Determination of rare earth elements U and Th in environmental samples by inductively coupled plasma double focusing sectorfield mass spectrometry (ICP-SMS) Thomas Prohaska,* Stephan Hann, Christopher Latkoczy and Gerhard Stingeder Institute of Chemistry, Vienna University of Agricultural Sciences – BOKU Wien, A-1190 Vienna, Austria Received 27th August 1998, Accepted 2nd November 1998 The excellent capability of high resolution inductively coupled plasma sectorfield mass spectrometry for measurements of the rare earth elements Th and U is demonstrated by investigating diVerent materials such as geological matrices (sediments, soils), plant tissues and marine animal tissues.Appropriate digestion of the samples resulted in complex matrices, especially in the case of silicate containing samples. The elemental loss in silicate residues of plant material was found to be up to 30% and therefore required HF-containing digestion methods. The high concentration of matrix elements leads to spectral interferences, which are investigated by measuring the elements with diVerent mass resolution. High mass resolution is shown to be a prerequisite for accurate determination of Sc and Y, respectively.Furthermore, eVects of non spectral interferences are investigated and could be properly corrected for by using 115In as internal standard. Moreover, the capability of a microconcentric nebulizer in combination with a membrane desolvation unit compared to a conventional microconcentric nebulizer is discussed with respect to suppression of spectral interferences. Oxide interferences could be reduced to a negligible amount, whereas it could be observed that high salt freight leads to a blockage of the membrane. Nowadays there is an increasing demand for rare earth interferences14–16 and even since the first investigations by ICP-MS numerous interference eVects have been reported.17,18 elements (REE) in new technology applications such as, e.g., Isobaric interferences, polyatomic ions and to a small extent electronics, optoelectronics, superconductors, supermagnets, (for Sc and Y) multiple charged ions are the most important lasers, computers, rechargeable hydride batteries, artificial spectral interferences, which are overcome usually by math- diamonds, glass and ceramics.Hence, REE technological ematical correction procedures.15,16,19 Obviously, one pos- materials are increasingly produced by industry and consesibility to deal with the problem of several spectral interferences quently also released into the environment where REE traces in ICP-MS is to separate the analyte peak and interference are bio-accumulated by organisms. Their bioavailability and peak in the mass spectrum by using high mass-resolution toxic properties are currently under investigation and REE instruments.20–22 However, there are several other instrumental are known to activate or inhibit metabolism or enzyme approaches which allow us to partially overcome this problem.activity.1–3 As a consequence, REE are already part of national One possibility to reduce spectral interferences originating and EU legislation. from the solvent is the formation of dry aerosols by means of In recent years attention has been drawn to the development a membrane desolvation unit. Membrane desolvation transfers and steady introduction of analytical methods suitable for solvents (e.g., H2O, HNO3) and dissolved gases (e.g., HF, quantification of Th and U traces.In general, spectrometric HCl, BF3, SiF4) into the gaseous phase by evaporation and determinations of REE in any matrices are characterized by separates them from the dried aerosol by diVusion through a several complicating features such as low REE concentrations membrane. Thus, mainly the analyte of interest is transported in most samples relative to the instrumental detection limits, into the plasma, whereas the concentration of possibly interfer- high concentration of matrix elements, often including major ing matrix-elements in the plasma is significantly reduced.elements (e.g., Fe, Mg, Si) and other minor and trace elements, Furthermore, enhanced signal intensities are observed comwhich can result in high levels of chemical or spectral inter- pared to conventional pneumatic introduction systems. ferences on the analyte determination. In order to overcome Another way to deal with spectral interferences is to separate some of these limitations, extensive sample pretreatment pro- the interfering matrix from the analyte of interest prior to cedures have been utilized to separate and preconcentrate REE ICP-MS analysis either in a batch process or by means of prior to analysis.4–7 oZine or online coupling of column separation methods.6,23–25 For many years atomic absorption spectrometry, X-ray Besides spectral interferences, systematic errors occur due fluorescence and inductively coupled plasma atomic emission to non-spectral interferences which are often referred to as spectrometry were the principal reference techniques to deter- ‘matrix-induced signal variations’ (suppression or enhancemine REE.8 Other important techniques are instrumental ment) or ‘matrix eVects’. These matrix eVects can be caused neutron activation analysis9,10 and isotope dilution mass by irreversible processes (e.g., clogging the nebulizer, cones, spectrometry.11 The introduction of inductively coupled torch) or reversible processes occurring only during the plasma mass spectrometry in the 1980s brought a remarkable measurement of the sample (e.g., change of nebulization change in REE analysis because of its very low detection eYciency and transport, ionization in the plasma, ion extraclevels, high sensitivity, large dynamic range, limited inter- tion).26 These signal variations have to be corrected for ference eVects and simplified sample preparation together with instrumental fluctuations by (i) internal stanprocedures. 