Development of a Gas Chromatography Inductively Coupled Plasma Isotope Dilution Mass Spectrometry System for Accurate Determination of Volatile Element Species. Part 1 I Selenium Speciation* Journal of Analytical Atomic Spectrometry STEFAN M. GALLUS AND KLAUS G. HEUMANNt Institute for Inorganic and Analytical Chemistry Johannes Gutenberg- University Mainz Becherweg 24 0-55099 Mainz Germany The determination of element species in the environment is often especially difficult due to their presence at low concentrations. The coupling of GC witb ICP-MS offers the advantage of transferring the total analyte into the ICP-MS instrument without any loss of analyte by nebulization. The application of IDMS results in relatively accurate results. The described GC-ICP-IDMS system consists of a gas chromatograph fitted with a capillary column for analytical separation and a diffusion cell that is used to exactly determine the mass discrimination factor for the isotope ratio measurement and to perform an element specific optimization of the plasma conditions.The construction of a relatively simple and low cost transfer line as well as the interface between GC and ICP-MS is described in detail. The applicability of the developed GC-ICP-IDMS system for the determination of volatile element species is demonstrated by the determination of selenite. Selenite is converted into a volatile piazselenol prior to determination. Selenate is determined after conversion into selenite. By applying 62% enriched 82Se selenite spike solution for the isotope dilution step the "Se 82Se and 78Se 82Se ratios respectively could be used for content calculation.Selenite (10 ng m1-I) was determined in a water sample with good agreement (within 1%) between the results obtained using the two isotope ratios. The accuracy of the results was demonstrated by the analysis of standard reference materials. The detection limit of the described method was found to be 0.02 ng ml-'. Keywords Inductively coupled plasma mass spectrometry; gas chromatographic coupling; isotope dilution; volatile element species; selenium speciation The behaviour and effect of an element e.g. in the environment essentially depends on the chemical form in which it exists. Different element species can occur due to different oxidation states or due to the co-ordination of an element to various compounds.The toxicity and volatility of many elements significantly varies from species to species; this is well known for mercury for example comparing the volatile and highly toxic dimethyl mercury with inorganic mercury compounds.' Element speciation is also an important key factor for a better understanding of the global geochemical cycle of an element.' Especially in the case of the global distribution of the trace element selenium the biogenic conversion of selenite in the ocean into the volatile dimethyl selenide is i m p ~ r t a n t . ~ ~~~ * Presented at the 1996 Winter Conference on Plasma Spectrochemistry Fort Lauderdale FL USA January 8-1 3 1996. To whom correspondence should be addressed. Owing to the great importance of element species in the environment even in trace and ultra-trace concentrations sensitive and reliable analytical methods are required.It is well known that the accuracy of analytical results is a great problem in trace element analysis. This becomes even more significant when element species have to be determined? However accu- rate analytical results are essential with respect to the judge- ment of the toxicity the bioavailability and the environmental behaviour of element species. On-line coupling of LC or GC with ICP-MS has become a powerful analytical method for element speciation in the past few years.'-' However coupling of chromatographic methods with ICP-MS does not necessarily guarantee accurate results. IDMS is a method of proven high accuracy which means that it is a technique for which the sources of systematic error are normally understood and controlled." This is the reason why IDMS is internationally accepted as a definitive meth- od." Recently Heumann and c o - ~ o r k e r s ~ ~ - ~ ~ have developed HPLC-ICP-IDMS methods which allow accurate determi- nation of element species in aquatic systems.Heavy metal complexes of humic acids and different inorganic and organic iodine species were determined by this combined IDMS tech- nique. Another