrants, and blanks was randomized to minimize potential bias arising from temporal effects. This same random measurement order was repeated 10 times, permitting precise measurement of the signal ratios. Data processing was performed external to the instrument software (which reports backgroundcorrected intensities), in a spreadsheet program of our own design. ICP-OES Samples and Experiment Design. All sample handlingsdilutions and addition of internal standardswas performed gravimetrically, with relative uncertainty from weighing of better than 0.1%. Operations where the weighing uncertainty is poorer than 0.1% were performed in multiple stages, with each stage having adequate precision to ensure the desired aggregate uncertainty. Internal-standard addition was accomplished by including the internal standard in a common diluent used for all samples, blanks, and calibrants. In future work, the internal standard for ICP-OES will be added to the Carius tubes by weight, eliminating potential bias associated with the quantitative transfer and subsequent dilutions. ICP-OES Calibrant Preparation. Four Al primary solutions were prepared from two samples, each of two different high-purity Al metal samples, reported to be >99.999% purity, exclusive of gases. The Al metal was dissolved in high-purity HCL, with HNO 3 added dropwise until visible reaction occurred. These solutions were heated in covered Teflon beakers overnight, at which point clear solutions were obtained. The four Li primary solutions were prepared from high-purity lithium carbonate, Li 2 CO 3, SRM 924a. This material was coulometrically assayed for CO 3 2- as well as analyzed for impurities. The isotopic abundance of Li in this material was measured and a Li atomic weight reported on the certificate. The Li 2 CO 3 was dissolved in dilute HNO 3. ICP-OES calibrants were prepared from these primary solutions, at mass fractions spanning the anticipated mass fraction of analytes in the samples. Eight multielement calibration solutions were prepared, from the eight primary solutions. Aliquots of two primary solutions (one Li and one Al) were weighed into a vessel and diluted, and a final weight was recorded for them. The same diluent, 50 µg/g Mn in 2% HNO 3, was used for the calibrants, samples, and blanks. The exact mass fraction of Mn in the diluent Table 2. TIMS Operating Conditions Sample Loading filament conditioning outgas at 4.2 A, 35 min 24 h in HEPA filtered air filament loading and drying 2.5 µl of 0.25 mol/l H 3PO 4 and 500 ng of Li loaded on Re side filaments, mixed, dried at 1.2 A for 5 min, repeated 3 times Instrument Configuration and Operation geometry 90 sector, 30-cm radius of curvature source slit mm collector slit mm ion detection Faraday cup 7 Li ion current during A, decaying to data collection ionizing filament heating increase 0.5 A every 2 min to 1.5 A, then protocol increase 0.1 A every 1 min to filament temperature of 1800 C (by optical pyrometer), final filament current A instrument pressure < Torr replicate measurements 8-10 Table 3. ICP Operating Conditions ICP Source Operating Parameters plasma flow 15 L min -1 auxiliary flow 0.5 L min -1 nebulizer flow 0.8 L min -1 power 1300 W sample uptake 1 ml min -1 autosampler probe rinse 15 s in 2% HNO 3 sample input type C concentric nebulizer, glass cyclonic spray chamber Spectrometer Operating Parameters signal measurement mode peak integration, high-resolution readout background correction manually selected, 2-point interpolation measurement time 10 s replicate measurements 7 Elements Measured nominal wavelength mass fraction Al nm 40 µg/g Li nm 10 µg/g Mn nm 50 µg/g is irrelevant, as is the exact mass fraction of Mn in any of the samples, standards, or blanks. Internal standardization requires only that the relative mass fraction of Mn is well-known (which it is, from the masses of diluent). RESULTS AND DISCUSSION Li Isotope Ratio. The lithium isotopic analyses of the samples are detailed in Table 4. The measured 6 Li/ 7 Li ratio was ( ((0.28%), and the calculated atomic weight was ( ((0.0069%). All isotopic values are tied to IRM 016 and its currently certified 6 Li/ 7 Li value. The nuclidic masses used in the atomic weight calculation were taken from Wapstra and Audi. 12 The individual Li isotopic composition determinations are presented in Figure 1. A multiplicative factor of ( was used to correct fractionation bias. The correction factor was determined (12) Audi, G.; Wapstra, A. H. Nucl. Phys. A 1993, A432, Analytical Chemistry C

182 Figure 1. 6 Li/ 7 Li isotope ratios. Error bars are Type A uncertainties, dashed line is the mean, dotted lines are the standard deviation of a single determination. Table 4. Li Isotope Ratio and Uncertainty Components 6 Li/ 7 Li isotopic ratio Type A uncertainties degrees of freedom sample variability fractionation correction Type A Type B uncertainties degrees of freedom IRM016 calibration material uncertainty Type B combined uncertainties degrees of freedom u c expansion factor 2 U U rel 0.44% from repeated measurement of IRM 016 during the analysis sequence. Perturbation of the Li isotope ratio by the cation separation was assessed by measuring the ratio of a sample of IRM 016, which had only been processed through the anion column, and comparing this ratio to that of a sample which had been through the complete process. No such perturbation was detected. A new measurement of the Li isotope ratio in IRM 016 has been reported ( ( ), but no new certificate has been issued. 13 The original certificate value ( ( ) was used in this work. If the new value were used, the 6 Li/ 7 Li ratio in the LiAlO 2 samples changes from to , the amountof-substance fraction of 6 Li to Li changes from 20.59% to 20.74%, (13) Qi, H. P.; Berglund, M.; Taylor, P. D. P.; Hendrickx, F.; Verbruggen, A.; De Bievre, P. Fresenius J. Anal. Chem. 1998, 361, and the atomic weight of the lithium changes from to The effect of the change in the atomic weight on the ICP- OES Li assay is insignificantssmaller than the Li assay expanded uncertainty by nearly an order of magnitude. ICP-OES Method. ICP-OES analysis with such small uncertaintyshigh-performance ICP-OES (HP-ICP-OES)sis implemented with the combination of several technical strategies. The strategies include those related to the measurement of signal, addressing most of the measurement-related sources of variability which contribute to uncertainty in the results, and an experiment design which permits quantification of uncertainty when uncertainty may not be dominated by measurement variability. Experience in our laboratory is that this approach often yields results which are limited by the inhomogeneity of, and our ability to prepare and handle, the primary and sample materials. HP-ICP-OES Signal Measurement. Variability introduced by the measurement of emission intensity for a given analyte in a given solution is managed in three ways: precision photometry, time-correlated internal standardization, and quiet sample input. Contemporary ICP-OES instruments typically use a crossdispersed Echelle configuration, with rectangular-format array detectors employed in the focal plane. Such a configuration affords appropriate spectral resolution, can be configured for excellent light-gathering (sensitivity), and offers flexible and simultaneous measurement of light at multiple wavelengths. 14,15 Performance benefits arising from simultaneous measurement at multiple wavelengths include: the ability to measure both line spectra and adjacent spectral regions to permit background correction, with no signal-to-noise penalty arising from signal fluctuation ( flickernoise ) between measurements; 16 the ability to measure analyte and internal-standard wavelengths simultaneously (time-correlated (14) Barnard, T.; Crockett, M.; Ivaldi, J.; Lundberg, P.; Yates, D.; Levine, P.; Sauer, D. Anal. Chem. 1993, 65, (15) Barnard, T.; Crockett, M.; Ivaldi, J.; Lundberg, P. Anal. Chem. 1993, 65, (16) Ivaldi, J. C.; Barnard, T. W. Spectrochim. Acta, Part B 1993, 48B, D Analytical Chemistry

183 internal standard), allowing signal-to-noise enhancement through the correlation of sample-carried fluctuation; 17,18,19 and, by maximizing measurement time while minimizing experiment elapsed time, the ability to reject the effect of drift in uncontrolled system parameters on the measured signals. The sample input system employed uses reliable nebulizer/spray chamber combinations with consistent transport characteristics, leading to emission signals with high signal-to-noise and fast sample washout with negligible sample carryover or memory effects. 20,21 Uncontrolled aspects of the measurement process changing with time over the course of an experiment cause signal drift, even when the signal is the ratio of analyte intensity to internalstandard intensity. In fact, the smaller the short-term variability of the measured signal, the more effect such drift has on the uncertainty of the resultsthat is, when the signal-to-noise ratio of a single measurement is sufficiently high, an experiment becomes dominated by long-term drift. Such is the case with an ICP-OES measurement which uses the signal-measurement approach described above, where the sources of short-term noise are minimized. Long-term changes in the plasma conditions or the spectrometer can cause the analyte and the internal standard to drift differently. A drift correction procedure developed for this circumstance has been described and is an integral part of the design for the high-precision ICP-OES measurement. 1 Using this approach, drift is mitigated and long measurement times (many hours) can be employed to aid the precision photometry. HP-ICP-OES Experiment Design. Because uncertainty in the results may not be dominated by measurement imprecision, consideration must be given to other possibly relevant sources of uncertainty. An assessment of these sources is essential if metrologically valid uncertainty estimates are to be obtained. Because some of these sources of uncertainty are not typically considered for ICP-OES analyses, they are described here along with an experiment design which permits their quantification. Sample heterogeneity is evaluated with replicate sampling; variability in sample dissolution is evaluated with replicate dissolution; variability in dilution and addition of internal standard is evaluated with replicate dilution; and variability in calibrant preparation is quantified with replicate calibrant preparation. In fact, there are two potential sources of variability in calibrant preparation: variability of the primary solution mass fractions (pure, single-element solutions prepared from well-characterized solidssprimary materials), due to material purity or dissolution, and variability in the calibrants (the working solutions used to calibrate instrument response) arising from their preparation (dilution and internal standard addition) from the primary solutions. A calibrant preparation strategy designed to assess both sources of variability is depicted in Figure 2. Calibrant preparation was crossed to separately detect the different sources of variability. Two calibrants are prepared from every primary solution, capturing dilution variability; to separate (17) Myers, S. A.; Tracy, D. H. Spectrochim. Acta, Part B 1983, 38B, (18) Mermet, J. M.; Ivaldi, J. C. Real-time Internal Standardization for Inductively Coupled Plasma Atomic Emission Spectrometry Using a Custom Segmentedarray Charge Coupled Device Detector, J. Anal. At. Spectrom. 1993, 8, (19) Ivaldi, J. C.; Tyson, J. F. Spectrochim. Acta, Part B 1996, 51B, (20) Ivaldi J. C.; Slavin, W. J. Anal. At. Spectrom. 1990, 5, (21) Hettipathirana, T. D.; Davey, D. E. Appl. Spectrosc. 1996, 50, Figure 2. The crossed calibrant design. variability in the primaries from variability in the calibrants, no two calibrants share identical primary materials. This permits deductive identification of errors in preparation and handling, and, though not ideal (it can be confounded by cascaded errors), the design adds robustness to the method. Several examples help illustrate the crossed-design concept: (1) Four primary solutions are prepared from aluminum metal: Al-1, Al-2, Al-3, and Al4. Primary solution Al-1 is at lower than nominal mass fraction. Both Cal-1 and Cal-2 will demonstrate the same inconsistent response for Al (lower slope of the background and blank-corrected calibration curve) when compared with the other calibrants. If primaries Li-1 and Li-2 are consistent, the Li response for these calibrants (Cal-1 and Cal-2) would be consistent with the other calibrants, suggesting no error in preparation of the calibrants from the primaries. The flaw is traced to primary Al-1. (2) While diluting Cal-3, a drop of diluent falls on the outside of the bottle. The mass fraction of Al and Li in Cal-3 will be higher than nominal and that of Mn lower than nominal. When comparing the response for Al in Cal-3 and Cal-4, and Li in Cal-3 and Cal-1, a similar relative discrepancy will be observed. The error in preparation is identified, and the results from Cal-3 are excluded. Where no significant blank exists, and spectral background correction is employed, the calibration function passes through zero. In this caseswhich applies in this analysisscalibration uncertainty is estimated directly as the standard uncertainty of the slope of the calibration curve obtained from the population of calibration solutions. HP-ICP-OES Bias. An experiment design incorporating replication as described permits the uncertainty due to variability to be quantified, but bias, uncertainty in any bias corrections, and uncertainty due to potential bias must be considered independently. Additive biases arise from unaccounted-for spectral background or signal arising from sources other than the samples or calibrants (such as from the blank). These biases are managed with spectral background correction and blank measurement. Multiplicative biases can arise from sample transport (changes in delivery of sample to the nebulizer, nebulization efficiency, injector tube orifice clogging, desolvation) or from excitation effects (changes in effective plasma power, plasma electron density, electron temperature). Multiplicative effects arising from sample transport are mitigated through the use of the time-correlated internal-standard approach described above and used here. As sample transport efficiency changes, the atom density in the plasma for both the analyte and the internal standard are typically affected to the same Analytical Chemistry E