8,12,13 However, ICP-MS analysis of REE is dardization,27–29 (ii) standard addition30 or (iii) isotope dilution,31 respectively. known to be hampered by unwanted spectral and non-spectral J. Anal. At. Spectrom., 1999, 14, 1–8 1A crucial point in ICP-MS analysis of REE is to find the acid) calibration solutions ranging from 0.01–2 ng g-1. appropriate digestion method in order to avoid incomplete Calibration standards were prepared by diluting a 10 mg g-1 digestion and too complex matrices causing severe interference REE multielement standard (SPEX, Metuchen, USA). 115In problems.Digestion of plants or animal tissues in was used as internal standard element and was prepared by HNO3–H2O2, for example, is known to lead to considerable diluting a 1000 mg g-1 standard solution (Merck, Darmstadt, formation of oxide interferences in ICP-MS.32 Especially in Germany).Other internal standard solutions (101Ru, 103Rh, the case of the determination of REE in silicate containing 187Re) were also prepared by diluting a 1000 mg g-1 standard materials, sample preparation is not straightforward. solution. The accuracy of the standards were cross checked by Incomplete digestion or loss of volatile species are two prob- single element standards if available. All standards were lems that face the analyst.12,33,34 DiVerent dissolution methods prepared by weighing.in open systems, closed vessels or in fusion are employed.35 For the preparation of HPLC eluents, lithium hydroxide Conventionally, the filtered silicate residue is dissolved in a Pt monohydrate pro analysi (Fluka, Buchs, Switzerland), oxalic crucible by using HF to dissolve SiO2 by volatilizing silicon acid pro analysi (Merck), diglycollic acid 98% (Acros Organics, as SiF4. Subsequently, the remaining metal fluorides are New Jersey, USA), tetramethylammonium hydroxide pentahyconverted (i) into nitrates by HNO3 and (ii) into chlorides by drate 97% (Fluka, Buchs, Switzerland) and purified water HCl and finally recombined with the filtrate.36 Alternatively, were used.a mixture of HF and HNO3 is used to digest the silicates. For digestion double sub-boiled HNO3 (Janssen Chimica, H3BO3 is added in a second step in order to prevent partial Geel, Belgium), H2O2 30% pro analysi (Merck), single subcoprecipitation of REE with CaF2 and other fluorides37 by boiled HCl pro analysi (Merck), H3BO3 pro analysi and HF forming a (BF4)- complex.As a main disadvantage, new 48% pro analysi (Fluka) were used. elements (B, F) are added to the matrix in high concentration The following samples were investigated: sediment material levels and consequently changing matrix-properties influence (Standard reference material ‘Canadian Stream Sediment both the occurrence of spectral and non-spectral interferences STSD1, STSD2’) and soil material (Standard reference matein ICP-MS analysis.Microwave induced decomposition was rial ‘Chinese Soil GBW07403’), aquatic plant material (‘Lemna found to be most eVective for rapid dissolution both of plant minor’) and mussel tissue (‘Mytilus edulis’) (both latter matematerial38 and geological material.39 The following work rials were provided by the ‘Mermayde Institute for monitoring demonstrates the potential of using a high mass resolution of water, sediment and biota’, Beregen, Netherlands, for an instrument in combination with two diVerent introduction intercomparison study within a certification campaign of the systems for accurate measurement of REE in various matrices EU).containing HNO3, HCl, HF and H3BO3. Sample preparation Experimental All digestions were performed in a closed vessel (PTFE vessels) Instrumentation microwave unit (MLS 1200mega, Leutenkirch, Germany). For method development the system was equipped with a pressure Measurements were carried out by means of a Finnigan MAT and temperature control vessel. ELEMENT (Finnigan MAT, Bremen, Germany) high reso- Sediments and soils were digested using (i) HNO3–HCl lution inductively coupled plasma sectorfield mass specand (ii) HNO3–HCl–HF and subsequently H3BO3 as trometer (ICP-SMS).The instrument is equipped with a F--complexing agent. double focusing mass analyzer consisting of a magnetic and electric sector field of reversed Nier–Johnson geometry. The Plant materials were digested with (i) HNO3–H2O2 and (ii) system allows three predefined nominal mass resolutions of HNO3–HF using again H3BO3 as F--complexing agent. 300, 3000 and 7500 by means of preselectable slits. The actual Marine animal tissues were digested using HNO3–H2O2. mass resolutions vary depending on the optimization of param- Table 1 shows the microwave program for sediments, soils, eter settings between 300–500 for low, 3500–4500 for medium plants and animal tissues, respectively.and 7500–8500 for high mass resolution. The system is operated under standard operating conditions, which are discussed elsewhere.40 As sample introduction systems the instument Results and discussion was equipped (i) with a microconcentric nebulizer (MCN 100, Selection of isotopes and isobaric interferences Cetac Techn. Inc., Omaha, Nebraska, USA) in combination with a double-pass Scott type spray chamber cooled at 4 °C Many isotopes of the REE are known to give rise to isobaric and (ii) with a microconcentric nebulizer in combination with interferences.Nevertheless, all these elements have at least one a heated Teflon spray chamber and a membrane desolvation isotope free of isobaric interferences which is suYciently system (MCN 6000, Cetac). abundant and hence suitable