single example of HPLC-ICP-IDMS is the analysis of trimethyl lead in rain water samples carried out by Brown et al.'' In all these cases 'real time' concentrations could be determined during the separation process without any external calibration which is normally the major problem in quantifying transient signals of chromatographic ICP-MS coupled systems.Although GC has been used in conjunction with ICP-MS ID techniques with GC-ICP-MS methods have not yet been applied. We therefore developed a GC-ICP-IDMS system for the accurate determination of volatile element species. The instrumental design of the GC-ICP-IDMS system and its fundamental principles are described in this paper. This system can be applied for the determination of all volatile element species if the corresponding spike compound is available the sample pretreatment including the ID step has been developed and the species to be determined is thermally stable under the GC conditions used. The reliability of this system is shown by the determination of selenite and selenate after specific chemical conversion of selenite into a volatile piazselenol.EXPERIMENTAL ICP-MS and GC Instrumentation The ICP-MS instrument used was a Perkin Elmer SCIEX ELAN 5000 quadrupole instrument with conventional equip- ment. The standard injector tube was replaced by a quartz Journal of Analytical Atomic Spectrometry September 1996 Vol. 11 (887-892) 887injector tube with an inner diameter of 1 mm (AHF Analysentechnik Tubingen Germany). The operating con- ditions which were generally used are summarized in Table 1. The gas chromatograph used was from Carlo Erba model 8160 equipped with an HP-1 capillary column (10 m x 0.53 mm; Hewlett Packard Waldbronn Germany) with a 2.65 pm thick film (dimethylpolysiloxane). A 1 m x 0.53 mm MXT guard column (intermediate polarity; deactivated fused-silica-coated steel capillary; Restek Bad Soden Germany) was attached between the on-column injector and the capillary column.For separation a helium (99.996% purity) gas flow of 15 to 20 ml min-' was used which did not influence the high argon gas flow rates (see Table 1) in the ICP-MS. When a toluene solution of piazselenol was injected on-column into the GC the following temperature program was applied injection at 110°C into the air pres- sure cooled injector followed by a temperature increase at 15 "C min-l to 130 "C (held for 2 min) to 200°C (held for 5 min) at 40°C min-l. Chemicals Nitric acid and hydrochloric acid (pro analysi Merck Darmstadt Germany) were purified by distillation under sub- boiling conditions in a quartz still.To obtain pure water deionized water was first purified by a two-fold distillation in a quartz still then distilled under sub-boiling conditions. 1,2-Diamino-4-trifluoromethylbenzene (Maybridge Tintagel UK) was purified as described in the literature.I6 Dimethyl selenide DMSe (99% purity; Strem Chemicals Newburyport MA USA) Toluol Lichrosolv dehydrated sodium sulfate (pro analysi; Merck) and selenium dioxide (puriss.; Fluka Neu Ulm Germany) were used without further purification. All gases used in this work were obtained from Linde Germany. Interface for Coupling of GC with ICP-MS The coupling unit is schematically represented in Fig. 1. The GC column is connected uia a six-port valve (C6WT Valco Schenkon Switzerland) to an MXT steel capillary (1.5 m x 0.53 mm).This is the same type of deactivated capillary as used for the guard column. The steel capillary is inserted through the injector tube of the plasma torch and ends about 3 cm in front of the load coil. The steel capillary column is inserted into a 1 m long transfer tube (inner diameter 1 mm) also made of steel which almost fits the outer diameter of the capillary. This transfer tube is heated directly by an electrical transformer (type 3234 D Statron Fiirstenwalde Gemany). The beginning of the transfer tube is connected via a modified Swagelok connector to a Table 1 ICP-MS operating conditions Rf generator Rf power/W Channeltron electron multiplier Dead time correction/ns Sampler and skimmer cone Torch Injector tube Argon flow rates11 min-l- Outer Intermediate Make up Oxygen flow rate11 min-' Scanning mode Dwell timelms Points per spectral peak Sweeps per reading Readings per replicate 40.68 MHz free running 1200 Galileo model 48 16B 80 Aluminium Standard Quartz 1 mm id 14.4 1.1 1.1-1.2 0.02 