184 degree, and the measurand, the background-corrected intensity ratio of analyte to internal standard, is unperturbed. It is possible for excitation effects to cause uncorrelated changes in the background-corrected emission intensities for the analyte and internal standard lines, which would perturb the measurand. Excitation sources in contemporary instruments are engineered to maintain constant plasma conditions, helping to ensure constant excitation conditions. Stability on the order of a part per thousand (measured as emission intensity from a line of the Ar support gas) is typical. Despite careful source engineering, bias will still arise from uncontrolled effects on excitation conditions, which most often arise because of the nature of the sample with which the plasma is seeded. These effects are observed when the sample contains a concentration of easily ionized elements sufficient to perturb the electron density and the ionization equilibria or when the solvent physical properties (volatility, dielectric, e.g., when an organic solvent contaminates an aqueous system) are different between samples and calibrants, causing a change in the effective plasma power. Multiplicative bias arising from excitation effects is disregarded in this application, because the analyte species are the bulk matrix species, present at mass fractions in the tens of micrograms per gram. Total mass fractions at such small levels do not significantly perturb the plasma conditions, and as care is taken to use the same high-purity solvent for all samples and standards; no concomitants are present. To assess the presence of multiplicative effects, control samples are often analyzed. In this experiment, replicate calibrants at different levels act as controls; should they yield consistent calibration relationships the presence of multiplicative biases can be ruled out. ICP-OES Results. The elapsed time for the ICP-OES measurements was 24 h 44 min. In this time, 10 measurements were made of each of the 16 sample solutions (8 samples 2 preparations), four blanks (2 blanks 2 preparations), and eight calibration standards. Instrument drift over this period of time was about (2% for the Al/Mn and Li/Mn ratios and (1% for the Li/Al ratio. The drift correction procedure was applied to all ratio measurements before summary calculations. ICP-OES Blanks. Blanks were not handled rigorously, but no significant analyte signal was detected in the four method blank solutions which were measured, and no blank corrections were made to the Li or Al assays. A more rigorous assessment of the blank would have included measurement of blank solutions in the experiment design in addition to the method blanks. Such a design would permit blank source apportionment and more careful assessment of assay uncertainty due to the blank. Here, for both the Li and Al assays, a Type B uncertainty component for the blank was estimated as the standard uncertainty of the four blank solutions which were measured. Both the Li and Al assay uncertainties are dominated by other effects and were unaffected by the inclusion of the blank variability. Because of the insignificance of the blank uncertainty, no component was carried forward to the amount-of-substance ratio uncertainty evaluation. Al Assay. The Al mass fraction results are detailed in Table 5. The measured Al mass fraction was ( ((0.14% relative uncertainty). The uncertainty reported is the expanded uncertainty and is dominated by the Type B estimate of the uncertainty in the purity of the calibration material. This is Figure 3. Li and Al assays. O Al mass fraction, b Li mass fraction. Error bars are Type A uncertainties, solid lines are means, dotted lines are the standard deviation of a single determination. Table 5. Aluminum Assay and Uncertainty Components mass fraction Al Assay Type A uncertainties degrees of freedom uncertainty due to replication variability uncertainty due to calibration Type A Type B uncertainties degrees of freedom uncertainty due to blank variability uncertainty due to calibration material purity Type B combined uncertainties degrees of freedom u c expansion factor 1.97 U U rel 0.14% estimated to be 0.1%, with a rectangular distribution, which is normalized to a standard uncertainty by dividing by 3. The selection of a primary material with a smaller assay uncertainty should permit Al mass fraction determination with sub-part-perthousand uncertainty. The Type A components are the uncertainty arising from the variability of repeated measures of the Al mass fraction and the uncertainty due to variability in the calibration factor (slope of the calibration curve). The uncertainty due to repeated measures includes the sample-to-sample and preparation-to-preparation (repeated preparation of the same sample) variability, both of which are easily detected in the graph of the Al and Li assays in Figure 3 (error bars are Type A standard uncertainties for each sample, which include measurement variability and calibration uncertainty). Because the standard uncertainties for each sample are small enough that sample-to-sample and preparation-topreparation variability is significant, the conservative approach to estimating the uncertainty treats each preparation as an indepen- F Analytical Chemistry

185 Table 6. Li Assay and Uncertainty Components mass fraction Li Assay Type A uncertainties degrees of freedom std error of replication uncertainty due to calibration Type A Type B uncertainties degrees of freedom uncertainty due to blank variability uncertainty due to calibration material purity uncertainty due to calibration material atomic weight uncertainty due to sample material atomic weight Type B combined uncertainties degrees of freedom u c expansion factor 2.12 U U rel 0.15% Table 7. Li/Al Ratio and Uncertainty Components Li/Al amt-of-substance ratio Li/Al Ratio degrees Type A uncertainties of freedom std error of replication uncertainty due to ratio calibration uncertainty due to Li calibration uncertainty due to Al calibration Type A Type B uncertainties degrees of freedom uncertainty due to assay of Al calibration material impurity uncertainty due to assay of Li calibration material uncertainty due to calibration material AW Type B degrees combined uncertainties of freedom uc expansion factor 2.04 U U rel 0.18% dent measure of the mean Al mass fraction, with the mean of the 16 samples reported with 15 degrees of freedom associated with the standard uncertainty. Uncertainty due to calibration is a measure of the variability of the calibration factor measured from the eight different calibrants, which were prepared from four primary solutions. No primary-to-primary effects were observed, so the eight calibrants are treated as independent measures of the calibration factor. The mean of these eight calibration factors is used for all calculations, and its standard uncertainty has seven degrees of freedom associated with it. Li Assay. The Li mass fraction results are detailed in Table 6. The measured Li mass fraction was ( ((0.15% relative). The expanded uncertainty for Li is dominated by the Type A components, which are the same sources of variability as described for the Al assaysreplication and calibration variability. For the Li assay, no preparation-to-preparation variability was detected, which permits detection of the sample-to-sample variability (see Figure 3). The conservative treatment is still to treat the measurements of duplicate preparations of eight samples as 16 independent measures of the Li mass fraction in the population of samples, as opposed to averaging the different preparations for each sample and summarizing these means. The mean of the 16 measures is reported with 15 degrees of freedom associated with the standard uncertainty. The Li assay has Type B components of uncertainty associated with uncertainty in the calibration material purity and, because the sample and calibration material have different atomic weights, uncertainty also associated with each atomic weight. The uncertainty in the calibration material purity is reported on the certificate for the reference material used and is divided by 2 to normalize the reported confidence interval to a standard uncertainty. The reported confidence intervals for the atomic weights are treated similarly. All of these components are insignificant with respect to the Type A components. Li/Al Stoichiometric Ratio. The Li/Al atom ratio was determined directly from the intensity ratios of Li to Al. Summary results are presented in Table 7. The results were well within the tolerance limits (between 0.92 and 1.00, 0.98 nominal) for the LiAlO 2 material. The measured ratio was ( ((0.18%). The significant components of uncertainty are the Type A measurement variability (both calibration and replication variability) and the Type B uncertainty in the Al calibration material. Determining the atom ratio directly from the Li/Al signal ratio is straightforward: the signal ratios for the samples and standards are measured, and a calibration curve is created from the standards. For this approach, this calibration curve relates the measured signal ratio (Li/Al) to the known atom ratio in the standards and is forced through zero, because the spectra are background-corrected and there is no significant blank. The atomic weights for Al and Li in the standards only are required to perform the calibration to atom ratio. The uncertainty in the Al atomic weight is insignificant, and for Li it is small compared with measurement variability. The procedure is robustsno sample preparation data (sample weights, internal standard weight, dilution factors) are required, making the procedure immune to blunders in handling and data recording. In fact, these results do not require quantitative transfer of the samples from the Carius tubes. Analytical Chemistry G

186 CONCLUSIONS The ICP-OES results presented here represent a new level of high performance when compared with the 1% uncertainties typically associated with this technique. We expect that this multielement method will transfer readily to the assay of most of the nearly 70 elements routinely measured by ICP-OES. The work described makes an important contribution to the successful implementation of the DOE commercial light water reactor manufacturing process for tritiumsin the form of a readily implemented analytical method using commercially available instrumentation and a reference material for process control. The analysis requires accurate sample handling; an accurate atomic weight of the sample; and the high-performance ICP-OES strategy, the performance of which rivals isotope-dilution TIMS and gravimetric analysis at a fraction of the cost. Additionally, this work demonstrates a systematic approach to the accurate and precise assay of materials which can be used as transfer standardsstraceability linkssto the primary standards employed. This concept is a useful tool for the practical realization of traceability in inorganic chemical metrology, which is essential to maintain long-term reference to the SI for amount-of-substance. ACKNOWLEDGMENT The authors would like to acknowledge the contributions of J. L. Mann to this work. Received for review January 31, Accepted April 29, AC H Analytical Chemistry PAGE EST: 7.3

187 Natural Remediation of Contaminants along the Forgotten River Stretch of the Rio Grande The purpose of this study was to measure the chemical and ecological gradient in the Forgotten River from Fort Quitman down to Presidio. This study included several sampling times during the year to asses the levels of metals, the chemical parameters and the impact on the benthic macroinvertebrate communities. As part of the investigation conducted under the TWRI- USGS grant that was provided, two sampling periods were performed, and a third one is planned for February The sampling times were modified from the original proposal due to the fact that seasonality could play an important role on modifying the hydrology of the river. Therefore, Figure 1. Forgotten River map showing approximate sampling sites. four sampling times per year, one in each season, for two years are planned to complete this project. The first sampling time was conducted in July 2003 and the second one in October The goal was to sample 6-10 site along the Forgotten River, but due to the inaccessibility of the roads, only 5 sites were sampled (Figure 1). These sites included one directly upstream from the cities of Presidio/Ojinaga (Site 1, GPS N , W ), one site near Candelaria (Site 2, GPS N , W ), two sites that were accessed through farms roads (Site 3 GPS N , W and Site 4 GPS N , W ), and one site at the International Boundary and Water Commission (IBWC) monitoring station at Fort Quitman (Site 5, GPS N W ). Each sampling period included measurements of water chemistry utilizing an YSI Model 85 oxygen, conductivity, salinity and temperature meter. These parameters were examined because they are relevant for metal speciation and bioavailability as well as for

188 existence of aquatic life. River physical parameters were also analyzed such as river width, depth and flow. This was done using a measuring tape and a graduated pole. The flow was approximated by allowing a floating object to flow for 10 meters and timing it, or where possible the measurements were takes from the IBWC real time measurements web page. These measurements will then be compared with other sites as well as with the same site at different sampling times. From the analysis of these parameters a correlation of metal concentration at different sites and chemical/physical parameters is expected. This will be important in determining if the presence of more water, i.e. more dilution, allows for the reduction of concentration of metals, along the gradient. Also, the parameters are going to form part of the analysis of the impact of the metals on the community. It might be possible to see that not only the presence of the metals can be disturbing the community but also the lack of water, the high salinity, or the temperature could be playing an important role. For the collection of water and sediments there were a total of two composite samples along the width of the river at two different points in every sampling site. The water was collected by grab using 500 ml plastic bottles. All of the samples were taken facing upstream to avoid contamination. After the collection, they were acidified using two milliliters of pure nitric acid to maintain the metals in solution. They were stored on ice until arrival to the laboratory where they were stored at 4 C. Before the metals were analyzed the samples were filtered using a 0.45µm Millipore filter, and then they were analyzed using a Perkin-Elmer Inductively Coupled Plasma spectrometer (Optima 4300 DV ICP-OES) (EPA 200.7). The metals that were analyzed (As, Cr, Cu, Ni, Pb, Zn, Cd, and Fe) were determined from looking at previous studies done in the river. The EPA found high levels of these metals in sites near El Paso/Juarez and downstream of the Forgotten River in sediments, but metals were also detected in water and fish tissue (EPA, 1997). Another study found high levels of lead and zinc in the El Paso/Juarez region, and the rest of the metals were also detected mainly in the sediments (Rios-Arana et al, 2003). In order to be able to determine if there is a concentration gradient due to the fact that the river flows undisturbed for 200 miles, the same metals that have been found in previous studies were analyzed. The sediments were collected using a bottom sampling dredge. They were placed in 500 ml plastic bottles, and stored on ice until arrival to the laboratory where they were kept at -20 C. To analyze the sediments following EPA method 200.7, 15 ml plastic tubes were filled with

189 sediments from each site, and then placed in a lyophilizer to freeze dry them for 48 hours. After the sediments were freeze dried, one gram per sample was microwave digested in pure nitric acid following EPA method The same metals that were analyzed in the water samples were Figure 2. Benthic macroinvertebrate sampling along the bank of the river in site 1 along the Forgotten River. also analyzed in the sediments utilizing the ICP/OAS. The collection of benthic macroinvertebrates was done according to EPA standards using a rectangular dipnet sampling along bank vegetation of the river (Figure 2). There were approximately 100 dips at each site. The macroinvertebrates were stored in glass jars containing 70% ethanol. In the laboratory, the organisms from each site were counted and identified to the family level. Benthic macroinvertebrates were used because they can be used as representatives of a particular site, in other words, drift of organisms from other areas is minimized. Analysis of biodiversity and species richness of benthic macroinvertebrates was done to determine the health of the system. The overall number of organisms was determined to show species richness at each site. For the biodiversity indices, the total number of species at each site was assessed. The numbers were predicted to be lowered as the site is more impacted, e.g. closer to El Paso/Juarez. The purpose of sampling macroinvertebrates was to determine if there is a shift in community compositions that is reflective of the metal pollution gradient. Several studies have shown that elevated levels of metals correspond with reduced species diversity and species richness (Clements et al, 2002). This part of the study will show the degree of impact of metals on community structure of macroinvertebrates following a pollution gradient. An interpretation of the results obtained from the indices, can be that the chemical/physical conditions of the river and the concentration of metals in the water and sediments are impacting the community structure of invertebrates.

190 Figure 3. Collection of fish using a seine at site 4 in the Forgotten River. The collection of fish, proved to be a much harder task than it was expected. The water levels at the different sampling sites were very low, and thus the fish that were collected were small. The sampling was done using seines, and fish of several species were collected at each site (Figure 3). The fish were euthanized with an overdose of MS-222 (1g/L) on site, and they were stored in dry ice until they were brought to the laboratory, where they were placed in a -80 C freezer. Biomarker expression, such as serum vitallogenin, hepatic metallothionein levels, and hepatic glutathione is still pending to be analyzed in these fish to assess metal exposure. Preliminary Results The chemical and physical parameters obtained from the two sampling times can be observed on table 1. From the measurements takes it can be observe that conductivity and salinity is much higher at site number 5, which is the site that is closest to El Paso. The other parameters seem to Chemical Parameters Physical Parameters July 29 & October 18 & Site 1 Site 2 Site 3 Site 4 Site 5 Site 1 Site 2 Site 3 Site 4 Site 5 Water Temperature ( C) Dissolved Oxygen (mg/l) Conductivity (us) Salinity (ppt) Flow estimate (cm/sec) N/A stagnant stagnant stagnant N/A Depth Average (cm) Width Average (m) Table 1. Chemical and physical parameters obtained from the two sampling times (July and October) at the five different sites along the Forgotten River stretch. N/A= Not available vary among the sites, but in general site 3 seems to have lower levels of the chemical

191 measurements taken. Further analysis and more sampling periods are necessary to determine if in fact a pattern or a gradient exists as you get further away from El Paso/Juarez. When total dissolved metals in the water were analyzed from the July sampling, the levels were undetectable for many metals except for Fe, which was also found at high levels during the October sampling (data not shown). When sediments were analyzed, many metals were detected as expected. Both sampling times reveal a pattern of higher concentration of metals in site 3. All the benthic macroinvertebrates were analyzed to family and the data is pending to be analyzed. Nevertheless, figure 5 shows how many families (showing diversity) were found at each site in July and October. The number of families seems to decrease with closeness to El Paso, but this needs to be further analyzed to see if in fact we find a correlation with the concentration of metals in the sediments and a decreased biodiversity Concentration (mg/kg) Pb Cr Cd Zn Cu Ni 5 0 Site 1 Site 2 Site 3 Site 4 Site 5 Sampling Sites (5 indicates close to El Paso, 1 indicates far from El Paso) Figure 3. Total recoverable metals from sediments obtained in July. Bars indicate mean ± SD.