for quantitative measurements All measurements were performed in class 10 000 clean room by ICP-MS.6,15,25,35,39,41 laboratory. Isotope-pairs with isobaric overlap require mass resolutions much higher than obtainable with high resolution instruments Samples, standards and reagents (from m/Dm=75 227 for 144Nd/144Sm to m/Dm=5 854 614 for 164Dy/164Er). Therefore, an isobaric overlap cannot be All chemical preparations were carried out in class 100 metal resolved.Nevertheless, it can be corrected for by measuring free clean benches. In all analytical work analytical reagent the intensity of an isotope of the interfering element, which is grade nitric acid (Janssen Chimica, Geel, Belgium) was itself free from interferences. In case of unknown matrices, the additionally cleaned by double sub-boiling distillation where isotopic pattern additionally allows one to estimate other the solvent is distilled by surface evaporation below the boiling possible spectral interferences.However, it is important to point in an ultrapure quartz apparatus (Milestone-MLS point out that even correct isotopic patterns for at least two GmbH, Leutkirch, Germany).diVerent isotopes are not an unambiguous indicator for not Water purified by (i) reverse osmosis subsequently passes having spectral interferences as shown later for Dy, e.g., where through a (ii) laboratory-reagent grade water system (F+L both investigated isotopes (161Dy and 163Dy) have a spectral GmbH, Vienna, Austria) and is finally purified by (iii) a interference of approximately the same extent. Table 2 lists the sub-boiling system (Milestone-MLS).External calibration was performed by aqueous (1% nitric isotopes selected for further quantification. 2 J. Anal. At. Spectrom., 1999, 14, 1–8Table 1 Digestion operation parameter Aquatic Sediment or Aquatic plant Sediment or soil Material plant (HF–digestion) soil (HF–digestion) Mussel tissue Sample dry weight/g 0.15 0.15 0.10 0.10 0.25 Digestion mixture 3 mL HNO3/ 3 mL HNO3/ 3 mL HNO3/ 3 mL HNO3/ 3 mL HNO3/ 0.5 mL H2O2 0.75 mL HF 2 mL HCl 2 mL HCl/ 0.5 mL H2O2 1 mL HF Final dilution weight/mL 20 20 20 20 20 Additional dilution factor None None 100 100 None Total dissolved solids (%) 0.75 1.55 0.005 0.015 1.25 Microwave digestion program [Time (Power)/min (W)] Step 1 1 (250) 1 (250) 1 (250) 1 (250) 1 (250) Step 2 2 (0) 2 (0) 2 (0) 2 (0) 2 (0) Step 3 15 (250) 15 (250) 15 (250) 20 (250) 8 (250) Step 4 5 (400) 5 (400) 5 (400) 6 (600) 5 (400) Step 5 5 (650) 5 (650) 5 (650) 5 (650) 5 (650) Step 6 2 (0) 2 (0) H3BO3 (40 g L-1) after digestion/mL None 4 None 5 None Step 7 15 (250) 15 (250) Table 2 Isotopes selected for analysis diVerent samples.Therefore, single element standard solutions were analyzed which contained the element that was expected Required to form interfering polyatomic oxide in the same concentration resolution to as found in the matrices. Since most of the oxide interferences Abundance Main Abundance separate cannot be resolved even by high mass resolution,43 correction Isotope (%) interferences (%) interference factors were determined by means of these correction solutions. 45Sc 100.00 28Si16O1H 91.99 1893 Additionally, the influence of sample introduction on oxide 89Y 100.00 178Hf++ 27.30 1348 formation was evaluated using two diVerent introduction 139La 99.90 123Sb16O 42.65 10 000 systems, the MCN100 and the MCN6000, respectively. The 140Ce 88.50 124Sn16O 5.78 57 000 oxide formation rate of both nebulizers is significantly diVerent 141Pr 100.00 (0.3% CeO/Ce for the MCN100 and 0.01% for the MCN6000). 143Nd 12.18 127I16O 99.76 13 701 145Nd 8.30 Generally, it has been observed that oxide interferences at 146Nd 17.19 130Te16O 34.41 12 177 the selected isotopes were negligible within the analyzed 147Sm 15.00 133Cs14N 99.63 22 954 samples and resulted in systematic errors<5% for the geologi- 149Sm 13.80 133Cs16O 99.76 8839 cal material using the MCN100 as nebulizer. REE determi- 135Ba14N 6.57 17 620 nation in plant material is known to pose an analytical 151Eu 47.80 135Ba16O 6.57 7828 challenge since problems arise due to high levels of Ba and 157Gd 15.65 141Pr16O 99.76 7334 158Gd 24.84 142Nd16O 27.03 7356 extremely low levels of REE .Thus, only for 151Eu the high 159Tb 100.00 143Nd16O 12.11 7709 levels of Ba (about 650 ng g-1) resulted in a significant BaO+ 161Dy 18.90 145Nd16O 8.27 8961 interference whereas a systematic error of only <5% was 163Dy 24.90 147Sm16O 15.03 8613 calculated for other isotopes, although significant oxide inter- 165Ho 100.00 149Sm16O 13.81 9049 ferences could have been expected (147Sm, 149Sm, 158Gd, 159Tb, 167Er 22.95 151Eu16O 47.66 9656 163Dy, 167Er and 174Yb).The BaO+ interference on 151Eu is 169Tm 100.00 153Eu16O 52.11 9349 171Yb 14.30 155Gd16O 14.69 9094 comprehensively discussed elsewhere15,16 and leads to a system- 172Yb 21.90 156Gd16O 20.42 8886 atic error of 45% for the investigated plant material. It had to 173Yb 16.12 157Gd16O 15.64 8941 be corrected for mathematically, although the separation of 174Yb 31.80 158Gd16O 24.81 8763 the 151Eu peak and the 135Ba16O+ peak in the mass spectrum 175Lu 97.41 159Tb16O 99.76 8524 is possible with a mass resolution of 7828, which is available 232Th 100.00 using the high resolution mode.Fig. 1 shows a high resolution 238U 99.27 spectrum (m/Dm=7500) of 151Eu in the plant matrix. It can Multiple charged ions The elements causing spectral interferences at m/2 were not found to be present at high levels in the samples under investigation. 