Peak hopping 10 1 at m/z=77 78 82 3 1 detector port of the GC which is heated to the same tempera- ture as the transfer tube.The temperature in the transfer tube is calibrated by a thermoelement against the applied electrical current. The temperature is found to be constant over the whole length of the transfer tube and could be reproducibly fixed by the electrical current. This allows adjustment of the temperature of the transfer tube during coupling experiments by the electrical current. No insulation is necessary when using this direct heating system to hold the temperature constant.However for safety reasons the transfer tube is covered by a PTFE tubing. The transfer tube ends at the connection to a Swagelok reducing union that fits into a laboratory-prepared glass fitting which is connected to the injector tube support. All connections are sealed with PTFE or graphite ferrules. For the introduction of piazselenol from the GC system into the ICP-MS the transfer tube was held at a temperature of 220 "C in the case of DMSe this temperature is 140°C. GC-ICP Coupling System for ID A schematic diagram of the complete instrumentation used for the determination of volatile element species by ID is shown in Fig. 2. An element species extracted into an organic solvent can be directly injected into the GC system by means of the on-column injector. The possible injection of extracts has the advantage that enrichment procedures by extraction and/or derivatizations of non-volatile species into volatile ones can be carried out prior to determination by GC-ICP-MS.Alternatively volatile element species can be directly intro- duced into the separation column. In this case cryo-focussing prior to sample introduction should be carried out to prevent unnecessary peak broadenings. The transfer tube leading to the ICP-MS instrument can either be switched to the GC capillary column or to a glass flow cell by the six-port valve. The flow cell with a volume of about 20 ml is laboratory prepared. It contains a diffusion cell consisting of a glass vial covered by a membrane which allows diffusion of the volatile compound into the flow cell.A 4ml screw cap vial (WISP Style Alltech Unterhaching Ger- many) with a PTFE-red butylgum septum (1 1 mm x 1.3 mm; Macherey & Nagel Diiren Germany) as membrane normally used as a container for GC samples is used as a diffusion cell. Both the capillary column and diffusion cell are heated by the GC oven. For calibration of the measured isotope ratios the diffusion cell containing pure chemicals of the element species to be determined is placed inside the flow cell. In the case of DMSe an oven temperature of 30°C is used for piazselenol the oven temperature is 85 "C. From here they are introduced in a controlled way into the helium gas flow entering the ICP-MS instrument under exactly the same conditions as the separated compounds from the capillary column.This is necessary because the actual mass discrimination factor in measuring isotope ratios with the quadrupole ICP-MS must always be determined for accurate results by ID. If the isotope ratio of the calibration compounds used in the diffusion cell are known for example by the natural isotopic composition of the elements or by determinations with other MS methods such as gas mass or TIMS which show only very low mass discrimination effects,17 the measured isotope ratio of the separated species in the sample can be corrected. Using the selenium isotope ratio determination as an example the mass discrimination factor can be calculated by the following equa- tion Mass discrimination factor ( 77Se 82Se) = ( 77Se 82Se),,ue/( 77Se 82Se),e,,u The introduction of compounds from the diffusion cell into 888 Journal of Analytical Atomic Spectrometry September 1996 Vol.11glass fitting injector tube support Swagelok heated transfer tube reducing union I I -H 3cm - 000 MXT steel capillary terminal to electrical I I transformer t u u 1 t plasma auxiliary I argon make up Fig. 1 Interface unit for coupling of GC with ICP-MS helium onalurnn injector GC oven I transformer transfer tu; Fig. 2. Schematic diagram of the GC-ICP-MS coupling system the ICP-MS is the best method for tuning the ion intensities. To reduce memory effects when different species are sub- sequently deposited in the diffusion cell the inner surface of the cell was deactivated by