192 Concentration (mg/l) Pb Cr Cd Zn Cu Ni Site 1 Site 2 Site 3 Site 4 Site 5 Sampling Sites (5 indicates closer to El Paso, 1 indicates far from El Paso) Figure 4. Total recoverable metals from sediments obtained in October. Bars indicate mean ± SD # of Families 10 8 July October Site 1 Site 2 Site 3 Site 4 Site 5 Sampling Site Figure 5. Total number of families of benthic macroinvertebrates present at each site in July and October.

193 The funds that were provided by this grant were used for the sampling times previously mentioned. A third sampling time will be taking place during February 20 th and 21 st using the rest of the funds. Also funds were used for supplies such as nets, bottles, nitric acid, and tubes. Thanks to this grant I was able to gather data to have a solid dissertation proposal presentation which I will defend at the end of this semester. The funds provided were essential in carrying out the activities mentioned in this report, and have left me with an incredible satisfaction and enthusiasm for what comes next; wrapping up the study and hopefully provide a baseline study to develop a restoration plan for the river which is much needed.

194 ORIGINAL ARTICLE Cardiovascular effects of Tacca integrifolia Ker-Gawl. extract in rats Nongyao Kitjaroennirut 1, Chaweewan Jansakul 2, and Prakart Sawangchote 3 Abstract Kitjaroennirut, N., Jansakul, C., and Sawangchote, P. Cardiovascular effects of Tacca integrifolia Ker-Gawl. extract in rats Songklanakarin J. Sci. Technol., 2005, 27(2) : Rhizome of Tacca integrifolia, a Thai folk medicinal herb, has been used for controlling blood pressure and improving sexual function in humans. However, the biological activities of this herb on the cardiovascular system have not yet been documented. In the present study, we investigated the cardiovascular effects of methanolic extract from the rhizome of this herb (Tacca extract). In the in vivo study, intravenous injection of the Tacca extract ( mg/kg) caused a decrease in both mean arterial blood pressure and heart rate of anesthetized rats (Nembutal sodium, 60 mg/kg, i.p.) in a dose dependent manner. Pretreatment of the animals with muscarinic receptor antagonist, atropine (1 mg/kg, i.v.), significantly reduced the hypotensive and the negative chronotropic activities of the Tacca extract. In the in vitro preparation, the Tacca extract ( mg/ml) caused a decrease in both force and rate of spontaneous contraction of isolated atria in a dose dependent manner. These effects were reduced by preincubation of the atria with atropine (10-7 or 10-6 M). For isolated blood vessels, the Tacca extract ( mg/ ml) caused vasodilation of endothelium-intact thoracic aortic rings pre-constricted with phenylephrine (3 1 M.Sc.(Physiology), 2 Ph.D.(Pharmacology), Assoc. Prof., Department of Physiology, 3 M.Sc.(Ecology), Department of Biology, Faculty of Science, Prince of Songkla University, Hat Yai, Songkhla, Thailand. Corresponding [email protected] Received, 5 July 2004 Accepted, 11 September 2004

195 Songklanakarin J. Sci. Technol. Vol.27 No.2 Mar. - Apr Cardiovascular effects of T. integrifolia extract in rats Kitjaroennirut, N., et al M). This effect disappeared after pre-incubation of blood vessels with atropine (10-6 M) or with N ω -nitro- L-arginine ( M), or by removing the vascular endothelium. The results obtained suggest that the hypotensive and negative chronotropic effects of the Tacca extract in the rat are due to the active components acting via the muscarinic receptors at the blood vessel to cause vasodilatation by stimulating the release of nitric oxide, as well as on the muscarinic receptors at the atria to cause the decrease of both rate and force of the atrial contraction. Key words : Tacca integrifolia, blood pressure, vasodilation, nitric oxide, acetylcholine receptors àõ πß «å π ÿµ å ««Ëπ ÿ ª» «à ß µ º Õß «à ππ ß «(Tacca integrifolia Ker-Gawl.) µàõà «À Õ Õ πàπÿ Á «. ß π π å « (2) : Àßâ Õß«à ππ ß «(Tacca integrifolia) ªìπ ÿπ æ π Àπ Ëß â À «ÿ «π À µ â ß æ ß æ» Õ à ß Áµ ß à ß π ß µ å æ ËÕæ Ÿ πå æ ÿ ß à «π È π» Èßπ È ß â» º µàõà «À Õ Õ Õß «à ππ ß «â«πõ» in vivo æ «à Àâ «à ππ ß «( /.) â ßÀ Õ Õ ÕßÀπŸ Á Ë â«nembutal sodium (60./., i.p.) º Àâ «π Õ Ë Õ µ µâπ ÕßÀ «ß ª µ «â âπ Õß Ë Àâ Àâ atropine (1./., i.v.) à µ«å Õß àõπ «à ππ ß «º Àâ «π Õ Õ µ µâπ ÕßÀ ««à ππ ß «πâõ ßÕ à ß π ß µ» in vitro æ «à «à ππ ß «( /.) º ÀâÕ µ «ß π À µ «â Õß Õß â π ÈÕ atrium ß ª µ «â âπ Õß ËÕ incubate â π ÈÕ atrium â«atropine (10-7 À Õ 10-6 M) àõπ Àâ º Àâ Õ µ «ß π À µ «â Õß Õß â π ÈÕ atrium πâõ ßÕ à ß π ß µ À» º Õß «à ππ ß «µàõà Õ Õ ß À à thoracic aorta æ «à «à ππ ß «( /.) º Àâ µ «ÕßÀ Õ Õ Ë ß ß å Õπ Ë Àâµ µ «Õ Ÿà àõπ â«phenylephrine ( M) µàº µ «ÕßÀ Õ Õ µàõ «à ππ ß «À ª â incubate À Õ Õ àõπ â«atropine (10-6 M) À Õ N ω -nitro-l-arginine ( M) À Õ å ÿà Õ Õ º Õß Ë â ß Àâ ÀÁπ«à º π «π À µ Õ µ µâπ ÕßÀ «Õß «à ππ ß «ªìπº π ËÕß ÕÕ ƒ Ï ßƒ Ϻà π ß muscarinic receptor ËÀ Õ Õ Àâ µ «ÕßÀ Õ Õ µÿâπ Àâ À Ëß Õß πµ ÕÕ å å ÿà Õ Õ Ë muscarinic receptor ËÀ «àߺ Àâ ÈßÕ µ «ß π À µ «ÕßÀ «ß Tacca integrifolia Ker-Gawl. belongs to the family Taccaceae, which is distributed predominately in tropical regions of Asia (Pengklai, 1993). In China, Tacca species have been used for the treatment of gastric ulcer, enteritis, and hepatitis (Dictionary of Chinese Medicinal Materials, 1977). In Thai herbal medicine, rhizomes of T. integrifolia are used for controlling blood pressure and

196 Songklanakarin J. Sci. Technol. Vol.27 No.2 Mar. - Apr Cardiovascular effects of T. integrifolia extract in rats 283 Kitjaroennirut, N., et al. improving sexual function (Chuakul et al., 2000). A numbers of Tacca spp. have been studied for their chemical constituents and their biological activities. For examples, the rhizome of T. plantaginea was found to contain several kinds of taccalonolides which are A, B, C, D, E, F, G, H, I, J, K, L, and M (Chen et al., 1987, 1988, 1989 and 1997; Shen et al., 1991 and 1996). Some other taccanolides (N, R, S, T, U, and V), were also found from root of T. paxiana (synonym T. chanterieri, Pengklai, 1993), the Vietnamese plant ( M uhlbauer et al., 2003). Among those taccanolides, the taccalonolide E and A, from the rhizome of T. chanterieri were studied for mitotic cell activities, and both have been found to cause an increased density of cellular microtubules in interphase cells and the formation of thick bundles of microtubules similar to the effects of Taxol (Tinley et al., 2003). Rhizome of T. chantrieri contains diarylheptanoids, seven diarylheptanoid glucosides, saponins, pregnane glycosides, chantriolide A and B and two withanolide glucosides (Yokosuka et al., 2002 and 2003). The diarylheptanoids and diarylheptanoid glucosides have cytotoxic activities against HL-60 human promyelocytic leukemia cells, HSC-2 human oral squamous carcinoma cells and normal human gingival fibroblasts. (Yokosuka et al., 2002). For T. integrifolia (synonym T. aspera, Pengklai, 1993), its rhizome contains ochratoxin A (Roy and Kumar, 1993), amino acids (Tiwari and Tripathi, 1980), n-triacontanol, castanogenin, betulinic acid, quercetin-3-α-arabinoside, and taccalin (Tripathi and Tiwari, 1981). However, no study has been reported on the cardiovascular activities of this plant. Thus, it is of interest to study whether the methanolic extract from T. integrifolia has any effects on blood pressure and heart rate of anesthetized rats, and which mechanisms would be involved for those activities. Materials and Methods 1. Preparation of crude methanolic extract of Tacca integrifolia Ker-Gawl. Tacca integrifolia was collected from the forest in Thepha District, Songkhla Province, southern part of Thailand. The specimen was identified and was deposited at the Prince of Songkla University Herbarium (Collecting No ). Fresh rhizomes (1.6 kg) of T. integrifolia were chopped into small pieces and were immersed in 100% methanol for three days. The clear methanol extract was collected and evaporated to dryness in vacuo, the residue was lyophilized to obtain a brown powder crude methanolic Tacca integrifolia extract (Tacca extract, 65 gm). Analysis of three ions by the inductively-coupled plasma atomic emission spectrometer (PERKIN-ELMER, Optima 4300 DV) found that the Tacca extract contained Ca , K , and Na ppm. The relative quantity of each inorganic ion in the Tacca extract powder was calculated to obtain the same amount of NaCl, KCl, and CaCl 2 to dissolve in distilled water for using as a vehicle control in the in vivo study. 2. Pharmacological studies of the methanolic Tacca extract In vivo preparation Male Wistar rats ( g) were anesthetized with Pentobarbital sodium (Nembutal sodium, 60 mg/kg, i.p.). An endotracheal tube (PE No. 50) was inserted into the rat's trachea for prevention of airway obstruction. For the actual experiment, two polyethylene catheters were used. One was cannulated through the right common carotid artery and connected to a pressure transducer (Model Stathum P23XL) and a Grass polygraph (Model 7D) for monitoring blood pressure and heart rate. Another polyethylene tube was cannulated through the left jugular vein for drug injection. Effects of Tacca extract on blood pressure and heart rate After equilibration of the animals for 40 minutes, the dose-response relationship to the Tacca extract ( mg/kg) or to the vehicle was determined. In another set of animals, after dissection and 40 minutes equilibration, the doseresponse relationship to the Tacca extract was studied after 20 minutes intravenous injection of

197 Songklanakarin J. Sci. Technol. Cardiovascular effects of T. integrifolia extract in rats Vol.27 No.2 Mar. - Apr Kitjaroennirut, N., et al. atropine (1 mg/kg). In vitro preparation The rats were decapitated with a guillotine. Both the left and the right atria were excised and mounted immediately in a 20 ml organ bath. For thoracic aorta, two adjacent rings were cut. In one ring, endothelium was removed mechanically by gently rubbing the intimal surface with a stainless steel rod, using the method of Jansakul et al., The thoracic aortic rings were placed in organ baths and attached to isometric force transducers and the signals were recorded on a polygraph. The organ bath contained Krebs-Henseleit solution of the following composition (mm): NaCl 118.3, KCl 4.7, CaCl 2 1.9, MgSO 4 7H 2 O 0.45, KH 2 PO , NaHCO , glucose 11.66, Na 2 EDTA 0.024, and ascorbic acid This solution was maintained at 37ºC and continuously bubbled with 95% O 2 and 5% CO 2. Prior to addition of drugs, tissues were equilibrated for 45 minutes under resting tension of 1.0 g for both atria and thoracic aortic rings. The Krebs solution was replaced every minutes. After equilibration, the presence of functional endothelium of the thoracic aortic ring was assessed in all preparations as follows: the thoracic aortic ring was pre-constricted with M phenylephrine until the response had plateaued (5-8 min), and dilator responses to M acetylcholine were recorded. Eighty to ninety percent vasodilatation to acetylcholine occurred with the endothelium-intact thoracic aortic rings. Effects of Tacca extract on rate and force of contraction of isolated atria After 45 min equilibration, cumulative doseresponse to the Tacca extract on the rate and force of spontaneous atrial contraction was studied. Following several washings and re-equilibration for 45 minutes, the atria were pre-incubated with atropine (10-7 or 10-6 M) for 30 minutes, after which the cumulative dose-response relationship to the Tacca extract was repeated. Effects of Tacca extract on thoracic aortic rings in vitro After 45 minutes re-equilibration, the cumulative dose-response to the Tacca extract of endothelium-intact or denuded thoracic aortic rings pre-constricted with M phenylephrine was studied. Following several washings, only the thoracic aortic rings with endothelium-intact were exposed to atropine (10-6 M) or N ω -nitro-l-arginine (L-NA), the nitric oxide synthase inhibitor (3 l0-4 M), for 40 minutes, then cumulative dilator responses to the Tacca extract of the thoracic aortic ring pre-constricted with phenylephrine were again assessed. Drugs The following drugs were used: phenylephrine chloride, N ω -nitro-l-arginine (L-NA), atropine sulphate and acetylcholine chloride, all Figure 1. Tracing recorded by a polygraph of different doses (mg/kg) of methanolic extract from Tacca integrifolia on heart rate (panel A) and on mean arterial blood pressure (panel B) in an anesthetized rat.