45Sc and 89Y can be interfered with by double charged Zr2+ and Hf2+ ions requiring a mass resolution of 12 644 or 1348, respectively. A screening of the samples for Zr and Hf revealed that the the concentration of these elements and consequently the mass peaks at m/2 are negligible. However, in the investigated plant material high levels of Ba were observed resulting in a Ba3+ interference at m/3 for 45Sc.The problem is discussed in more detail later. Oxide interferences on mass 139–175 Oxide interferences by REE, well discussed in the Fig. 1 Mass spectrum (m/z=150.87–150.95) of 151Eu in high literature,15,16,42 were investigated considering the diVerent mass resolution (m/Dm=7500) of a plant material digested in HNO3–HF–H3BO3. c(Eu)=0.06 ng g-1; c(Ba)=650 mg g-1. concentration levels of the interfering parent element in the J. Anal. At. Spectrom., 1999, 14, 1–8 3be seen that a possible BaN+ interference is present on the shoulder between the BaO+ and the Eu+ peak which would require a mass resolution of 13 757.However, the low concentration of Eu in plant material makes the quantification in the high mass resolution mode problematic due to low sensitivity and high relative standard deviation (RSD) of the signal. In the case of the marine animal tissue all isotopes including 151Eu led to deviation of the accurate value of <5%. The interferences were also observed to be negligible in case of the sediment and the soil sample even if significant oxide interferences were observed for other ICP-MS instruments.41 Although several elements also form hydroxide molecules within this mass range, the levels of formation are usually 10% of the oxide formation and are evidently negligible in the present study.41 Oxide formation strongly depends on the plasma and operating conditions and both vary with the diVerent types of Fig. 2 Mass spectrum (m/z=44.94–45.03) of 45Sc in medium mass ICP mass spectrometer; it is usually subject to changing resolution (m/Dm=3000) of a 1 ng g-1 standard solution (1% nitric operating parameters. Therefore, the oxide ratio (CeO/Ce) was acid), blank solution (1% nitric acid), digested plant material measured prior to analysis to recalculate the correction factors (HNO3–HF–H3BO3—matrix) and a HNO3–HF–H3BO3—blank with respect to the changed oxide ratio which varied from 0.1 digestion solution.to 0.4% (for CeO/Ce) in case of the MCN100 operating parameters. The use of the MCN6000 significantly reduced the oxide Table 3 Possible interferences on 45Sc by the matrix elements Ba, Si, interferences to a negligible amount due to membrane desolv- O, H, N, F and B ation. Oxide formation rates of about 0.01% (CeO/Ce) were usually observed. Therefore, correction for oxide interferences Element Mass Dm (Sc) Resolution was not necessary by using the MCN6000 as the introduction system.Nevertheless, since the high salt transfer to the nebul- 45Sc+ 44.95591 135Ba3+ 44.96856 0.01265 3555 izer, especially in case of the undiluted digestion solution of 29Si16O+ 44.97141 0.01550 2901 the plant material (salt concentration #1.6%), leads to a 28Si17O+ 44.97606 0.02015 2232 blockage of the membrane after about 10–20 consecutive 28Si1H16O+ 44.97967 0.02376 1893 measurements and subsequently to an increase of the oxide 12C1H16O2+ 44.99765 0.04174 1078 ratio from 0.01% to 0.3%, a continuous check of the oxide 12C14N19F+ 45.00147 0.04556 988 ratio is necessary.Therefore, we checked the ratio (CeO/Ce) 28Si1H314N+ 45.00349 0.04758 946 16O10B19F+ 45.00625 0.05034 894 continuously during data acquisition. 1H14N216O+ 45.00888 0.05297 850 12C21H219F+ 45.01406 0.05815 774 Interferences on 45Sc 1H216O211B+ 45.01479 0.05888 765 45Sc was expected to be influenced by a significant SiOH+ 1H14N11B19F+ 45.01861 0.06270 718 12C1H314N16O+ 45.02147 0.06556 687 interference due to high levels of Si in the sediment, the soil 12C210B11B+ 45.02225 0.06634 679 and also the plant matrices.Furthermore, the use of HF as 1H316O210B+ 45.02625 0.07034 640 digestion agent leads to an increased background of Si released 1H214N10B19F+ 45.03007 0.07416 607 from the quartz torch by remaining free HF (even if the major 12C1H311B19F+ 45.03120 0.07529 598 part is complexed by H3BO3) in sediment, soil and plant 1H314N3+ 45.03270 0.07679 586 matrices.Since the required resolution is 1894 for the main 1H414N16O11B+ 45.03861 0.08270 545 12C1H410B19F+ 45.04266 0.08675 519 abundant 28Si16O1H+ interference, 2233 for 28Si17O+ and 2902 1H411B219F+ 45.04834 0.09243 487 for 29Si16O+, 45Sc was measured in medium resolution mode (m/Dm=3000). The spectrum of the plant material in Fig. 2 shows unequivocally that besides the above listed interferences additional interferences are present. The major matrix elements interfering signal is changing significantly with respect to the H, B, C, N, O, F and Si lead to numerous complex molecular sample material.This clearly proves that measurement in low ions. Table 3 shows a list of possible interferences of the prior resolution mode or by means of quadrupole based instruments listed elements considering only the major isotopes. The prodwith subsequent blank correction solely, leads to systematically ucts of all these main matrix elements are usually di- or wrong values for Sc.The presence of a variety of spectral triatomic ions. It is very diYcult, however, to determine their interferences makes a correction even more diYcult. Therefore, probability and intensities, and this can only be done experit is evident that high mass resolution is a key prerequisite for imentally under typical operating conditions.45 The influence separation of the spectral interferences from the 45Sc signal. of the triple charged Ba3+ ion at mass 44.97 was studied in When using the MCN6000 as introduction system, we could more detail for the plant material.By investigating single observe an increase of the signal intensity for 45Sc by a factor element