silylation. Silylation of the OH groups at the glass surface was carried out by reaction with dimethyl silicon dichloride as described in the literature." To complete the removal of the compounds introduced for signal optimization the GC oven was heated for 30 min at 200 "C and the transfer tube at 220°C.Isotope Ratio Determinations For isotope ratio determinations of selenium the isotopes 77Se 78Se and 82Se were selected. The 74Se isotope was not measured because of its low natural abundance. Interference was pre- sent by the dimeric molecular ions of the argon plasma gas 36Ar40Ar+ and 40Ar2+ on 76Se and 80Se respectively. Mass number 83 has always been controlled for possible corrections at mass number 82 which can be interfered by krypton impurities in the gas flows.Fig. 3 shows the results of the measurement of selenium isotope intensities at mass numbers 77 78 and 82 after separating the piazselenol compound described later by the GC system. The piazselenol compound for this measurement was of natural isotopic composition. The time resolved chromatogram demonstrates that ion intensities at mass numbers 77 and 82 can only be detected at the specific retention time of the piazselenol. In ICP-MS mass number 77 is very often interfered by 40Ar37C1+. One of the great advan- tages of chromatographic coupling methods in connection with ICP-MS is therefore the separation of chlorine containing compounds which prevents interferences of plasma gas mol- ecular ions such as ArCl' with isotopes of the analyte com- pound.In Fig. 3 a small contribution of the 38Ar40Ar+ dimer loo Ooo 000 000 000 0 1,7% I r 80 100 120 140 20 40 60 I Retention tirnek Fig. 3 Selenium isotope measurements within a GC-ICP-MS chroma- togram of piazselenol to the total ion current at mass number 78 is seen. This relatively constant background intensity can easily be corrected for the 78Se+ ion current in piazselenol because in the time resolved chromatogram all Ar2+ plasma gas ions appear at any time whereas the isotope peak of the element species only appear at the specific retention time of the compound to be determined. For calculation of the isotope ratios it is better to use the total area of an isotope peak instead of the peak height. Only in this case could reproducibilities in the range of 0.6 to 2% RSD be achieved from a GC chromatogram for both selenium isotope ratios 77Se 82Se and 78Se 82Se.The piazselenol was injected on-column into the GC system as a solution in toluene. Toluene and most other organic solvents appear in the first few seconds of the chromatogram with the temperature program described. To prevent carbon deposition on the cones the organic solvent had to be oxidized by oxygen gas. For the separation of piazselenol oxygen gas (purity 99.995%) was introduced with the make up argon flow for the first 50 s after injecting the sample. The use of oxygen of lower purity should be avoided because of its relatively high krypton content which causes interference. It was necessary to stop the O2 gas flow after this time because a distinct depression of the selenium isotope intensities by a factor of about three was found under these conditions compared with use of a pure argon plasma gas.This effect is not described in the literature. Relatively high solvent volumes of up to 6 pl could be applied when oxygen was introduced in the beginning of the separation. This also increased the detection power of the method com- pared with volumes of about 1 pl normally used for similar separation problems. As can be seen from the intensity curve of mass number 78 (Fig. 3) the oxygen gas also reduces the occurence of Ar2+ dimers. However the relatively low con- tribution of this ion to the 78Se+ peak under non-oxygen conditions can be easily corrected. Data processing was per- formed using inhouse programs as no commercial software for the evaluation of ID analysis of transient signals is available.Journal of Analytical Atomic Spectrometry September 1996 Vol. 1 1 889Selenite Spike Solution The 82Se isotope was selected as a spike for ID. A 45mg portion of metallic grey selenium enriched in 82Se by more than 91 YO (Chemotrade Diisseldorf Germany) was dissolved in 15 ml of concentrated nitric acid. After dissolution the acidic solution was diluted by 200ml of pure water and completely reduced to selenite by adding concentrated hydro- chloric