198 Songklanakarin J. Sci. Technol. Vol.27 No.2 Mar. - Apr Cardiovascular effects of T. integrifolia extract in rats 285 Kitjaroennirut, N., et al. obtained from Sigma, U.S.A. All drugs, and the Tacca extract were dissolved in distilled water, when used in in vivo and in vitro studies. Statistical analysis The changes in blood pressure and heart rate were recognized as the difference between the steady pressure before and the lowest pressure after injection. The blood pressure were recorded in mmhg as systolic pressure (SP) and diastolic pressure (DP) and were expressed as mean arterial pressure (MAP), which was calculated as DP + 1/ 3(SP-DP). Relaxation responses were expressed as a percentage of the induced tension which existed at the start of a relaxant concentration-effect experiment. All results are expressed as means ± SEM of 6-8 experiments (n=6-8). Student's paired or unpaired t-tests or one-way ANOVA were used for statistical analysis. In all cases, a p-value of 0.05 or less was considered statistically significant. Results Effects of Tacca extract on blood pressure and heart rate in vivo preparation The effects of the Tacca extract on arterial blood pressure and heart rate is shown in Figure 1 and the changes in mean arterial blood pressure (MAP) and heart rate (H.R.) is shown in Figure 2. Basal mean arterial pressure and heart rate of anesthetized rats both of the control and the experimental groups are similar (vehicle control group, MAP=152.85±7.81 mm Hg, H.R.= ±12.65 beats/min, n = 7; experiment group, MAP =144.92±5.6 mm Hg, H.R.= 455.0±10.62 beats/ min, n = 8). The Tacca extract ( mg/kg) caused a decrease in mean arterial blood pressure and heart rate in anesthetized rats in a consistent dose-dependent manner, while the vehicle, which has the same ion-concentration as those of the Tacca extract, did not have significant effects on blood pressure or heart rate. The lowest dose of the Tacca extract (0.04 mg/kg) caused a decrease in mean blood pressure and heart rate by 51.14±2.71 mmhg and 28.75±4.18 beats/min, respectively, whereas the highest dose (40 mg/kg) caused decrease in blood pressure and heart rate by ±5.72 mmhg and 86.88±13.78 beats/min, respectively. The highest dose caused cardiac arrest after injection of the Tacca extract for 1-2 min. The incidence of cardiac arrest of the rats was 25%. The data from these animals were excluded. Figure 3 shows the effects of the Tacca extract on mean arterial blood pressure and heart rate after blocking the muscarinic receptors with atropine. Blocking the muscarinic receptors by atropine significantly decreased the lowering effect of blood pressure and heart rate by the Tacca extract. Effects of Tacca extract on rate and force of isolated atrial contraction in vitro The effects of the Tacca extract on force and rate of spontaneous contraction of isolated atria are shown in Figures 4 and 5. The Tacca extract caused a dose-dependent decrease in rate and force of atrial contraction. Pre-incubation of the atria with the muscarinic antagonist, atropine, caused a decrease in the negative chronotropic and inotropic effects of atrial contraction by the Tacca extract (Figure 5). Effects of Tacca extract on thoracic aortic rings in vitro As shown in Figure 6, Tacca extract caused vasodilatation of the endothelium-intact thoracic aortic ring pre-constricted with phenylephrine in a dose-dependent manner. This effect disappeared after pre-incubation of the endothelium-intact aortic rings with atropine, L-NA, or by removal of the vascular endothelium. Discussion The Tacca extract caused a decrease in both mean arterial blood pressure and heart rate of anesthetized rats in a dose-dependent manner. Pretreatment of the animal with atropine, the muscarinic receptor antagonist, significantly reduced the hypotensive and negative chronotropic activities of the extract. This finding suggests a possible involvement of the Tacca extract on the

199 Songklanakarin J. Sci. Technol. Cardiovascular effects of T. integrifolia extract in rats Vol.27 No.2 Mar. - Apr Kitjaroennirut, N., et al. Figure 2. Effects of methanolic extract from Tacca integrifolia on mean arterial blood pressure (A) and heart rate (B) in anesthetized rats compared to the vehicle (control). Each point represents the mean±sem of data from 6-8 experiments. *Statistically significant lower than those of control group (p-value < 0.05). Figure 3. Effects of methanolic extract from Tacca integrifolia on mean arterial blood pressure (A) and heart rate (B) in anesthetized rats before ( ) and after blocking with atropine ( ). Each point represents the mean±sem of data from 6-8 experiments. * Significantly lower than those obtained with atropine (p-value < 0.05). muscarinic receptor. The extract may act as a muscarinic agonist, causing directly a decrease of both rate and force of atrial contraction. In order to prove this possibility, the in vitro preparations were investigated. The Tacca extract reduced both force and rate of spontaneous contraction of isolated atria in a dose-dependent manner. Preincubation of the atria with atropine significantly reduced the negative inotropic and chronotropic effects on the atrial contraction of the extract.

200 Songklanakarin J. Sci. Technol. Vol.27 No.2 Mar. - Apr Cardiovascular effects of T. integrifolia extract in rats 287 Kitjaroennirut, N., et al. Figure 4. Tracing of rate (panel A) and force (panel B) of spontaneous atrial contraction in responses to cumulative doses of the methanolic extract from Tacca integrifolia (Tacca extract, mg/ml). Figure 5. Effects of methanolic extract from Tacca integrifolia on force (A) and rate (B) of spontaneous contraction of isolated atria in the absence ( ) or presence of atropine (Atro 10-7 M or 10-6 M). Each point represents the mean±sem of data from 6-7 experiments. *Significantly lower than those obtained with atropine (p-value < 0.05).

201 Songklanakarin J. Sci. Technol. Cardiovascular effects of T. integrifolia extract in rats Vol.27 No.2 Mar. - Apr Kitjaroennirut, N., et al. Figure 6. Effects of methanolic extract from Tacca integrifolia on thoracic aorta rings with endothelium-intact ( ), endothelium-denuded ( ), endothelium-intact in the presence of atropine ( ) or endothelium-intact in the presence of L-NA ( ) in vitro. Each point represents the mean±sem of data from 6-7 experiments. In 1980, Furchgott and Zawadzki demonstrated that the vasodilation of thoracic aorta (in vitro) by acetylcholine is mediated via endothelium-derived releasing factors, later known as nitric oxide (Palmer, 1987), and is attenuated after removal of the endothelial cells. Nitric oxide is produced from L-arginine by the enzyme nitric oxide synthase in the endothelial cells that line the inner surface of blood vessels. Inhibition of the activity of nitric oxide synthase causes an increase in animal blood pressure (for review see, Moncada, 1991). Therefore, the hypotensive activity of the Tacca extract may be due to the active component(s) of the Tacca extract acting via the muscarinic receptors at the blood vessels, after which nitric oxide is released. In order to prove this possibility, we performed experiments in vitro, using isolated thoracic aortic rings. The Tacca extract caused vasodilation of endotheliam-intact, but not of denuded- thoracic aortic rings pre-constricted with phenylephrine. This result suggests that the vasodilator effect of the Tacca extract is endothelium-dependent. On the other hand, pre-incubating the thoracic aortic rings with atropine, a muscarinic receptor antagonist, or with L-NA, a nitric oxide synthase inhibitor (Moncada, 1997) completely blocked the vasodilator activity of the Tacca extract. These findings suggest that the Tacca extract causes vasodilation by stimulating the release of nitric oxide from the vascular endothelium via muscarinic receptors. In conclusion, Tacca extract exerts a hypotensive and negative chronotropic effects in anesthetized rats. The mechanisms probably involve the active component(s) acting via the muscarinic receptors at the vascular endothelium and causing vasodilatation by stimulated release of nitric oxide, as well as acting via the muscarinic receptors at the atria to cause a decrease both in rate and force of spontaneous atrial contraction. Acknowledgements This study was supported mainly by to the Faculty of Science, Prince of Songkla University, Hat Yai, Thailand, and partially by the Department of Physiology, Faculty of Science. The authors would also like to thank to Dr. Alan Geater for proof-reading the manuscript.

202 Songklanakarin J. Sci. Technol. Vol.27 No.2 Mar. - Apr Cardiovascular effects of T. integrifolia extract in rats 289 Kitjaroennirut, N., et al. References Chen, Z.L., Wang B.D., and Chen M.Q Steroidal bitter principles from Tacca plantaginea structures of taccalonolide A and B. Tetrahedron Letters, 28(15): Chen, Z., Wang, B., and Shen, J Taccalonolide C and D, two pentacyclic steroids of Tacca plantaginea. Phytochemistry, 27(9): Chen, Z., Shen, J., and Gao, Y Some chemical reactions of taccalonolide A - a bitter substance from Tacca plantaginea. Heterocycles, 29(11): Chen, Z.L., Shen, J.H., Gao, Y.S., and Wichtl, M Five taccalonolides from Tacca plantaginea. Planta Medica, 63(1): Chuakul, W. et al Encyclopedia: Medicinal plants, Vol. 2. Bangkok: Department of Pharmaceutical Botany, Faculty of Pharmacy, Mahidol University, p.207. Dictionary of Chinese Medicinal Materials; Shanghai Scientific and Technological Press: Shanghai, 1977, 2: Furchgott, R.F. and Zawadzki, J.V The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature, 288: Jansakul, C., Boura, A.L.A., and King, R.G Effects of endothelial cell removal on constrictor and dilator responses of aortae of pregnant rats. J of Autonomic Pharmacology, 9: Moncada, S., Palmer, R.M.J., and Higgs, A.E Nitric oxide: physiology, pathology, and pharmacology. Pharmacological Reviews, 43(2): Moncada, S., Higgs, A., and Furchgott, R XIV. International union of pharmacology nomenclature in nitric oxide research. Pharmacological Reviews, 49(2): M uhlbauer, A. et al Five novel taccalonolides from the roots of the Vietnamese plant Tacca paxiana. Helvetica Chimica Acta, 86(6): Palmer, R.M., Ferrige, A.G., and Moncada, S Nitric oxide release accounts for the biological activity of endothelium-derived relaxing factor. Nature, 327(11): Pengklai, C Taccaceae. In Flora of Thailand, 6(1): 1-9. Roy, A.K. and Kumar, S Occurrence of ochratoxin A in herbal drugs of Indian origin - a report. Mycotoxin Research, 9(2): Shen, J., Chen, Z., and Gao, Y The pentacyclic steroidal constituents of Tacca plantaginea: taccalonolide E and F. J of Chemistry, 9(1): Shen, J., Chen, Z., and Gao, Y Taccalonolides from Tacca plantaginea. Phytochemistry, 42(3): Tinley, T.L. et al Taccalonolides E and A: Plantderived steroids with microtubule-stabilizing activity.cancer Research, 63(12): Tiwari, K.P. and Tripathi, R.D Amino acid content in Tacca aspera. Vijnana Parishad Anusandhan, 23(2): Tripathi, R.D. and Tiwari, K.P Phytochemical investigation of the roots of Tacca aspera. Planta Medica, 41(4): Yokosuka, A., Mimaki, Y., and Sashida, Y Spirostanol saponins from the rhizomes of Tacca chantrieri and their cytotoxic activity. Phytochemistry, 61(1): Yokosuka, A., Mimaki, Y., and Sashida, Y Steroidal and pregnane glycosides from the rhizomes of Tacca chantrieri. J of Natural Products, 65(9): Yokosuka, A., Mimaki, Y., and Sashida, Y Chantriolides A and B, two new withanolide glucosides from the rhizomes of Tacca chantrieri. J of Natural Products, 66(6):

203 Journal of Chromatography A, 928 (2001) locate/ chroma Determination of organo-zinc based fungicides in timber treatments employing gas chromatographic analysis with mass selective detection and/ or inductively coupled plasma atomic emission spectroscopy * Duncan A. Rimmer, Paul D. Johnson, Stephen D. Bradley Health and Safety Laboratory, Broad Lane, Sheffield S3 7HQ, UK Received 13 April 2001; received in revised form 24 July 2001; accepted 24 July 2001 Abstract A method for the determination of zinc octoate (zinc 2-ethylhexanoate) and acypetacs zinc in occupational hygiene samples and wood treatments formulations is described. The zinc carboxylates are liquid liquid partitioned between toluene and 1 M HCl, with the liberated acids being extracted into the toluene and zinc (chloride) into the acid. The carboxylic acids are then methylated using trimethylsilyldiazomethane methanol and the resultant methyl esters are selectively and sensitively analysed by gas chromatography with mass selective detection (GC MS). Alternatively, the zinc content of the acid extract can be analysed by inductively coupled plasma atomic emission spectroscopy (ICP AES). GC MS is the preferred method of analysis for zinc octoate, where a single analyte (methyl-2-ethylhexanoate) is produced for analysis. Because acypetacs zinc contains a complex mixture of carboxylates, quantitative GC MS analysis of the methyl esters produced is impractical and ICP AES is the preferred method for quantitation. In this case, GC MS can be used to confirm the identity of the product used. The analysis of occupational hygiene samples (cotton pads, gloves and socks as well as Tenax tubes and GF/A filters) spiked with metal carboxylates is demonstrated. Recoveries around 70 90% and reproducibilities of 5 23% (n56 8) were typically achieved for the determination of tin octoate (a surrogate for zinc octoate) at spiking levels ranging from 4 to 190 mg per sampling device. Recoveries around % and reproducibilities of 10 12% (n55 6) were typically achieved for acypetacs zinc at spiking levels ranging from 100 mg per sampling device. Reaction yields for the octoate methylation reaction were in the region of 85 87%. The method was used to monitor for occupational exposure to zinc octoate and acypetacs zinc during the application of wood treatments to fences. Crown copyright 2001 Published by Elsevier Science B.V. All rights reserved. Keywords: Acypetacs zinc; Zinc octoate; Pesticides 1. Introduction *Corresponding author. Tel.: ; fax: address: [email protected] (D.A. Rimmer). Zinc 2-ethylhexanoate (commonly called zinc octoate) and acypetacs zinc are widely used active ingredients in wood preservatives and to a lesser / 01/ $ see front matter Crown copyright 2001 Published by Elsevier Science B.V. All rights reserved. PII: S (01)