standards it could be proved that 1 mg g-1 Ba leads of 8 and a reduction of spectral interferences even if still to signal intensities corresponding to about 0.1 ng g-1 Sc. The significantly present. The Si–OH interference was reduced Ba level in the investigated plant material was about 10-fold, whereas the Si, B, and F containing interferences were 600 ng g-1.Furthermore, the Ba3+ signal in Fig. 2 is overonly reduced by a factor of 5. The Ba3+ interference remained lapped with 29Si16O+. nearly unaVected. The results showed clearly that even the use It is noteworthy that even after correction by internal of a nebulizing system with membrane desolvation could not standardization, the interfering signal of the method-blank is reduce the interferences to a negligible amount for 45Sc and significantly lower than that originating from the digested therefore requires high mass resolution for accurate sample.Even the interference of the standard-blank is lower than for the standard itself. Moreover the intensity of the quantification. 4 J. Anal. At. Spectrom., 1999, 14, 1–8Fig. 3 Mass spectrum (m/z=88.86–89.04) of 89Y in medium mass resolution (m/Dm=3000) of a 1 ng g-1 standard solution (1% nitric acid), blank solution (1% nitric acid), digested plant material (HNO3–HF–H3BO3—matrix) and a HNO3–HF–H3BO3—blank digestion solution.The 1% nitric acid blank solution is <20 cps and therefore not visible in this spectrum. Interferences on 89Y, 141Pr, 161Dy, 163Dy and 165Ho Since significant variations of the concentration levels in the plant material digested in HNO3–HF–H3BO3 were found for 89Y, 141Pr, 161Dy, 163Dy and 165Ho in low and high mass resolution, these isotopes were investigated with respect to possible spectral interferences. Fig. 3 shows the spectra of 89Y in the HNO3–HF–H3BO3-matrix and in 1% HNO3, respectively (medium resolution, m/Dm=3000). By calculating the accurate mass of all possible element combinations (about 150 for the major isotopes, only) in a mass range from 88.90–89.03 u, most of the interfering signals could be identified as complex combinations of H, O, B, F, N, Cl and Ar. The same eVect can be observed for 141Pr, 161Dy, 163Dy and 165Ho.Again as a consequence we proved that blank correction solely leads to erroneous quantification of these elements. It is furthermore important to point out that even the measurement of diVerent isotopes does not give evidence of spectral interferences if not measured in high mass resolution, as demonstrated for the element Dy. Both Dy isotopes 161Dy and 163Dy which were chosen for quantification were found to be interfered with to the same extent. Fig. 4 Comparison of the quantified concentration levels in the When using the MCN6000 as introduction system, the above digestion solution of a plant material for 89Y, 141Pr, 161/163Dy and mentioned spectral interferences could be reduced to a negli- 165Ho measured by means of the MCN100 in low resolution (m/Dm= gible amount.Accordingly, these isotopes could be measured 300), medium resolution mode (m/Dm=3000) and by means of the nearly free of interferences in low mass resolution when using MCN6000 in low resolution (m/Dm=300), respectively.the MCN6000 as introduction system. Again, the high salt transfer to the nebulizer is blocking the membrane and leads to a deterioration of the reduction of spectral interferences. an isotope, which is (i) not present in the sample, (ii) has no Due to the high concentration levels of REE in soil and obvious spectral interferences and (iii) shows the same eVect sediment the digested samples were diluted 100-fold, reducing of signal variation as the analyte of interest.In this study the eVect of spectral interferences. Thus, ICP-MS analysis 101Ru, 103Rh, 115In and 187Re were tested regarding their revealed only negligible production of molecular ions even suitability as internal standard by investigating the matrix when using the MCN100. suppression of both the single isotopes of interest and the Fig. 4 summarizes the eVect of spectral interferences on the internal standard. For that purpose a multielement REE result of the above mentioned elements and shows clearly the solution was digested using HF–H3BO3 and a 110 ng g-1 good agreement of the results obtained with the MCN100 in internal standard solution was added 1+9 in one experiment mass resolution m/Dm=3 000 and with the MCN6000 in mass prior to digestion and in another experiment after digestion resolution m/Dm=300.to avoid systematic errors caused by elemental loss during digestion. The analysis was performed using the MCN100 as Non-spectral interferences introduction system.Since no elemental loss could be observed, the deviation of the signal intensity could be ascribed unam- Since the use of an internal standard is the most used and biguously to non-spectral interferences. Comparing the signal most attractive method for the correction of non-spectral intensities of the digested standard to those of an aqueous interferences in routine analysis, several internal standards multielement solution showed that all investigated isotopes have been investigated with respect to their applicability in were subject to the same signal suppression as 115In.For all REE analysis by ICP-MS according to previous studies.41,46,47 The method involves the addition of a specified amount of elements under investigation the signal intensities of the J. Anal. At. Spectrom., 1999, 14, 1–8 5digestion agent. Within our investigations digestion was performed both with and without HF in order to estimate elemental loss within the silicate residues.Indeed, REE-concentrations measured in the HNO3–HCl-matrix compared to the HNO3–HCl–HF–H3BO3-matrix