acid. The isotope ratio 77Se 82Se of this solution was found to be about 1 150; this composition is too extreme for precise isotope ratio determinations by ICP-MS. By adding selenite of natural isotopic composition and by further addition of concentrated hydrochloric acid a stock solution with a more optimized isotopic distribution of about 1:20 for 77Se 82Se (isotopic abundances 77Se = 3.13% 78Se = 9.73% "Se = 61.9%) was prepared. The concentration of this stock solution was determined by inverse IDMS to be (5.94f0.14) x 1014 selenium atoms per gram using a standard selenite solution of natural isotopic composition.It was found that selenite is stable in this stock solution for many months at 5°C in the dark in a PFA container. In ICP-MS the precision of the isotope ratio measurement is better the closer the isotope ratio is to unity. The isotope ratios of the isotope diluted samples should therefore not distinctly exceed the range between 0.1 and 10. This means that the optimum spike addition depends on the concentration of the species compound to be determined in the sample.I About 2 ml sample I Sample Treatment for Selenite and Selenate Speciation with If non-volatile element species such as selenite are to be determined by GC-ICP-IDMS they must be converted into a volatile compound which is thermally stable under the GC and transfer tube conditions. The specific formation of different piazselenols from selenite was used in the past for the determi- nation of selenite by GC separation and ECD or electron impact ionization MS.19-22 For high detection sensitivities by ECD nitro- or halogeno-groups must be substituted at the piazselenol. A more element-specific detection compared with ECD is possible by ICP-MS. Chlorine and bromine substituted piazselenols as described by Dilli and Sutiknolg and applied by Tanzer,20 are not acceptable in ICP-IDMS because 77Se will be interfered by 40Ar37C1 and high ion intensities of 79Br and 81Br can contribute to the neighbouring 78Se and 82Se isotopes.Nitro-compounds as used by Reamer and Veillon,22 have relatively long retention times so the 1,2-diamino-4- trifluoromethylbenzene was selected as the compound for forming piazselenol with Se'" in acidic solutions by equation (2). This 5-trifluoromethylpiazselenol has relatively short reten- tion times of about 100 s under the GC conditions used in this work (see Fig. 3). GC-ICP-IDMS + 2 H 2 0 + H30+ The different sample treatment steps for selenite speciation in aquatic systems by converting selenite into the fluoropiazse- lenol subsequent extraction of this compound into toluene and determination by GC-ICP-IDMS is schematically rep- resented in Fig.4. The derivatization is carried out in polypro- pylene centrifuge tubes with screw caps (Sigma-Aldrich Deisenhofen Germany). About 2 ml of the sample are diluted by adding about 10ml of pure water. The dilution step is necessary to obtain optimum acidic conditions for reaction (2) Dilution with about 10 ml pure water I 1 Addition of "Se0s2- spike I I Measurement of nSe?2Se and "Se?*Se (correction by mass discrimination factor) I Fig. 4. Sample treatment for selenite speciation in aquatic systems by GC-ICP-IDMS and to avoid reduction of selenate after addition of the spike solution acidified with concentrated hydrochloric acid. Afterwards about 1 ml of the 82Se-enriched selenite spike is added to make total use of one of the main advantages of the IDMS technique which is that after ID has taken place loss of substance has normally no effect on the analytical res~1t.l~ The amount of spike added is optimized with respect to the isotope ratio of the isotope diluted sample.Then 20Opl of a 10 mmol 1- solution of 1,2-diamino-4-trifluoromethylbenzene in 0.25 mol 1-l hydrochloric acid is added and the solution is heated at 70 "C for 30 min. Under these conditions the selenite is totally converted into the corresponding pia~selenol.~'.~~ A single step extraction of the piazselenol is subsequently carried out with 1 ml of toluene. The toluene phase is dried with about 50mg of dehydrated sodium sulfate. To increase the injected selenium amount about 90% of the toluene is evaporated at 80°C after removal of the sodium sulfate with 0.45 pm PTFE syringe filters (Sartorius Gottingen Germany).Portions (1-6 p1) of this concentrated extract are introduced into the on-column injector of the GC system. Selenate is determined by reduction to selenite at 80°C for 1-4 h with about 5 moll-' hydrochloric acid after addition of the 82Se0,2- pike.