204 210 D.A. Rimmer et al. / J. Chromatogr. A 928 (2001) extent in surface biocides. They are effective against procedure for the determination of zinc octoate and fungi, and are often co-formulated with other active acypetacs zinc residues in occupational hygiene ingredients such as permethrin to broaden the range samples, employing an alternative methylation proof effective protection. Additionally, zinc octoate and cedure which had been previously applied to the other metal octoates have uses in the paint and analysis of fatty acids [19] and phenoxy acid herpolymer industries where they are utilised as pigment bicides [20,21]. The solvent extracted analytes are dispersants, paint driers and polymerisation catalysts converted to free carboxylic acids by partitioning [1 4]. with HCl, then derivatized to their methyl ester Zinc octoate is a single compound, whereas derivatives, employing trimethylsilyldiazomethane, acypetacs zinc is a metal salt of complex mixtures of then quantitatively (octoate) or qualitatively C8 to C10 linear and branched chain saturated (acypetacs) analysed by gas chromatography with carboxylic acids [5]. The branched chain acids of mass selective detection (GC MS). Acypetacs zinc acypetacs are a mixture of approximately equal parts was determined by quantifying the zinc content of by mass of: (i) acids in which the main chain is the HCl extract using inductively coupled plasma dialkyl substituted on the second carbon atom and atomic emission spectroscopy (ICP AES). GC MS (ii) acids in which the second carbon atom is either was required in order to determine the analytes in the unsubstituted or monoalkyl substituted. Both (i) and presence of the complex hydrocarbon base solvent (ii) acids may be further alkyl substituted on the which is present in the formulated products. Atomic third or higher carbon atoms. absorption spectroscopy may be a suitable alternative The Health and Safety Executive (HSE) regularly to ICP AES. carries out surveys to assess the occupational expo- Due to the non-availability of a pure standard, sure of workers and consumers to pesticides under method development (spiked recovery) experiments normal use conditions [6 9]. Exposure surveys are were carried out using tin octoate as a surrogate conducted following recommended procedures [10 reference material for zinc octoate. No such pure 12], employing a number of sampling devices, which surrogate exists for acypetacs zinc, and hence re- 2 include cotton pads (100 cm ) positioned on the covery experiments were performed using a commerclothing of the operator, cotton gloves and socks cially purchased formulation following determination worn under protective gloves and shoes, and a of its zinc content. Application of the method is pumped GF/A/ Tenax sampler (a glass fibre filter demonstrated with the analysis of samples for zinc backed up with a Tenax filled glass tube). Chemical carboxylates collected during a survey investigating analysis of the devices, following the pesticide the occupational exposure to pesticides during treatment operation, enables calculation of the der- amateur fence painting operations [6]. mal and respirable exposure of the operator to the pesticide. A recent survey, investigating operator exposure during the application of amateur use 2. Experimental pesticides, required the development of reliable methods for zinc octoate and acypetacs zinc Chemicals and solutions There are many derivatization methods available which enable the gas chromatographic determination 2-Ethylhexanoic acid, tin(ii) salt (tin octoate), 2- of acids and their salts [13]. More specifically, ethylhexanoyl chloride, methyl octanoate, trimethylseveral methods have been reported in the literature silyldiazomethane (2.0 M solution in hexane), difor the determination of metal octoates [14], chlorodimethylsilane (99%) and hydrochloric acid caprylates [15] and 2-ethyl hexanoic acid [16 18]. (37%, %), were obtained from Aldrich UK. The most popular approaches involve derivatization All solvents were Distol (pesticide residue) grade of the acid or carboxylates to their pentafluorobenzyl and were obtained from Fisher Scientific, UK. esters [15,16], methyl esters [18] or through silyla- Methyl 2-ethylhexanoate (methyl octoate) could not tion [18], prior to gas chromatographic analysis. be obtained from a commercial source and was This paper describes the development of a new synthesised from 2-ethylhexanoyl chloride, as de-

205 D.A. Rimmer et al. / J. Chromatogr. A 928 (2001) tailed later in this paper. A solution of zinc at a gloves (RS Electrical Components, UK), GF/A concentration of 1000 mg/ml in 1% w/w hydrochlo- filters (Whatmans, UK) and Tenax tubes (SKC, UK) ric acid was purchased from Aldrich UK. Ronseal were employed for all laboratory and field experi- Wood Preservative (listed as 9.1% w/ w zinc octoate) ments. and Cuprinol Garden Fence & Shed Preserver (listed Teflon bottles of varying sizes, ml, were as 10% w/ w acypetacs zinc), the products applied obtained from Fisher Scientific UK. Ten ml Reactiduring the survey, were purchased from DIY retail- Vials with teflon/ silicone backed screw caps were ers. supplied by Pierce & Warriner UK. Six calibration standards of methyl octoate were All glassware was deactivated by rinsing thorprepared gravimetrically in cyclohexane to give oughly with a 5% solution of dichlorodimethylsilane concentrations in the range mg/ ml. A stan- in cyclohexane. The glassware was then rinsed three dard spiking solution of tin octoate was gravimetri- times in cyclohexane and washed (end-capped) with cally prepared at mg/ ml in cyclohexane. All methanol. solutions were stored in the dark at 2 88C. A TurboVap solvent evaporator (Zymark, UK) was Four calibration standards of zinc and a calibration employed for sample concentration, where required. blank were prepared in 1 M HCl to give con- Two GC MSD systems were utilised in the study. centrations in the range mg/ ml. Cuprinol Both consisted of a Hewlett-Packard 5890 Series II Garden Fence & Shed Preserver [stated to contain gas chromatograph fitted with a Hewlett-Packard 10% w/w acypetacs zinc (or 85 g/l) equivalent to HP-5 MS column (cross linked 5% phenyl silicone, 2% w/ w zinc (or 17 g/ l)] was found to contain 30 m30.25 mm30.25 mm film thickness). One GC 1.83% w/ w zinc (or 15.6 g/ l) by ICP AES analysis. was interfaced to a Hewlett-Packard 5970 Series This was used as the standard spiking solution of mass selective detector, the other to a Hewlett-Packacypetacs zinc. ard 5972 Series mass selective detector. Both systems were controlled by Hewlett-Packard G1034C 2.2. Synthesis of methyl octoate MS ChemStation software. The injection (splitless) and transfer line temperatures were 250 and 2808C, respectively, and the oven temperature programme Fifty ml (1.24 mol) of methanol was added was 608C for 1 min, ramping at 208C/min to 3008C cautiously down a reflux condenser into a 250-ml and holding for 1 min. Total run time is 14 min. round bottomed flask containing 10 g (61.5 mmol) of Helium ( %) was used as the carrier gas and 2-ethylhexanoyl chloride. The mixture was refluxed electronic pressure control in constant flow mode for 1 h and the excess methanol was removed by delivered 0.98 ml/ min. Selected Ion Monitoring rotary evaporation to leave 6.74 g (69%) of methyl (SIM) data was collected between 3.5 and 5.5 min, octoate which was a clear liquid. GC MS analysis of monitoring ions with m/z587, 102 and 130, for the headspace (25 ml, collected by syringe) from methyl octoate. Scan data was obtained for the above the product showed only methyl octoate qualitative analysis of acypetacs methyl, but SIM (methyl 2-ethylhexanoate) to be present. This result analysis could be used at lower levels, monitoring was checked by performing GC MS headspace ions with m/z574, 101 and 102 which are repreanalysis of the other possible contaminants, i.e. 2- sentative of fragment ions observed for branched C8 ethylhexanoyl chloride, methanol and 2-ethyl hexaand C9 methyl esters present in acypetacs (745 noic acid. The analysis results confirmed that the 1? 1 [CHO] 3 6 2, 1015[CHO] and 1025 methyl ester product was indeed free from con- 1? [C5H10O 2] ). taminants. Computer mass spectral library searching A Perkin-Elmer Optima 3000DV ICP AES confirmed the identity of the methyl octoate product. equipped with a Perkin-Elmer AS 91 autosampler was utilised throughout the study. The ICP AES 2.3. Instrumentation and apparatus was operated in radial viewing mode with a radio frequency (RF) generator power of 1300 W and Cotton pads (Phillip Harris Medical, UK), cotton nebuliser and plasma gas flow-rates of 0.80 and 15

206 212 D.A. Rimmer et al. / J. Chromatogr. A 928 (2001) l/ min, respectively. Emission wavelengths of sorbent to give 3.63 mg spiked. The spiked samples and nm were monitored. were left for 2 h at room temperature to allow the solvent to evaporate Procedures For acypetacs zinc, 1 ml of Cuprinol Garden Fence & Shed Preserver was added to pad, glove and Fig. 1 presents schematically the process of ex- sock devices producing a spike equivalent to 78 mg traction and analysis of the organo-zinc based fun- of acypetacs zinc (i.e mg of zinc and 62.4 mg gicides. of acypetacs) per device. Additionally, controls were assessed whereby clean, empty, 30 ml Teflon bottles Spiked samples were spiked with 1 ml of sample. The spiked For tin octoate, the sampling devices were spiked samples were left for 2 h at room temperature to by syringe, with various volumes of the standard allow solvent to evaporate. GF/A filters and Tenax spiking solution. Spiking levels were: 250 ml added sorbent were not assessed for this analyte. to pads to give 36.3 mg spiked; 625 ml added to gloves to give 90.8 mg spiked; 1300 ml added to Field samples socks to give mg spiked; 25 ml added to GF/A Field samples were exposed during amateur applifilters to give 3.63 mg spiked; 25 ml added to Tenax cation of timber treatments and were collected following standard occupational exposure survey protocol [10 12]. The samples were stored frozen in individual plastic bags prior to extraction, derivatization and analysis Extraction All devices were placed into Teflon bottles and toluene was added to extract the pesticide. Pads and controls were extracted with 20 ml, gloves with 50 ml, socks with 200 ml, and GF/A and Tenax 2 ml. The bottles were sealed then placed in an ultrasonic bath for 1 h Concentration In order to improve sensitivity, some of the zinc/ tin octoate extracts were concentrated using the Zymark TurboVap sample evaporator, prior to acid partitioning. Ten ml of pad extracts and 20 ml of the glove and the sock extracts were concentrated to 1 ml. Fig. 1. Quantitative and qualitative analysis of occupational hygiene sampling devices exposed to zinc carboxylate containing timber treatments Acid partitioning Portions of the extracts, were removed from the bottles and placed into Reacti-Vials. Four ml was collected for the acypetacs zinc exposed samples and 1 ml for the concentrated zinc/tin octoate samples (2 ml for GF/A and Tenax). One ml of 1 M HCl was added and the Reacti-Vials were shaken vigorously for 2 min and allowed to separate into two layers. The acid phase was removed from the vials by Pasteur pipette prior to ICP AES analysis to de-

207 D.A. Rimmer et al. / J. Chromatogr. A 928 (2001) to GC vials for analysis of methyl ester contents by GC MS. 3. Results and discussion 3.1. Tin and zinc octoates Fig. 2 shows the selected ion chromatogram (SIC) results for the methyl octoate derivative in a glove Fig. 2. GC MSD selected ion chromatogram for methyl octoate field sample which was found to contain 176 mg of m/z 102 for a real sample (i.e. left hand glove), showing the zinc octoate. The chromatogram shows that the effectiveness of SIM in resolving the analyte from the matrix. methyl octoate peak is well resolved from interferences which are probably co-extracts from the pad termine the zinc content (acypetacs zinc only). Then, and glove sampling devices. All calibration graphs 0.5 ml of the acid extracts from the spiked samples were linear (correlation coefficients of 1.000) over had to be diluted prior to analysis, to give an the standard range i.e mg/ ml. effective (standardised) 100 ml volume relative to The results of the spiking experiments are prethe spike (i.e. dilutions of 1:20 were used for pads, sented in Table 1. The mean recoveries, around 1:8 for gloves and 1:2 for the socks). Acid extracts 70 90%, are within what are generally considered from field samples were analysed directly. acceptable limits, i.e %. The lowest result The remaining toluene layer was dried using was obtained for the GF/A filters which may be due sodium sulphate prior to derivatization. to the active sites found on these devices. This proposal is supported by the fact that the highest Derivatization recovery is obtained for the inactive Tenax sorbent. Aliquots of the dried toluene extracts (2 ml for The estimated limits of detection (33signal/ noise), pads, gloves, socks and controls, and 1 ml for Tenax which range from 5 to 460 ng/device, are adequate and GF/A) were placed into clean Reacti-Vials. for determining the levels of exposure typically Methanol (450 ml for pads, gloves, socks and experienced [6 9]. Analysis of blank samples recontrols, and 225 ml for Tenax and GF/A) was added vealed no trace of the pesticide. along with trimethylsilyldiazomethane (2.0 M), 50 The reaction yield for the derivatization step was ml for pads, gloves, socks and controls, and 25 ml for determined using tin octoate due to the non-availabil- Tenax and GF/A. The total volumes were 2.5 ml for ity of a pure standard of zinc octoate. This was found pads, gloves, socks and controls, and 1.25 ml for to be around 85 87% which, although lower than Tenax and GF/A giving a dilution factor of 1.25 for would be desired, is more than acceptable. There is all devices. The Reacti-Vials were capped and soni- no reason to expect that the reaction yield for the cated for 30 min. The solutions were then transferred derivatization of the zinc octoate would be sig- Table 1 Tin octoate recovery data (determined as methyl octoate) from spiking experiments on occupational hygiene devices Device Number of Theoretical content Mean analysis 23 SD % Recovery a Limit of detection replicates (mg/ device) result (mg/ device) 623 SD (ng/ device) Pad Glove Socks Tenax GF/A filter a All limits of detection were determined on the 5972 series MSD.