show significant losses of the investigated elements amounting to about 50% for U, 35% for Yb and Lu, 30% for Tm and Sc, 25% for Y and Er, 15% for La, Ce, Dy, Ho and Th and less than 10% for Nd, Sm, Eu, Gd and Tb. The same investigation was performed for the plant material. When comparing the two digestion procedures for the plant material significant losses of Sc (40%), Y (30%), Ce (10%), Nd (10%), Gd (15%), Er (35%) and Th (10%) were observed within the silicate residues when using only HNO3–H2O2 as digestion agents.For the other elements Fig. 5 Signal supression of 115In in a HNO3–HF–H3BO3—matrix. the diVerence was <5%. Since it can be expected that silicates The intensity is normalized to the maximum intensity obtained for an do not come from the plant tissues directly but are subject to aqueous standard solution (infinite dilution). The dilution represents the dilution of a matrix consisting of 3 mL HNO3, 0.75 mL HF and contamination of roots by soil or of leaves by mud in case of 4 mL H3BO3 (40 g L-1) by ultrapure water.(The final dilution to aquatic plant material, it is important to trace the sources of 20 g for the measurement of the plant material corresponds to a silicate residues in plant material. Therefore, future analysis dilution of 152 .The dilution of 0 represents an infinite dilution and of plant material has to focus on the investigation of silicate corresponds to an aqueous standard solution.) residues to find out if the elemental content in silicate residues is significant for the actual plant tissues or remaining contami- digested sample were reduced to 45–55% compared to the nation and if it is possible to separate remaining silicate signal intensities in an aqueous matrix.Only 238U showed a contaminants from the plant tissue by sequential digestion. slightly less pronounced signal suppression (from about 60–65%). Therefore it is evident that 115In is an appropriate Memory eVects internal standard. 238U had to be corrected further for the less pronounced matrix suppression. Final recovery rates were An overestimation of elemental concentration can result from found to be 100±5% after internal standardization for all elements adsorbed within the system (tubing, membrane) isotopes under investigation. Ru, Rh and Re turned out to be which are washed out in subsequent samples.REE were less suitable for internal standardization, since these elements observed not to show a significant memory eVect. For the showed lower signal suppression than the REE, but these MCN100 washout times of <30 s and of 1 min were required internal standards revealed better results for 238U. As for the MCN100 and MCN6000, respectively. Only Th showed expected,48 the signal suppression clearly depends on the a significant memory eVect, which would necessitate washout matrix concentration amounting to only 10% for the diluted times of about 20 min in case of the MCN6000 when using sediment and soil digestion samples.the conventional operation parameters (see Fig. 6). To over- Fig. 5 shows as an example that the matrix supression on come this problem, a washing solution of 10% HNO3 was 115In is not linearly related to the matrix (HF–H3BO3) concen- used with significantly increased pump speed to reduce washout tration.The same eVect was found for the other isotopes of times to 4 min for Th using the MCN6000. interest. Furthermore, the nebulizer gas flow rate was observed to be the parameter, which aVects most severely the occurrence Limits of detection of non-spectral interferences and is studied in more detail Limits of detection (DL) are calculated for mg kg-1 (ppm) elsewhere.49,50 The observed constant matrix supression indidry sample considering spectral and non-spectral interferences: cates according to Vanhaecke et al.51 that space eVects in the plasma are not solely the origin of the observed signal decrease.DL=(3×s/k)×DW/(SW×1000) Moreover, nebulizer eVects caused by changing nebulizing s, standard deviation (n=25) of the digestion blank signal eYciency with respect to viscosity and salt concentration are (cps); k, slope of the blank-corrected calibration curve expected to have a strong impact on non-spectral interferences. (cps ppb-1); DW, dilution weight (=20 g); SW, sample weight The nebulizing eYciency is again strongly influenced by the (=0.20 g).nebulizer gas flow rate when using a microconcentric nebulizer. When using the MCN6000 a reproducible reversed eVect could be observed. The signal intensities were up to 30% higher for all elements in the digested matrix compared to the aqueous standard solutions. This eVect is not fully explainable, yet.In general the main problem of the MCN6000 is an irreversible non-spectral interference and arises in the analysis of samples with high salt concentrations (e.g., digested plant material: salt concentration=1.55%). The high salt concentration leads to a blocking of the membrane and deterioration of the oxide formation ratio from 0.01% to about 0.3% (comparable to the conventional MCN100) within a sequence of about 20 consecutive measurements.Also the signal intensity of the element of interest was decreasing by a factor of 5–8, leading to a signal intensity comparable to the conventional MCN100. Silicate residues Fig. 6 Washout curve of an aqueous standard containing 5 ng g-1 Sediments and soils contain a significant amount of silicates 232Th under standard operating parameter using the MCN100 and MCN6000, respectively. and therefore require a digestion procedure with HF as 6 J. Anal. At. Spectrom., 1999, 14, 1–8Table 4 Limits of detection (mg kg-1 dry weight) for the investigated isotopes using the MCN100 and the MCN6000 in low resolution mode (LR) and