^^,^' Afterwards the solution is diluted with water to obtain about 0.5 mol 1-1 acidic solution and steps 4 to 8 of Fig. 4 are then carried out as described before. With reference to the larger volume of the aqueous phase 2ml of toluene are used for extraction.The selenate content is deter- mined by subtraction of selenite from the total selenium content after reduction of selenate.However for the determi- nation of selenate in natural waters by this procedure it must be shown that no organoselenium compounds are present which can be converted into inorganic selenium under the sample treatment condition^.^^,^^ The piazselenol is separated from other compounds by the capillary column and then analysed by ICP-MS for its 77Se 82Se and 78Se 82Se isotope ratios respectively. Prior to each set of measurements the actual mass discrimination factor is determined by introducing piazselenol of natural isotopic composition from the diffusion cell into the ICP-MS for correction of the measured isotope ratio of the isotope diluted sample. The evaluation of the selenite concentration for IDMS 890 Journal of Analytical Atomic Spectrometry September 1996 Vol.11analyses in general is carried out as described in more detail el~ewhere.~.'~ RESULTS AND DISCUSSION GC-ICP-MS Coupling System One of the great advantages of coupling GC with ICP-MS is that the total amount of the separated analyte is used for detection whereas in the case LC coupled with ICP-MS more than 95% of the analyte is lost when a conventional nebulizer is used. For example a selenite solution introduced by a conventional nebulizer into the ICP-MS used in this work results in a detection limit of about 1 ng ml-'. Application of a highly efficient direct injection nebulizer in connection with anion-exchange separation resulted in detection limits for selenite and selenate of 7-8 ng m1-'.28 On the other hand after conversion of selenite into fluoropiazselenol and analysis by the described GC-ICP-IDMS system the detection limit was 0.02 ng ml-' determined by the measurement of the 77Se 82Se ratio.A comparison of technical details of GC-ICP-MS systems described in the literature with the corresponding coupling system used in this work is summarized in Table 2 (see also Fig. 1 and Fig. 2). Using steel transfer lines instead of those made of PTFE or glass has one advantage in that they give good and fast heat transfer which is especially important if temperature programs are to be applied for the transfer line. With a temperature limit of 400°C deactivated steel as a capillary material is applicable at a broader temperature range than PTFE.Another advantage of the steel capillary is its mechanical stability. Vibrations at the end of the capillary column in the torch (Fig. l) which produce unstable ion currents can be almost totally prevented. The direct heating of the transfer tube needs less instrumentation and is faster than indirect heating by an aluminium rod. The transfer tube is also heated much more homogeneously than with a resist- ance wire. It could be shown that the temperature in the whole transfer tube was constant by direct heating without any insulation of the tube. Optimization of the plasma (gas flows and torch position) can be carried out with volatile species of the element to be determined. By introducing these compounds from the diffusion cell into the mass spectrometer (Fig.2) with the same gas flows and under the same temperature conditions as used for the GC separation process the instrumental conditions for the special analytical problem can be tested best. This is never possible if the instrumental conditions for a direct introduction of gaseous compounds into the plasma torch are checked by a liquid introduction (nebulizer) system. However determination of the actual mass discrimination factor for an isotope ratio measurement which is necessary for the IDMS technique can only be obtained by a compound of the same element. Table 2 Comparison of different GC-ICP-MS coupling systems Technical detail Transfer capillary Heating of transfer tube Optimization of plasma and other instrumental conditions This work Deactivated steel Direct heating of steel tube by electrical current By volatile element- specific compounds using a diffusion cell under GC conditions Literature PTFE,'v2' quartzg Aluminium rod,30 indirect heating by resistance Gas flow with Hg,30 injection of standard solution (nebulizer system)' Isotope Dilution Technique As mentioned accuracy in element speciation is a general problem. In the case of coupling systems calibration can especially be influenced by matrix effects and the treatment of the sample.For example it was shown that in HPLC-ICP-MS different matrix compositions in the various chromatographic fractions of separated element species can place limitations on the ac~uracy.'