208 214 D.A. Rimmer et al. / J. Chromatogr. A 928 (2001) reaction is employed under mild conditions. This reaction would not be limited to the determination of tin and zinc octoates in occupational hygiene samples but should be applicable for many more sample types, including environmental matrices, providing the analytes can be extracted into toluene Acypetacs zinc Fig. 3 shows SIC results (m/z 74, 101 and 102) for the same acypetacs zinc extract following acid partitioning, (a) underivatized and (b) derivatized. This example clearly illustrates that derivatization is required in order to derive any qualitative information on the carboxylate species present. This could be of use, for example, in determining if a product contained zinc octoate, zinc naphthenate or acypetacs zinc. Identification of the peaks as either C7, C8 or C9 was established from the highest mass? 1 ions observed in the mass spectra ([M H ] 5m/z 143, 157 and 171, respectively) and by comparison with the retention time for methyl octanoate. Methyl octanoate being unbranched would be expected to have a longer retention time than its branched C8 methyl ester analogues and this is observed in Fig. 3c). The average chain length for acypetacs was estimated to be C8.5. Unfortunately the complex Fig. 3. Qualitative GC MS (selected ion chromatograms: m/z5 nature of the products obtained prevents quantitative 74, 101 and 102) analysis of acypetacs zinc extracts. GC MS from being performed. ICP AES was used to produce quantitative data nificantly different to that of tin octoate. The reaction on the zinc content of the acid extract. The recovery yield may explain the somewhat low overall re- results for the zinc analyses are presented in Table 2. coveries achieved in the spiking experiments (i.e. The recoveries which range from 102 to 106% are recovery result5recovery1reaction yield). There- within what are generally considered acceptable fore, recovery correction must be applied to any limits, i.e %. The estimated limits of deresults obtained using this method. tection (33signal/ noise), which range from 4 to 40 The derivatization reaction utilising trimethyl- ng/ device (assuming that sample extracts undergo a silyldiazomethane, offers significant benefits over 10-fold concentration and that a 1:1 solvent acid other reported methods in that a safe and efficient partition is used) are adequate for determining the Table 2 Zinc recovery data from acypetacs zinc spiking experiments on occupational hygiene devices Device Number of Theoretical zinc Mean analysis result: 23 SD % Recovery Limit of detection replicates content (mg/ device) zinc content (mg/ device) 623 SD (ng/ device) Control N/A Pad Glove Sock

209 D.A. Rimmer et al. / J. Chromatogr. A 928 (2001) levels of exposure typically experienced [6 9]. Anal- from the zinc results using the conversion factor ysis of blank samples revealed no trace of the described in the previous section. pesticide. All calibration graphs were linear, with a correlation coefficient of or better, over the standard range, i.e. 0 5 mg/ml. Conversion of zinc to acypetacs zinc results was 4. Conclusions performed, assuming the average acypetacs chain length to be C8.5. The acypetacs zinc in this case The method described in this paper provides a would have an average molecular mass of 365 simple route for the determination of both tin and Daltons, with a zinc content of 17.8%. This is zinc octoates and acypetacs zinc from occupational consistent with the analysis result for the bulk hygiene samples. formulation where we detected 1.83% zinc in a The acid partitioning step has been shown to be an stated content of 10% acypetacs zinc (i.e. the zinc efficient way to isolate the metal and organic acid content of zinc octoate was determined at 18.3%). components prior to analysis by two different ana- Thus zinc results were converted to acypetacs zinc lytical techniques. through a multiplication factor of Trimethylsilyldiazomethane is an effective reagent for the methylation of the extracted carboxylic acids and a safer alternative to diazomethane. Subsequent 3.3. Analysis of field samples GC MSD (SIM) analysis of the organic extract following derivatization can be used to quantify Table 3 presents the data obtained from the and/ or identify the resultant methyl esters. Good analysis of survey samples exposed to organo-zinc recoveries have been achieved reproducibly for metal compounds, employing the methods described in this octoates and the limits of detection are acceptable for paper. The exposure data derived is consistent with the residue analysis required to determine occupadata obtained for the other analytes which were tional exposure. ICP AES analysis of the acid applied during this study [6]. The acypetacs zinc extracted zinc has been shown to provide similarly values detailed in Table 3 have been determined good quantitative results, but this route would re- Table 3 Analysis results for field samples collected during survey of amateur exposure to fence painting biocides Sample type Location Results [mg/ device (unless stated)] Zinc octoate Acypetacs zinc a a Bulk (% w/ w) 9.6 (9.1) 10.1 (10) GF/A Breathing zone 5 83 Tenax Breathing zone NQ ND Pad Head 1 mg/ device 11 Pad Chest (outer) Pad b Chest/ ankle (inner) 9 11 Pad Right wrist 4.6 mg/ device 6 Pad Left thigh Pad Left ankle Pad Back Glove Right hand mg/ device Glove Left hand Sock Right foot NQ 50 Sock Left foot ND, not detected; NQ, detected but not quantified (below calibration range). a Stated content. b Different positions were used to assess clothing penetration; chest for zinc octoate, ankle for acypetacs zinc.

210 216 D.A. Rimmer et al. / J. Chromatogr. A 928 (2001) quire qualitative GC MS analysis if speciation (identification) is needed. It is expected that the method described here would be applicable to many more acid based organo-metallic species and could be extended to include a wider range of samples, such as environmental samples and bulk formulations. Acknowledgements The authors would like to thank the Health Directorate of the Health and Safety Executive for sponsoring this work. References [1] P. Kirk-Othmer, 4th ed., Encyclopedia of Chemical Technology, Vol. 25, Wiley Interscience, New York, 1998, p [2] P. Kirk-Othmer, 4th ed., Encyclopedia of Chemical Technology, Vol. 24, Wiley Interscience, New York, 1998, p [3] Maldeep Catalysts Pvt. Ltd., Plot No. 2101, Panoli Industrial Estate, G.I.D.C. District, Bharuch, Gujarat ( com). [4] G. Schwach, J. Coudane, R. Engel, M. Vert, Polym. Bull 32 (1994) 617. [5] Acypetacs internet reference: co.uk/ acypetacs.html [6] A. Garrod, R. Guiver, D.A. Rimmer, Ann. Occup. Hyg. 44 (2000) 421. [7] A. Garrod, M. Martinez, J. Pearson, A. Proud, D.A. Rimmer, Ann. Occup. Hyg. 43 (1999) 543. [8] A. Garrod, D.A. Rimmer, L. Robertshaw, T. Jones, Ann. Occup. Hyg. 42 (1998) 159. [9] D.M. Llewellyn, A. Brazier, R.H. Brown, J. Cocker, M.L. Evans, J. Hampton, B.P. Nutley, J. White, Ann. Occup. Hyg. 40 (1996) 499. [10] WHO, Standard Protocol, 1982, VBC/ [11] Health and Safety Executive, Methods for the Determination of Hazardous Substances, MDHS 94, HSE Books, Sudbury, [12] Health and Safety Executive, Dermal Exposure to Non- Agricultural Pesticides: Exposure Assessment Document, EH74/ 3, HSE Books, Sudbury, [13] B. Blau, P. Halket, Handbook of Derivatives for Chromatography, 2nd ed., Wiley Interscience, New York, [14] J.B. Lucas, Health Hazard Evaluation Determination, Report No. HHE , Cook Paint and Varnish Company, North Kansas City, Missouri, Health Evaluation Services Branch, NIOSH, [15] L. Ying-Che, J. Chromatogr. 361 (1986) 279. [16] S. Kroger, Analyst 114 (1989) [17] T. Gorski, T.J. Goehl, C.W. Jameson, B.J. Collins, J. Bursey, R. Moseman, J. Chromatogr. 509 (1990) 383. [18] S. Pennanen, S. Auriola, A. Manninen, H. Komulainen, J. Chromatogr. 568 (1991) 125. [19] N. Hashimoto, T. Aoyama, T. Shioiri, Chem. Pharm. Bull. 29 (1981) [20] D.A. Rimmer, P.D. Johnson, R.H. Brown, J. Chromatogr. A 755 (1996) 245. [21] P.D. Johnson, D.A. Rimmer, R.H. Brown, J. Chromatogr. A 765 (1997) 3.

211 JOURNAL OF APPLIED PHYSICS VOLUME 93, NUMBER MAY 2003 Structure and magnetic properties of rf thermally plasma synthesized Mn and Mn Zn ferrite nanoparticles S. Son, a) R. Swaminathan, and M. E. McHenry Department of Materials Science and Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania Presented on 13 November 2002 Plasma synthesis has previously been shown to be a viable route to producing nanocrystalline magnetite and Ni ferrite nanoparticles. In this work nanocrystalline powders of Mn and Mn Zn ferrites have been synthesized using a 50 kw 3 MHz rf radio frequency induction plasma torch. We investigate these materials for soft magnetic applications. High-energy ball milled Mn Fe powders and Mn Zn Fe powders 10 m in the stoichiometric ratio of 1:2 were used as precursors for the ferrite synthesis. Compressed air was used in the oxygen source for oxidation of metal species in the plasma. X-ray diffraction patterns for the plasma-torched Mn ferrite and MnZn ferrite powders were indexed to the spinel ferrite crystal structure. An average grain size of 20 nm was determined from Scherrer analysis confirmed by transmission electron microscopy studies. The particles also exhibited faceted polygonal growth forms with the associated truncated cuboctahedral shapes. Room-temperature vibrating sample magnetometer measurements of the hysteretic response revealed saturation magnetization M s and coercivity H c of Mn ferrite are emu/g and 20 Oe, respectively. The Néel temperatures of Mn ferrite powders before and after annealing 500 C, 30 min were determined to be 200 and 360 C, respectively. Inductively coupled plasma chemical analysis and energy dispersive x-ray analysis data on the plasma-torched powders indicated deviations in the Mn or Zn contents than the ideal stoichiometry. MnZn ferrite was observed to have anéel temperature increased by almost 400 C as compared with as-synthesized Mn ferrite but with a larger coercivity of 35 Oe American Institute of Physics. DOI: / In recent years, as electronic devices have become more miniaturized, offering increasingly high levels of performance, the trend has been towards much higher-signal frequencies. Ferrites have been of great importance in highfrequency soft magnetic applications owing to their large resistivities, low power losses, and high permeabilities. Mn ferrites and MnZn ferrites have especially come to light as essential materials for higher-frequency up to MHz applications such as inductors, transformers, and yoke coils. 1 3 Mn ferrites and MnZn ferrites possessing the cubic spinel structure are described by the formula A B 2 O 4 where A and B refer to tetrahedral and octahedral cation sites, respectively, in a face-centered cubic anion oxygen sublattice. The type of cations and their distribution between the two interstitial sites in these ferrites determine many of the intrinsic magnetic properties of the ferrites. 4 Interestingly, some nanocrystalline spinel ferrites show different cation distributions and, as a consequence, diverse resulting magnetic properties when compared with their bulk counterparts. 5,6 It should be noted that the equilibrium distribution of cations in the spinel structure depends on ionic radii, electronic configuration, electrostatic energies, and polarization effects. However, using non-equilibrium processing routes, one can achieve nonequilibrium distribution of cations in the spinel ferrites. 7 a Electronic mail: [email protected] rf inductively coupled plasma synthesis has been previously shown to be a viable route to producing nanocrystalline magnetite and Ni ferrite nanoparticles. 8 In this study, our goal is to produce nanocrystalline Mn ferrites and MnZn ferrites from metallic precursors by rf plasma torch synthesis and characterize their structural and magnetic properties. Two types of precursor samples were prepared for plasma synthesis. Precursor sample 1 was a mixture of pure Mn 45 m powder and pure Fe powder 6 10 m at the atomic ratio of 1:2 i.e., that of the desired cation ratio. Sample 2 was a mixture of Mn, pure Zn dust 10 m, and Fe with overall Mn:Zn:Fe ratio 1:1:4. Both samples 1 and 2 were ball milled for two hours using a high-energy ball mill SPEX 800M Mixer/Mill, SPEX CertiPrep Inc. to ensure homogeneous mixing. An rf induction thermal plasma system consisting of a TEKNA PL50 type plasma torch head with a 50 kw, 3 MHz power supply was used for producing the nanocrystalline ferrite powders with the ball-milled metallic precursor samples. Argon gas was used as the plasma gas and the plasma sheath gas consisted of a mixture of Ar and H 2 gas. All the precursor powder was injected through the plasma jet stream by Ar, a carrier gas. After the induction plasma was established, compressed air was introduced into the reactor as an oxidizer. More detailed processing procedures can be found elsewhere. 9,10 X-ray diffraction XRD of both the precursor powders and the as-synthesized powders were performed on a Rikagu diffractometer using Cu K radiation 1.54 Å. Scanning /2003/93(10)/7495/3/$ American Institute of Physics

212 7496 J. Appl. Phys., Vol. 93, No. 10, Parts 2& 3, 15 May 2003 Son, Swaminathan and McHenry FIG. 1. X-ray diffraction patterns: a Mn Fe raw powder; b Mn Fe ball-milled powder; and c plasma-torched Mn ferrite powder. FIG. 2. X-ray diffraction patterns: a Mn Zn Fe raw powder; b Mn Zn Fe ball-milled powder; and c plasma-torched MnZn ferrite powder. electron microscopy Philips XI-30 FEG model and highresolution transmission electron microscopy TECNAI F20 model with energy dispersion x-ray EDX and electron energy-loss spectroscopy EELS were used to examine the morphology and composition of particles. The chemistry of the ferrites was further studied by inductive coupled plasma ICP chemical analysis Perkin-Elmer Optima 2000DV ICP- OES. The magnetic properties of the samples were obtained with a high-temperature vibrating sample magnetometer VSM; Lake Shore Model 7300, with 1000 C oven assembly. The formation of nanoparticles in the gas phase involves the reaction of a supersaturated metal vapor with the oxygen containing species and rapid cooling to room temperature by the cold inert gas Ar or further on the surface of watercooled chamber. Hence, this kind of processing is inherently nonequilibrium. At the extreme temperatures in the plasma jet, vaporized metal atoms react with oxygen and are cooled rapidly, forming ferrite embryos within a very short time interval. In this supercritical region, numerous nuclei or embryos formed coalesce together, forming the stable nuclei from the embryos. The growth of the particles stops when they travel out of the growth zone where there is no further supply of atoms, embryos, or nuclei. The temperatures in the growth zone are high. Finally, the particles are quenched by a cooled solid substrate that is a wall of reactor and metal filter. The temperature difference between the plasma plume and the cooled substrate has been estimated to be a few thousand Kelvin. 10 The ferrite nanoparticles are collected at room temperature. Figures 1 and 2 show the XRD patterns of samples 1 and 2, respectively. After high-energy ball milling, the diffraction peaks broadened a versus b in both figures, which indicated that the ball-milled particle size was significantly reduced as compared with that of the mixture of precursor powders. XRD peaks on plasma-torched powders revealed that most of the metallic precursor powders were transformed into nanocrystalline ferrites. All major peaks were indexed to the standard pattern for Mn ferrite and MnZn ferrite, respectively. The average particle size of as-produced ferrite powder was estimated from the Scherrer formula. 11 The 311 ferrite peak was chosen for calculating the average particle size. The Scherrer analysis showed that the average grain size of all the samples were reduced from 10 m starting powder size to 20 nm after plasma synthesis; Mn ferrite and MnZn ferrite samples have an average size of about 15 and 21 nm, respectively. TEM analysis of 346 randomly selected Mn ferrite particles shows that the average size is 9.2 nm and standard deviation is 9.2 nm. MnZn ferrite has an average size of 9.6 nm in TEM analysis of 385 randomly selected particles and its standard deviation is 10 nm. ICP analysis on both samples indicated that the powders did not result in full stoichiometric alloying after the plasma synthesis. The EDX results showed that we were able to synthesize MnZn ferrites close to the stoichiometric Mn 0.5 Zn 0.5 Fe 2 O 4, but the Mn ferrites were off stoichiometry (MnFe 2 O 4 ) Table I. TEM observations of the ferrite nanoparticles revealed the faceting with polygonal growth forms, most of them exhibiting 111 faceting with truncated cuboctahedral shapes Fig. 3. For picturing polygonal faceting in the synthesized powders, EELS is a good tool which has the capability of TABLE I. Physical properties of the as-produced ferrite particles. Mn ferrite MnZn ferrite Average particle size nm Lattice parameter Å ICP result at. % Fe and Mn a Fe 78.1, Zn 17.6, and Mn 4.3 Fe 74.1 and Mn 25.9 Fe 80, Zn 16.3, and Mn 3.7 EDX result at. % Fe 76.66, Zn 18.75, and Mn 4.59 Coercivity, H c Oe Saturation magnetization M s emu/g Néel temperature C a There is also 2.3 at. % Ni in the Mn ferrite.