the MCN100 in medium resolution (MR) for diVerent digestion methods (HNO3–H2O2 and HF) Isotope MCN100 (LR) MCN100 (LR) MCN6000 (LR) MCN100 (MR) MCN100 (MR) (HNO3/H2O2) (HF) (HF) (HNO3/H2O2) (HF) 45Sc — — 9E- 03 6E-03 5E-02 89Y 7E- 05 5E-01 8E-04 2E-03 1E-03 139La 1E-04 3E-03 1E-03 2E-03 2E-03 140Ce 2E-04 2E-03 2E-03 9E-03 3E-03 141Pr 7E-05 7E-03 3E-04 8E-04 1E-03 145Nd 3E-04 3E-03 6E-04 1E-02 4E-03 146Nd 1E-04 2E-03 1E-03 1E-02 3E-03 147Sm 1E-04 3E-03 1E-03 1E-03 5E-04 149Sm 8E-05 3E-03 2E-03 4E-03 3E-03 151Eu 6E-05 7E-04 8E-05 4E-03 4E-04 157Gd 1E-04 2E-03 1E-03 5E-03 1E-03 158Gd 9E-05 7E-04 5E-04 6E-03 2E-03 159Tb 1E-05 5E-04 2E-04 1E-03 4E-04 161Dy 1E-04 6E-03 4E-04 1E-03 9E-04 163Dy 5E-05 2E-02 8E-04 3E-03 4E-04 165Ho 2E-05 5E-03 2E-04 9E-04 2E-04 167Er 5E-05 4E-03 5E-04 2E-03 2E-03 169Tm 2E-05 2E-04 6E-05 6E-04 1E-04 171Yb 6E-05 5E-04 3E-04 3E-03 1E-04 172Yb 3E-05 6E-04 2E-04 NDa ND 173Yb 3E-05 2E-03 2E-04 ND ND 174Yb 2E-05 6E-04 3E-04 2E-03 1E-03 175Lu 1E-05 2E-04 7E-05 6E-04 9E-05 232Th 9E-04 1E-03 1E-03 5E-03 3E-03 238U 2E- 05 2E-04 2E-04 4E-04 3E-04 aNot determined.The detection limits were calculated on the basis of 25 behavior of an element, this element is usually part of additional analytical investigations to determine the accuracy of the applied independent analysis (including blank digestion in diVerent vessels by microwave digestion and subsequent analysis) method before drawing conclusions of an anthropogenic impact of REE on the sample.Thus chondrite normalization can be performed on three diVerent days. Table 4 shows the LOD for diVerent matrices and diVerent an additional assessment of analytical accuracy. Fig. 7 depicts the normalized distribution of the elemental nebulizers, respectively. content of REE in STSD1 of the certified values, values of Sen Gupta and Bertrand39 and the values obtained within this Chondritic pattern study by using the MCN100 and MCN6000, respectively.Normalization of the measured REE pattern to a known The normalization shows that the certified values for standard allows a better insight into geological processes. Yb and certainly for Lu might be too high. The slope of the Additionally, anthropogenic REE-contaminations or pollution plot from La to Ce seems also to be slightly too steep.The can be identified. Chondritic meteorites are used as standards, latter may also concern values obtained by Sen Gupta and because they represent primitive solar material that may be Bertrand.39 The MCN100-values behave discretely, but part of parental earth (values for chondritic meteorites used the MCN6000-values for Sm and Eu are slightly too high, for normalization; see Table 5).52 If the chondrite-normalized which could be caused by unresolved interferences.patterns of REE-concentrations result in a smooth curve, These results indicate that for Sm and Eu systematically anthropogenic impact can be excluded, although curves may wrong values could be found within the sediment matrix. vary in their slopes. Furthermore, Ce and Eu may display Comparable results were found for the other sediment and anomalous behavior because of their diVerent oxidation states. soil samples. Also the chondritic pattern of the animal and When the analytical chemist is faced with an anomalous Table 5 Values for chondritic meteorites used for normalization Element Concentration/mg kg-1 La 0.329 Ce 0.865 Pr 0.122 Nd 0.630 Sm 0.203 Eu 0.077 Gd 0.276 Tb NDa Dy 0.343 Ho 0.076 Er 0.226 Tm ND Fig. 7 Chondrite normalized plot of results for sediment STSD1 for Yb 0.220 Lu 0.034 the MCN100 in medium resolution (m/Dm=3000) mode, the MCN6000 in low resolution mode (m/Dm=300), the certified values aNot determined in this publication.46 and literature values.36 J.Anal. At. Spectrom., 1999, 14, 1–8 74 P. Henderson, Rare Earth Element Geochemistry, Elsevier, plant tissue indicate too high values for Lu. This is proved Amsterdam, 1984. and properly corrected for by reference measurements. Eu 5 H. Haraguchi, A. Itoh, C. Kimata and H. Miwa, Analyst, 1998, showed a significant deviation from the expected smooth curve 123, 773. in the case of plant tissue for the values obtained by low 6 X. Cao, G.Zhao, M. Yin and J. Li, Analyst, 1998, 123, 1115. resolution mode and subsequent correction for BaO+. 7 B. Li, Y. Zhang and M. Yin, Analyst, 1997, 122, 543. 8 I. B. Brenner and H. E. Taylor, Crit. Rev. Anal. Chem., 1992, 23(5), 355. Conclusion 9 R. A. Schmitt, R. H. Smith, J. E. Jasch, A. W. Mosen, D. A. Olehy and J. Vasilevskis, Geochim. Cosmochim. Acta, 1963, 27, 577. Spectral interferences are mainly important when encountering 10 P. Voldet, Trends Anal. Chem., 1984, 1, 262.low concentrations of REE in undiluted complex matrices 11 C. C. Schnetzler, H. H. Thomas and J. A. Philpotts, Anal. Chem., (e.g., digested plant material ), since a part of these inter- 1967, 39, 1888. 12 K. E. Jarvis, Chem. Geol., 1988, 68, 31. ferences are reduced to a negligible amount if the solution can 13 V. Balaram, Trends Anal. Chem., 1996, 15(9), 475. be diluted 100 fold, which is easily possible for digested 14 K. Shinotsuka and M. Ebihara, Anal. Chim. Acta, 1997, 338(3), sediment and soil samples containing high levels of REE.For 237. Sc and Y even the diluted digestion solution of soil and 15 P. Dulski, Fresenius’ J. Anal. Chem., 1994, 350, 194. sediment require correction for spectral interferences as is 16 M. A. Vaughan and G. Horlick, Appl. Spectrosc., 1990, 44, 587. obvious in high resolution spectra. It is evident that spectral 17 R. S. Houk, V. A. Fassel, G. D. Flesch, H. J. Svec, A. L. Gray and C. E. Taylor, Anal. Chem., 1980, 52, 2283.interferences are very complex, depending on the matrix, and 18 A. L. Gray, Analyst, 1975, 100, 289. simple mathematical correction procedures lead inevitably to 19 J. Goossens, L. Moens and R. Dams, Talanta, 1994, 41/2, 187. erroneous results. The varying amount of matrix induced 20 U. Gießmann and U. Greb, Fresenius’ J. Anal. Chem., 1994, 350, spectral interferences in blanks, calibrants and samples, 186. respectively, makes as a consequence high mass resolution a 21 L.Moens and N. Jakubowski, Anal. Chem., 1998, 70(1), 251A. prerequisite for accurate determination of REE in a variety of 22 B. Schnetger, Fresenius’ J. Anal. Chem., 1997, 359, 468. 