~,'~ When applying ID only an isotope ratio and not an absolute amount or concentration must be deter- mined.As the isotopes of an element and of element species respectively show identical behaviour during (chemical) treat- ment of the sample and also with respect to matrix effects the isotope ratio is not influenced by these parameters. Possible chemical isotope effects which are normally in the range of less than 1 per ml can be neg1e~ted.I~ However the mass discrimination effect especially in ICP-MS cannot be neglected for accurate analyses by ICP- IDMS. The determination of natural 78Se "Se isotope ratios by nitrogen MIP-MS also shows mass bias.32 Catterick et ~ 1 . ~ ~ stated that the determination of the mass discrimination factor is an important topic for accurate ICP-IDMS analyses. The mass discrimination factor with respect to equation (1) must therefore be determined by a corresponding calibration com- pound. This can be best carried out by continuous introduc- tion of a volatile compound from a diffusion cell into the mass spectrometer under the same conditions as for the GC separation (Fig.2). The mass discrimination factors for the selenium isotope ratio measurements of this work are listed in Table 3. From these results one can see that this factor for both isotope ratios determined 77Se 82Se and 78Se "Se is independent of the selenium compound when applying piazselenol and dimethyl selenide respectively. The mass discrimination factors were calculated on the basis of the natural isotopic composition of selenium with 'true' values recommended by IUPAC.34 The factor for selenium isotopes was found to be relatively high by about 3% per mass unit but was constant at least for a whole set of measurements and the day-to-day variation was not significant. This enables the necessary corrections to be made for accurate IDMS results. The reason for this relatively high mass discrimination effect could not be explained up to now.A possible discrimination by the detector dead time could be neglected because count rates were usually less than 100000 s-' at mass number 78. Also mass calibration and ion lens potentials were corrected before each set of measurements. However the finding of relatively high mass discrimination factors was confirmed by V O ~ I ~ ' who found the mass discrimi- nation factor for zinc to be 3% (measurement of 64Zn:66Zn) and for zirconium to be 2.3% ("Zr 94Zr) per mass unit using the same ICP-MS with a conventional nebulizer. Selenium Speciation by GC-ICP-IDMS Important selenium species which can be found in the environ- ment are listed in Table 4 together with possible chromato- graphic ICP-MS coupling systems for determination.When applying ID in connection with the GC-ICP-MS an exactly known quantity of the corresponding spike enriched with 82Se should be added to the sample at the beginning of the analytical procedure. The sample treatment process must guarantee that no conversion of one selenium species into another one takes place and that derivatization processes are specific for only one species. Two different water samples the NIST standard reference material SRM 1643b Trace Elements in Water and the selenium species reference material CRM 602 from BCR Brussels were analysed by GC-ICP-IDMS for selenite and selenate using the sample treatment procedure represented in Fig.4 (Table 5 ) . Journal of Analytical Atomic Spectrometry September 1996 Vol. 11 891Table 3 Mass discrimination factors for selenium isotope ratio measurements by GC-ICP-MS Isotope ratio Selenium compound Piazselenol Dimethyl selenide ~~ Measured Na t Mass discrimination factor 77Se ”Se 0.762 0.874 1.15 78Se 82Se 2.44 2.72 1.12 77Se 82Se 0.766 0.874 1.14 78Se ”Se 2.46 2.72 1.11 Table 