213 J. Appl. Phys., Vol. 93, No. 10, Parts 2& 3, 15 May 2003 Son, Swaminathan and McHenry 7497 FIG. 5. Shapes and size distribution of a Mn ferrite particles and b MnZn ferrite particles T.O.: truncated octahedron, T.C.: truncated cube. FIG. 3. TEM images: a Mn ferrite particles; b MnZn ferrite particles; and c and d EELS thickness profile of Mn ferrite particles. getting the thickness profile of particles. EELS spectrum shows the energy loss due to both elastic and inelastic scatterings in TEM. Since the amount of all inelastic scattering increases with the specimen thickness, one can get the thickness profile of the particle by measuring the ratio of zero-loss peak intensity to total intensity in the spectrum. 12 The determination of thickness profile using EELS is shown in Fig. 3 d. For faceted particles with cubic symmetry and with exclusively 100 and 111 faces, the possibilities for particle morphologies include cube, octahedron, truncated cube, truncated octahedron, and cuboctahedron Fig. 4. In our observations, we seldom see perfect cubes or octahedra and an energetic argument to be published later shows that the perfect cuboctahedron is also rare. We thus categorized the particles by the frequency of occurrence of the truncated cubic and truncated octahedral shapes at different particle sizes. The degree of truncation describes the fraction of 100 and 111 faces in any particle Fig. 5. In both Mn ferrites and Mn Zn ferrites, in the size range of nm, we see primarily truncated octahedra, which are seen in the TEM micrographs as six-sided figures which are projections of a truncated octahedron along the 111 direction. In the size range 12 nm, in the case of Mn Zn ferrite we see more of the truncated cubes eight-sided polygons when projected along the 100 direction than the truncated octahedron and exactly the opposite in the case of Mn ferrite. This analysis is FIG. 4. Truncation of cube and octahedron: a octahedron; b truncated octahedron; c perfect cuboctahedron; d truncated cube; and d cube. important since 100 and 111 faces in the ferrite structure are terminated by different polyhedra and the degree of truncation determines the surface anisotropy. Magnetic hysteresis and magnetization versus temperature curves were measured using VSM at room temperature. Coercivities values H c and Néel temperatures of both ferrites are shown in Table I. The saturation magnetization M s and the coercivity H c of Mn ferrite are 23.6 emu/g and 20 Oe, respectively. The Néel temperature of as-produced Mn ferrite is around 200 C. It is interesting to know that after furnace annealing of the powder at 500 C for 30 min in open atmosphere, the Néel temperature became 360 C. The change of Néel temperature might be ascribed to the cation redistribution through heating process. Our results clearly demonstrate the viability of plasma torch synthesis for production of ferrite nanoparticles as did the previous report on Ni ferrite synthesis. 7 Still more research such as using alloy powder for precursor, controlling of atmosphere in the reaction chamber and thermodynamic calculation of oxide formation will be required for optimum control of the stoichiometry of the particles. The authors thank T. Nuhfer and T. Ohkubo for helping with TEM analysis. E.E. Carpenter is gratefully acknowledged for the ICP results. This work was sponsored by the Air Force Of Scientific Research, Air Force Material Command, USAF, under Grant No. F M. Sugimoto, J. Am. Ceram. Soc. 82 2, K. Ono et al., Proceedings 8th International Conference Ferrites, Kyoto and Japan, 2000, pp P. I. Slick, in Ferromagnetic Materials: A Handbook on the Properties of Magnetically Ordered Substances, edited by E. P. Wohlfarth North- Holland, Amsterdam, 1980, Vol. 2, Chap. 3, pp E. C. Snelling, Soft Ferrites: Properties and Applications Butterworths, London, 1988, p D. J. Fatemi et al., J. Appl. Phys. 85 8, C. Rath et al., J. Appl. Phys. 91 4, T. Seshagiri Rao, in Ferrite Materials Science & Technology, edited by B. Viswanathan and V. R. K. Murthy Narosa, 1990, Chap. 1, pp. 2 17, and Chap. 3, pp S. Son et al., J. Appl. Phys , Z. Turgut, Ph.D. thesis, CMU M. I. Boulos J. High Temp. Chem. Processes 1, B. D. Cullity, Elements of X-ray Diffraction, 2nd ed. Addison-Wesley, Reading, MA, 1978, pp D. B. Williams and C. B. Carter, Transmission Electron Microscopy: A Textbook for Materials Science Plenum, New York, 1996, pp

214 FIELD SCALE STUDY RESULTS FOR THE BENEFICIAL USE OF COAL ASH AS FILL MATERIAL IN SATURATED CONDITIONS, VARRA COAL ASH BURIAL PROJECT, WELD COUNTY, COLORADO Joby L. Adams and James W. Warner Colorado Groundwater Resource Service, Inc. Fort Collins, Colorado Abstract The Varra Coal Ash Project (Varra Project) is an ongoing study to determine the feasibility of using coal ash to reclaim flooded gravel mine quarries in Weld County, Colorado. The use of coal ash as fill in saturated environments is discouraged by most regulatory agencies, and there are few studies documenting the effects of coal ash in wet systems. Nearly four years were required to obtain the required permits for conducting this field scale study to assess potential impacts of large-scale coal ash reclamation on groundwater resources. The field scale study consisted of placing 400 tons of two types of Class F coal ash in a trench excavated to seven feet below the water table at the Varra gravel quarry near Longmont, Colorado. The trench was immediately adjacent to the quarry pond proposed for reclamation. The State of Colorado and Weld County approved the permits required to conduct the field study, based on column leaching studies and local surface and groundwater investigation results and safeguards implemented for the project. The trench was divided into two ash cells measuring 10 feet in width, 45 feet in length, and 11 feet in depth. A 15- foot native soil divider was left between the ash cells. Twelve groundwater monitoring wells were installed up, cross, and down gradient of the ash deposit. Two monitoring wells were installed within the trench to monitor water quality of the pore waters within the ash. Groundwater monitoring wells were placed down gradient of the coal ash trench at a spacing of 10, 25, 45, 50 (point of compliance (POC) distance), and 120 feet. Groundwater monitoring was conducted weekly for the first month, monthly to the end of the first quarter, and quarterly till project termination. Surface and groundwater samples were analyzed for 29 elements and ions such as alkalinity, chloride, fluoride, nitrate, nitrite, and sulfate. Analytical data generated from the field study indicate that the leaching characteristics of coal ash used in this study are relatively benign. The most mobile and prevalent constituent of concern appears to be boron. Molybdenum, sulfate, selenium chloride, and fluoride had elevated concentrations in water samples obtained from the ash; however, the levels associated with these elements or ions dropped to below regulatory or background levels within a month of ash placement. Drinking water standards were not exceeded at point of compliance wells 50 feet down gradient of the trench. With the exception of boron and nitrite, water quality samples obtained from the ash for the last sampling event met drinking water standards. These data indicate that large-scale ash reclamation may be feasible at this location. Experiment Surface water samples were collected from the pond proposed for reclamation (immediately adjacent to the coal ash trench) between May 1999 and August The samples were obtained to establish background water quality and assess any impacts of coal ash burial in groundwater. All samples were filtered to 0.45 microns and analyzed for eight major ions and 29 elements. Between September 2001 and February 2002, twelve soil borings were drilled with a truck mounted, hollow stem power auger. The soil borings were advanced to between 8 and 17 feet below ground surface (bgs). Soil sampling was conducted in accordance with ASTM:D Using this method representative soil samples were obtained by advancing a two-inch outside diameter split barrel sampler ahead of the auger bit. Local soil lithology descriptions were recorded during drilling activities. As the samples were obtained in the field, they were examined and described in accordance with ASTM:D In general, soil conditions consist of two to five feet of finegrained alluvial deposits of brown clay, silt, and fine-grained sands, which are underlain by six feet of gravel with sand. Bedrock, which consists of a dense, dark gray shale (Pierre Shale) underlies the sand and gravel deposits. 151

215 Figure 2 depicts soil conditions based on boring log data. All of the twelve soil borings drilled at the site were completed as two-inch diameter groundwater monitoring wells. All monitoring wells were constructed with 0.01-inch factory slotted PVC well screen with a blank PVC riser. All monitoring wells were completed within 4"x 4" above grade monuments. The depth of the wells varied between 8 and 15 feet below ground surface (bgs). The material that the water table is located in can be described as sand with gravel with the depth to water varying between 1.5 and 13 feet below grade. The Pierre Shale underlies the unconsolidated alluvial deposits and extends to beyond the depths explored during this project. The depth to the Pierre Shale within the study area ranges between 8 feet and 15 feet below ground surface. Over the entire site, the saturated thickness of the aquifer averages 5.61 feet. The groundwater flow direction within the study area is from north to south and is reflective of local topography. The hydraulic gradient varied between and ft/ft during the report period. Figure 3 depicts groundwater contours for February 2002, and Figure 3a depicts groundwater contours for May Two slug tests were utilized to determine the hydraulic conductivity of the upper portion of the aquifer and one slug test was performed in the coal ash trench. Water level measurements and times were recorded using an in situ Troll SP4000 pressure transducer. The test results will be input into standardized software, utilizing the Bouwer Rice Method, to determine hydraulic conductivity of the aquifer. Hydraulic conductivity values (K) for the aquifer varied over an order of magnitude, ranging from 20 to 203 feet/day. The hydraulic conductivity of the coal ash was calculated to be 1.22 feet per day. Using Darcy s Law the average groundwater flow rate for the aquifer was calculated to be 2.5 feet per day. The seepage velocity within the coal ash was feet per day (using an effective porosity of 0.27 and 0.40 for the aquifer and ash, respectively). On February 14 and 15, 2002 a 100-foot by 10-foot trench was excavated to 11 feet bgs using a Hitachi track driven loader. The ash was placed in two cells measuring 10 feet in width, 45 feet in length, and 11 feet in depth. A 15-foot native soil divider was left between the ash cells. Concurrent with the excavation, 200 tons of Class F coal ash with gypsum and 200 tons of coal ash with sodium were transported to the site, mixed, and placed in the excavation. Mixing was performed by use of a front-end loader, and a track driven excavator was used to compact the ash after placement. The trenches were filled to ground surface with ash, and a two-foot native soil cover was placed over the ash to preclude wind or water erosion. Water quality samples were collected from site monitoring wells on a periodic basis in order to evaluate possible changes in water quality. The sampling frequency was weekly for the first month, biweekly for the second month, monthly to the end of the first quarter, and then quarterly until project termination. All samples were filtered to 0.45 microns prior to analysis. Samples were analyzed for alkalinity as bicarbonate and carbonate, chloride, fluoride, nitrate, nitrite, and sulfate, along with other metals aluminum, antimony, arsenic, barium, beryllium, boron, calcium, cadmium, chromium, cobalt, copper, iron, lead, lithium, magnesium, manganese, mercury, molybdenum, nickel, phosphorous, potassium, selenium, silver, sodium, thallium, titanium, uranium, vanadium, and zinc. Elements were analyzed with a Perkin-Elmer Optima 2000 ICP/AES. Ions were analyzed with a Dionex LC90 ion chromatograph. Field parameters such as ph, conductivity, and temperature were measured with a HyDAC digital conductivity, ph, and temperature meter. Analytical procedures for water and soil sample collection and analysis were performed in accordance with USEPA guidelines described in SW 846 (Test Methods for Evaluating Solid Waste/Physical/Chemical Methods, 3 rd ed.). 3 Results and Discussion Surface water samples were obtained to document surface water quality of the gravel pond adjacent to the coal ash trench. Six samples were obtained between February 22 and May 13, A total of 12 samples were obtained for analyses between May 1999 and August All samples were obtained at the location depicted on Figure 1. The samples were analyzed for all constituents of concern identified for this project. Elemental analytical results are presented in Table 1. Major ions results are presented in Table 2. A review of tables 1 and 2 shows that all surface water samples (identified as pond) exceeded the standard for sulfate. Pond samples exceeded the standard for iron, lead, manganese, and nitrate for one sampling event prior to coal ash placement. The primary standard for selenium was exceeded on three sampling events. However, the elevated selenium levels are considered to be naturally occurring as sampling was conducted prior to and concurrent with coal ash placement. The anticipated travel time from the coal ash trench to the pond sample point is on the order 152