23 J. A. Barrat, F. Keller, J. Amosse�, R. N. Taylor, R. W. Nesbitt diVerent matrices. Therefore, also in the case of technological and T. Hirata, Geostandard Newslett., 1996, 20(1), 133. materials ( like oxides with very low REE concentration) 24 V. K. Panday, K. Hoppstock, J. S. Becker and H.J. Dietze, At. spectral interferences remain an analytical challenge in future Spectrosc., 1996, 17(2), 98. applications. 25 M. Bettinelli and S. Spezia, At. Spectrosc., 1995, 16(3), 133. The MCN6000 was observed to eliminate the matrix solvent 26 R. F. J. Dams, J. Goossens and L. Moens, Mikrochim. Acta, 1995, and volatile species very eYciently and is therefore a powerful 119, 277. 27 C. Vandecasteele, M. Nagels, H. Vanhoe and R. Dams, Anal. tool to reduce spectral interferences (includes O, H, B, N, F Chim.Acta, 1988, 211, 91. containing species) to a negligible amount without using high 28 F. Vanhaecke, H. Vanhoe, R. Dams and C. Vandecasteele, mass resolution, except for 45Sc and also partially 89Y. Talanta, 1992, 39, 737. Moreover, the signal intensity of the analyte of interest is 29 J. J. Thompson and R. S. Houk, Appl. Spectrosc., 1987, 41, 801. increased by a factor of 5–8 and the higher sensitivity leads 30 Z. Peng, H. Klinkenberg, T.Berren and W. Van Borm, Spectrochim. Acta, 1991, 46, 1051. to lower LOD. However, higher RSD values of the signal 31 K. Y. Patterson, C. Veillon, P. B. Moser-Veillon and G. F.Wallau, (about 4–6%) compared to the MCN100 (1–3% RSD) are Anal. Chim. Acta, 1992, 258, 317. observed. As a drawback, blocking of the membrane by high 32 S. Chenery, T. Williams, T. A. Elliott, P. L. Forey and salt concentration as found in undiluted digestion solutions L. Werdelin, Mikrochim. Acta, 1996, 13, 259.decreases the performance of the desolvation unit and conse- 33 R. C. Hutton and A. N. Eaton, J. Anal. At. Spectrom., 1988, quently the reduction of polyatomic interferences. This has to 3, 547. 34 M. Totland, I. Jarvis and K. E. Jarvis, Chem. Geol., 1992, 95, 35. be checked for by, e.g., determining the oxide ratio (Ce/CeO) 35 V. Balaram, S. L. Ramesh and K. V. Anjaiah, Fresenius’ J. Anal. during the measurement. Since the sensitivity is higher with Chem., 1995, 353, 176.the MCN6000, it is possible to dilute the digestion samples 36 U. R. Kunze, Grundlagen der quantitativen Analyse, Thieme, five-fold and still obtain the detection limits comparable to Stuttgart, 1990, p. 220. the MCN100. This reduces the blockage of the membrane; 37 F. Panholzer, Spezielle Aufschlußverfahren fu�r anorganische however, a thorough rinsing is recommended after 10–20 Materialien, script Moderne Aufschlußtechnik, Vienna, 1995. 38 J. S. Alvarado, T. Neal, L. L. Smith and M. D. Erickson, Anal. consecutive measurements. Since isobaric interferences cannot Chim. Acta, 1996, 322, 11. be separated even when using high mass resolution, presepara- 39 J. G. Sen Gupta and N. B. Bertrand, Talanta, 1995, 42, 1595. tion of the REE by chromatographic methods is the only 40 G. Stingeder, T. Prohaska, Ch. Latkoczy and W. W. Wenzel, solution, e.g., when undisturbed isotope ratio measurements Proceedings of the Fourth International Conference on the are the point of investigation. Biogeochemistry of Trace Elements, Berkeley, USA, 1997, p. 391. 41 S. Z. Zhang, X. Q. Shan and H. Z. Zhang, At. Spectrosc., 1997, 115In was furthermore shown to be an appropriate isotope 18(5), 140. for the correction for non-spectral interferences for the equip- 42 H. Longerich, Can. J. Spectrom., 1986, 31, 111. ment used. It is highly recommended for the accuracy of 43 S. Augagneur, B. Me�dina, J. Szpunar and R. Lobinski, R., J. Anal. analytical data to test the applicability of an internal standard At. Spectrom., 1996, 11, 713. for all elements of interest with the used ICP mass 44 A. R. Date, Y. Y. Cheung and M. E. Stuart, Spectrochim. Acta, spectrometer, especially if a new matrix is investigated. Part B, 1987, 42, 3. 45 A. R. Date and A. L. Gray, Applications of inductively coupled plasma mass spectrometry, Blackie, New York, 1989. We gratefully acknowledge Alexandra Pederzolli and Gunda 46 K. Shinotsuka and M. Ebihara, Anal. Chim. Acta, 1997, 338, 237. Ko�llensperger for their support. 47 J. J. Thompson and R. S. Houk, Appl. Spectrosc., 1987, 41, 801. 48 H. Falk, R. Geerling, B. Hattendorf, K. Krengel-Rothensee and K. P. Schmidt, Fresenius’ J. Anal. Chem., 1997, 359, 352. References 49 J. A. Olivares and R. S. Houk, Anal. Chem., 1986, 58, 20. 50 I. Rodushkin, T. Ruth and D. Klockare, J. Anal. At. Spectrom., 1 Y. Nakamura, Y. Tsumura, Y. Tonogai, T. Shibata and Y. Ito, 1998, 13, 159. Fund. Appl. Toxicol., 1997, 37(2), 106. 51 F. Vanhaecke, J. Riondato, L. Moens and R. Dams, Fresenius’ 2 S. R. Higsmith and M. R. Head, J. Biol. Chem., 1983, 258, 52. J. Anal. Chem., 1996, 355, 397. 3 H. R. Wen and R. S. Chen, in Metal Ions in Biology and Medicine, 52 K. E. Jarvis and I. Jarvis, Geostandard Newslett., 1988, 12(1), 1. ed. P. Collery, P. Bra�tter, V. N. Bra�tter, L. Khassanova and J. C. Etienne, John Libbey Eurotext, Montrouge, 1998, vol. 5, ch. 5, p. 199. Paper 8/06720A 8 J. Anal. At. Spectrom., 1999, 1