4 Selenium species in the environment References Species Trimethylselenonium ion Selenoamino acids Selenium-HUS compounds Selenite selenate Dimethyl selenide DMSe Dimethyl diselenide DMDSe Dimethylselenone Possible ICP-MS method IC-ICP-M S * RPC-ICP-MSt RPC-ICP-M S IC-ICP-MS or GC-ICP-MS after derivatization GC-ICP-MS GC-ICP-MS GC-ICP-MS * IC =ion chromatography t RPC = reversed-phase chromatography Table 5 Selenite and selenate concentrations (in ng m1-l) in certified reference water samples by GC-ICP-IDMS NIST SRM 1643b BCR CRM 602 77Se 82Se 78Se 82Se 77Se 82Se 7sSe 82Se Selenite 10.6f0.3 10.7k0.5 5.0k0.7 5.1k0.8 Selena te < 0.02 < 0.02 7.4+ 1.3 7.42 1.1 G C- IC P- ID MS- Certijied values- Selenite - Selenate - Total selenium 9.7 + 0.5 5.8 k 0.4 7.7 If 0.7 - For both samples the reliability of the results was proved by measuring both isotope ratios 77Se 82Se and 78Se 82Se.The results listed in Table 5 show that the calculated concentrations on the basis of both measured isotope ratios are identical within the limits of the given standard deviations (1s) of 0.3 to 0.5% (three independent determinations).The accuracy of the results is shown by the agreement of the GC-ICP-IDMS data with the certified values. CONCLUSION The GC-ICP-IDMS system described here allows in principle accurate determinations of all volatile element species if a corresponding spike compound is available. Most spike com- pounds for element speciation can easily be synthesized as has been demonstrated in this work for selenite. The costs for an isotopically enriched spike are usually negligible compared with the other costs of the analysis because less than 1 pg of a spike compound is necessary per analysis.We thank the ‘Deutsche Forschungsgemeinschaft’ for financial support. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Organometallic Compounds in the Environment ed. Craig P. J. Longman UK 1986. Frimmel F. H. Fresenius’ J. Anal. Chem. 1994 350 7. Tanzer D. and Heumann K. G. Atmos. Environ. 1990,24A 3099. Heumann K. G. in Metal Speciation in the Environment ed. Broekaert J. A. C. Giicer S. and Adams F. Springer Heidelberg 1990 NATO AS1 Series vol. G23 pp. 153-168. Crews H. M. Dean J. R. Ebdon L. and Massey R. C. Analyst 1989 114 895. Gercken B. and Barnes R. M. Anal. Chem. 1991 63 283. Hansen S. H. Larsen E. H. Pritzl G. and Cornett C. J. Anal. At. Spectrom. 1992 7 629. Feldmann J. Grumping R. and Hirner A. V. Fresenius’ J.Anal. Chem. 1994,350 228. Prange A. and Jantzen E. J . Anal. At. Spectrom. 1995 10 105. De Bikvre P. Fresenius’ J. Anal. Chem. 1990 337 766. Heumann K. G. Mass Spectrom. Rev. 1992 11 41. Heumann K. G. Rottmann L. and Vogl J. J. Anal. At. Spectrom. 1994 9 1351. Rottmann L. and Heumann K. G. Anal. Chem. 1994,66 3709. Rottmann L. and Heumann K. G. Fresenius’ J. Anal. Chem. 1994,350 221. Brown A. A. Ebdon L. and Hill S . J. Anal. Chim. Acta 1994 286 391. Al-Attar A. F. and Nickless G. J. Chromatogr. 1988 440 333. Heumann K. G. in Inorganic Mass Spectrometry ed. Adams F. van Grieken R. and Gijbels R. Wiley New York 1988 Knapp D. R. Handbook of Analytical Derivatization Reactions Wiley New York 1979 p. 20. Dilli S. and Sutikno I. J . Chromatogr. B 1984 300 265. Tanzer D. PhD Thesis University of Regensburg 1990. Elaseer A. and Nickless G. J. Chromatogr. A 1994 664 77. Reamer D. C. and Veillon C. Anal. Chem. 1981 53 2166. Johannson K. and O h A. J. Chromatogr. 1992,598 105. Bye R. and Lund W. Fresenius’ J. Anal. Chem. 1988 332 242. Pettersson J. and Olin A. Talanta 1991 38 413. Tanzer D. and Heumann K. G. Anal. Chem. 1991,63 1984. Ornemark U. and Olin A. Talanta 1994 41 1675. Shum S. C. K. and Houk R. S. Anal. Chem. 1993 65 2972. Hintelmann H. Evans R. D. and Villeneuve J. Y. J. Anal. At. Spectrom. 1995 10 619. Kim A. W. Foulkes M. E. Ebdon L. Hill S. J. Patience R. L. Barwise A. G. and Rowland S . J. J. Anal. At. Spectrom. 1992 7 1147. Peters G. R. and Beauchemin D. Anal. Chem. 1993 65 97. Yoshinaga J. Shirasaki T. Oishi K. and Morita M. Anal. Chem. 1995 67 1568. Catterick T. Handley H. and Merson S. At. Spectrom. 1995 16 229. IUPAC Commission on Atomic Weights and Isotopic Abundances Pure Appl. Chem. 1991 63 991. Vogl J. personal communication University of Mainz 1996. pp. 301-376. Paper 6/01 71 5K Received March 11 1996 Accepted June 24 1996 892 Journal of Analytical Atomic Spectrometry September 1896 Vol. 11