216 of 56 days, assuming no retardation or dilution effects. All of the pond samples that exceeded the selenium standard were obtained prior to the 56-day trench to pond travel time. At the time of this writing, no surface water quality issues can be associated with coal ash placement. Groundwater quality samples were collected from all monitoring wells at the same sample frequency identified for surface water samples. The samples were submitted to a contract laboratory and analyzed to determine constituents identified previously. The laboratory results are summarized in Table 1 and Table 2. A review of Table 1 and Table 2 shows that all wells not completed within the coal ash trench exceeded the standard for sulfate during every sampling event, with the levels varying between 600 and 1,961 mg/l. Nitrate levels varied between 4.8 and 36 mg/l, and every well located outside the ash trench exceeded the 10 mg/l standard. The nitrite standard was exceeded in monitoring wells MW-5 and MW-6 on sampling events one and four, respectively. With the exception of one well, all wells exceeded the standard at least once for manganese and selenium during the project. Manganese levels varied between not detected and 1.73 mg/l. Selenium varied between and mg/l. The average ph of groundwater in non-ash wells was 7.4. Monitoring wells MW-11 and MW-12 were located within the coal ash trench (ash wells). The ash wells had sulfate concentrations that closely approximated wells installed outside the trench. The ash wells also had elevated nitrate concentrations; however, with time nitrate concentrations declined and nitrite concentrations increased, with a mean concentration of 3.46 mg/l. The mean boron concentration in the ash wells was 20 mg/l with all other wells having a mean concentration of 0.69 mg/l. Significant increases in molybdenum and selenium were observed in the ash wells; however, concentrations of both elements decreased to background levels within two months of the ash placement. Chloride and fluoride concentrations were elevated in samples obtained from the ash wells with levels exceeding both standards on occasion. The mean ph of groundwater in the trench was 9.8 with a range minimum and maximum of 7.4 and 11.9, respectively. Table 3 presents selected water quality comparison of ash and nonash wells. Conclusion As expected, the coal ash trench acted as an impermeable barrier and diverted groundwater between and around the coal ash cells. The change in hydraulic head in the vicinity of the trench was on the order of 0.43 to 0.52 feet. The hydraulic conductivity of the ash was calculated to be 1.22 feet per day, which is between one and two orders of magnitude less than the permeability of the surrounding aquifer. Other than the change in hydraulic head at the trench, no other hydrogeologic effects were noted. As previously documented in the column leaching studies, the most mobile elements in the coal ashes used in this study were boron, molybdenum, and selenium. Sulfate is also mobile; however, background sulfate levels are very high and any contribution from the coal ash cannot be determined. In the column study, molybdenum and selenium levels dropped off sharply with passing pore volumes and reduction in ph, while boron levels appeared to be less affected by ph changes. To date, these trends are occurring in samples obtained from the coal ash trench as well. Analytical data document concentration spikes and reductions for the mentioned constituents that appear to be the result of geochemical processes. An unexpected occurrence, which was not observed in the column study, is the reduction of nitrate to nitrite within the ash. Nitrite was detected at elevated levels within the trench and above background levels in down gradient wells MW-5 and MW-6. A water quality graph for MW-5 (attachment) depicts boron, nitrite, and selenium. There appears to be a direct correlation between nitrite and boron concentrations. Water quality monitoring documented high (above regulatory standards) natural levels of manganese, selenium, sulfate, and nitrate. Weld County established the downstream point of compliance at 50 feet down gradient from the trench. Wells were installed downstream of the trench at 10 (MW-6), 25 (MW-5), 40 (MW-2), and 50 feet (MW-4 and MW-7). Monitoring wells MW-7 and MW-4 are points of compliance and have not been affected by the presence of coal ash within the water table. Wells MW-6 and MW-5 had apparent impacts as a result of the ash placement. Nitrite levels in samples obtained from MW-5 were in excess of the 1.0 mg/l standard in four of six sampling events. The standard for boron was exceeded on one occasion. At the time of this writing, both boron and nitrite have declining levels. The nitrite standard was exceeded in samples obtained from MW-6 on one occasion (the first sampling event). Boron levels in samples obtained from MW-6 exceeded the 5 mg/l standard during the 153

217 last sample event. The potential of using coal ash for a large-scale reclamation appears promising as field and analytical data indicate limited impacts from the coal ash placement. Concentrations for all of the elements or compounds of concern have dramatically decreased within and outside the ash deposit within a relatively short period of time. Boron, ph, and selenium levels have reduced to below regulatory limits in samples obtained from MW-12 and both ash cells have exhibited sharp reductions in elemental concentrations to date. The seepage velocity within the ash is on the order of ten feet per year, and the concentration reductions within the ash cannot be attributed to flushing or simple dilution. If this trend occurs in much larger scale deposits, then using coal ash to reclaim saturated quarries appears viable. References 1. ASTM D1586, Standard Test Method for Penetration Test and Split Barrel Sampling of Soils, Vol ASTM D2488, Description and Identification of Soils (Visual-Manual Procedure), Vol U.S. EPA, 1990, Test Methods for Evaluation Solid Waste: Physical/Chemical Methods. Joby L. Adams is co-founder and Vice President of CGRS, Inc., a national consulting firm. He has been a practicing hydrogeologist for more than 14 years. His principal areas of specialization are defining aquifer hydrogeologic properties and solving groundwater contamination problems. He has been involved with coal combustion byproduct leaching studies for more than four years and is the Principal Investigator for the Varra Coal Ash Project. He holds M.S. and B.S. degrees from Colorado State University and holds professional registrations in six States. 154

218 Digestion and Characterization of Ceramic Materials and Noble Metals S. Mann*, D. Geilenberg*, J.A.C. Broekaert*, P. Kainrath**, and D. Weber* *FB Chemie, Analytische Chemie, University of Dortmund, Otto-Hahn-Strasse 6, D Dortmund, Germany **Bodenseewerk Perkin Elmer GmbH, PO Box , D Überlingen, Germany INTRODUCTION In the synthesis and industrial production of ceramics, oxide materials, and noble metal alloys, the analytical characterization of main, minor, and trace elements is very important before their final utilization as high-purity materials or semiconductors (1). The characterization is necessary for either the determination of the exact stoichiometry or of impurities that influence the properties of the materials. The characterization of ceramics, oxide materials, and noble metal alloys was found to be difficult because of the high resistance of these materials to thermal and chemical attack, even to the attack of concentrated acids. Solid state analyses or nondestructive methods, such as X-ray fluorescence, glow discharge optical emission spectrometry, glow discharge mass spectrometry, etc., are either not sensitive enough or cannot be used because there are no reference materials available. For this reason, wet chemical digestion methods are predominantly used for routine analysis. In the case of fusion or digestion with acids in open systems, the possibility of contamination or loss of highly volatile elements, depending on the samples or the elements of interest, must be considered. Acid digestion in an open system as well as the digestion in a closed system (pressure digestion with conventional heating) is in most cases very time-consuming. In several publications (2 7) it has been shown that digestion at high temperature and high pressure in AAtomic Spectroscopy SVol. 19(2), March/April 1998 ABSTRACT This paper discusses the digestion and characterization of ceramic materials and noble metal alloys, and the advantages and limitations of several digestion methods. For the ceramic materials, three commercial pressure digestion systems were used with conventional, microwaveassisted heating, and decomposition via alkali fusion. Several noble metals and their alloys were digested in a microwave system and alternatively in a highpressure, high-temperature asher. The sample weight and especially the gas phase decomposition were optimized. The different digestion systems and some analytical results are presented. 62 closed vessels is the best method for the complete mineralization of inorganic as well as organic compounds. Temperatures of o C, which are necessary for the complete mineralization, can only be reached at pressures of bar in closed vessels. Microwave-assisted pressure digestion is an efficient alternative to the conventional, heated decompositions. INSTRUMENTATION Microwave Digestion The concept of the Perkin-Elmer Multiwave microwave digestion system (Perkin-Elmer, Norwalk, CT USA) is described in detail by Kainrath et al. (8). Figure 1 (left section) shows the power temperature curve obtained from the decomposition of the ceramic samples. Fig. 1. Microwave power and temperature for a digestion procedure of a ceramic material.

219 A tomic S pectroscopy Vol. 19(2), Mar./Apr It can also be seen at which time the pressure limit of 35 bar or the maximum temperature (200 C as in this case) was reached in one of the vessels. The temperature is measured during the digestion procedure in all vessels and is shown in the graph of Figure 1 (right section). For this example, only three vessels were used. The vessels in positions 1 and 3 contained the sample and the vessel in position 5 contained a blank solution. In the two vessels containing the sample, nearly the same temperature and therefore the same pressure was reached. The advantage of this form of presentation and control is that differences in sample constitutions, sample weights, and the composition of the acid mixtures as well as leakage of a vessel can be observed. High Pressure Asher Digestion For some very specific applications, e.g., the platinum group elements and their alloys, the temperatures and digestion times reached by microwave-assisted pressure digestion are not sufficient. In these cases, the conventional heated HPA-S High Pressure Asher (Perkin-Elmer, Norwalk, CT USA) can be used. The quartz vessels of the high pressure asher, filled with sample and acid, are closed with a quartz lid, which is fixed by a quality-controlled PTFE tape wrapped tightly around the lid and the top of the vessel. The vessels are placed inside an autoclave. The autoclave is closed and filled with nitrogen. The nitrogen pressure counteracts the reaction pressure inside the quartz vessels. Especially for the platinumgroup elements and their alloys, the gas phase decomposition was developed. A small vessel with solid potassium chlorate is suspended above the sample solution within the digestion vessel which, as a result of heating, forms elemental chlorine with chloric acid. The elemental chlorine reacts with the sample to be digested. This method is described in detail by Knapp et al. (9). APPLICATIONS Digestions of several coarsely powdered ceramic materials were performed by using the Multiwave Microwave digestion system, the conventional heated HPA-S High Pressure Asher digestion system, and DAB III, the classical digestion by salt fusion (see Tables I IV). For optimization of the digestion parameters, sample weights, acid mixtures, temperatures, and time TABLE I Instrumental Parameters Power Number Material Volume (ml) Instrument (W) of vessels of vessels of vessels Multiwave TFM / 100 / Microwave (unpulsed power quartz 50 (20) Digestion System control) HPA-S 4 x quartz / / High Pressure (conventional glassy carbon 20 Asher heating) DAB III PTFE 250 (conventional heating) TABLE II Pressure Digestions with High Concentrated Acids Digestion Sample HF a HCl b HNO c 3 H 2 SO 4d Time Sample method weight (mg) (ml) (ml) (ml) (ml) BN DAB III h Multiwave min Si 3 N 4 DAB III h Multiwave min Si B N C DAB III h (10) Multiwave min ZrO 2 Multiwave min TiO 2 Multiwave min a 38%, p.a., J. T. Baker c 65%, puriss., Riedel de Haën consumption were varied. The digestion conditions for pressure digestions with concentrated acids are shown in Table II. For salt fusion, 0.5 g of an equimolar mixture of Na 2 CO 3 and K 2 CO 3 (both pro Analyse grade, Merck, Darmstadt, Germany) and 50 mg of the sample were mixed and heated in a platinum crucible with a Bunsen burner for 10 minutes. For silicon-containing samples, the cooled melt was dissolved in either 5 ml nitric acid or a mixture of 3 ml nitric acid and 2 ml hydrofluoric acid. b 37%, puriss., Riedel de Haën d 95 97%, puriss., Riedel de Haën 63

220 TABLE III Digestion of Platinum Group Elements with HPA-S System Weight Reagents Temperature program Notes Sample (mg) ( C) (min) ( C) Pt, Pd, Os ml HCl (37%) mL vessel 4 ml HNO 3 (67%) 0.7 g KClO 3 Pt - Ir ml HCl (37%) mL vessel 4 ml HNO 3 (67%) modified 0.7 g KClO 3 heating block Rhodium ml HCl (37%) mL vessel Ruthenium 0.7 g KClO 3 Iridium ml HCl (37%) mL vessel 0.7 g KClO 3 TABLE IV Comparison of Several Digestion Methods for Nitride Ceramics Sample Digestion system Si (%) B (%) BN a DAB III (n=11) 46 ± 2 Multiwave (n=6) 46.0 ± 0.4 Salt fusion (n=9) 42.3 ± 0.8 Si 3 N b 4 DAB III (n=10) 60 ± 1 Multiwave (n=8) 59 ± 1 Salt fusion (n=8) 60 ± 2 Si-B-N-C DAB III (n=10) 31.3 ± ± 0.2 Multiwave (n=7) 33.4 ± ± 0.2 Salt fusion (n=12) 33 ± ± 0.3 a Manufacturer s specification: 42.5% B. b Manufacturer s specification: 60.06% Si (corresponding to the stoichiometry). Table III shows the conditions used for the digestion of the platinum group elements and their alloys. The advantages of the conventional heated HPA-S High Pressure Asher are the higher temperatures that can be reached and set very exactly in contrast to the microwave-assisted system, and the long digestion times that are possible. The modified heating block used for a Pt Ir 10 alloy with a smaller contact surface between heating block and digestion vessel inhibits the deposit of sample material at the rim of the fluid surface. RESULTS All materials were completely digested using the methods described. The solutions were clear and colorless. Principally, the biggest advantage of microwaveassisted pressure digestion with the Multiwave microwave system is the much shorter digestion time required in comparison to the conventional heated pressure digestion, thus resulting in a considerable time-savings. The reproducibility of the analytical results is somewhat better for digestions with the Multiwave microwave system than with the other methods. While the noble metals Au, Ag, and Pt can be easily digested by microwave-assisted pressure digestion, Ir and its alloys cannot be dissolved in microwave systems. The gas phase decomposition in the conventional, heated high pressure system is a very successful method for the digestion of these compounds without contamination and losses. The HPA-S high pressure, high temperature asher can be successfully used for ceramics and noble metal alloys up to sample weights of 120 mg because of the ability to reach high temperatures and long digestion times. For higher sample weights, even this method may fail. The gas phase decomposition with elemental chlorine as the reacting compound enables the digestion of highly inert materials. CONCLUSION Sample preparation using the Multiwave microwave high pressure digestion system provides high quality digestates for inorganic materials, because of the high temperatures that can be reached in this system, corresponding to a working pressure of 35 bar in TFM vessels. Another advantage of this system is that one can judge the quality of the digestions because of the ability to measure the temperature in each vessel. The sample throughput and the low amount of acid required make this method very attractive for routine analysis. High pressure, high temperature digestion offers an alternative to effect long digestion times at high temperatures or very slow heating procedures for highly reactive samples. With this method, even samples that normally can be dissolved only with conventional fusion or with acid mixtures unsuitable for atomic spectroscopy are accessible. 64

221 A tomic S pectroscopy Vol. 19(2), Mar./Apr REFERENCES 1. J.A.C. Broekaert, T. Graule, H. Jenett, G. Tölg, and P. Tschöpel, Fresenius Z. Anal. Chem. 332, 825 (1989). 2. G. Knapp, Mikrochim. Acta 2, 445 (1991). 3. E. Sucman, M. Sucmanova, O. Celechovska, and S. Zima, Colloqium Atomspektrometrische Spurenanalytik 6, 617 (1991). 4. P. Schramel, S. Hasse, and G. Knapp, Fres. J. Anal. Chem. 326, 142 (1987). 5. G. Knapp, B. Maichin, and F. Panholzer, Colloquium Atomspektrometrische Spurenanalytik 6, 571 (1991). 6. J. Borszeki, P. Halmos, E. Gegus, and P. Karpati, Talanta 41, 1089 (1994). 7. F. Panholzer, G. Knapp, P. Kettisch, and A. Schalk, Colloquium Atomspektrometrische Spurenanalytik 6, 633 (1991). 8. P. Kainrath, P. Kettisch, A. Schalk, and M. Zischka, LaborPraxis 11, 34 (1995) 9. G. Knapp and P. Kettisch, Colloquium Analytische Atomspektrometrie 1993, H.P. Baldus, O. Wagner, and M. Jansen, Mater. Res Soc. Symp. Proc. 271, 821 (1992) 11. S. Mann, D. Geilenberg, J.A.C. Broekaert,and M. Jansen, J. Anal. At. Spectrom. 12, 975 (1997) 65