Application of a nitrogen microwave-induced plasma mass spectrometer as an element-speciÆc detector for arsenic speciation analysis Amit Chatterjee,* Yasuyuki Shibata, Jun Yoshinaga and Masatoshi Morita National Institute for Environmental Studies, Environmental Chemistry Division, Environmental Chemodynamics Section, 16-2 Onogawa, Ibaraki 305 0053, Tsukuba Science City, Japan Received 24th June 1999, Accepted 14th October 1999 A high power nitrogen microwave-induced plasma (1.3 kW) mass spectrometer (N2-MIP-MS) was successfully coupled with an HPLC system using a silica-based cation-exchange column.It was examined as an elementspeciÆc detector for its applicability to the optimization and determination of seven arsenic compounds [As(V), methylarsonic acid (MA), dimethylarsinic acid (DMA), arsenobetaine (AB), arsenocholine (AC), trimethylarsine oxide (TMAO) and tetramethylarsonium ion (TMI)]. The system is a promising alternative ion source for mass spectrometry for elemental speciation analysis. The MIP was stable with a pyridine mobile phase for up to 6 h.Replacing the MIP-MS fabricated nebulizer (concentric) and sample input tubing (PTFE) with an ICP-MS (PMS-2000) nebulizer (concentric) and PEEK tubing increased the ion signals for anionic and cationic arsenic compounds by 17±30 and 21±25%, respectively. PEEK tubing additionally increased the separation efÆciency for the arsenic compounds. The detection limits of As(V), MA, DMA, AB, TMAO, AC and TMI obtained with the optimized HPLC-N2-MIP-MS system were 0.68, 0.95, 2.01, 0.92, 22.1, 1.31 and 1.75 mg l21, respectively.The repeatability (RSD for three successive analyses) and reproducibility (RSD for three successive analyses performed on three different days) achieved were 0.7±9.22 and 6.5±11.4%, respectively, for the seven different arsenic compounds. No detectable spectroscopic interference of 40Ar35Clz was observed with a high chloride matrix (10 000 mg l21). The developed HPLC-N2-MIP-MS method was successfully applied to the determination of arsenic compounds, principally AB, in NIES Candidate CRM-18 Freeze-Dried Human Urine (134°6 mg l21).The results agreed reasonably well with the HPLC-ICP-MS values. Introduction Trace element speciation can generally be realized by complex analytical systems, in which the species information is generated by some kind of separation method. The detection of the separated species is performed by a spectroscopic analytical method, resulting in atomic or molecular information.Chromatographic separation with element-speciÆc detection is clearly a favorable method for the determination of different forms of an element.1 Chemical speciation of arsenic compounds is a topic under extensive study, because of the very rich chemistry, the diverse toxicity of its compounds, and the dramatic differences in metabolism of the various arsenic species. The powerful speciation techniques developed during the past few years have played an essential role in providing information for understanding the distribution and fate of arsenic in biological and environmental systems.2,3 The combination of chromatographic separation with elementspeciÆc spectrometric detection has proved to be very useful for the speciation of trace levels of arsenic compounds. In particular, liquid chromatographic separation with inductively coupled plasma mass spectrometry (ICP-MS),2±6 electrospray mass spectrometry (ESMS),7 capillary electrophoresis,8 and atomic spectrometry1,9±11 has played an important role in chemical speciation studies.In the development of such hyphenated techniques the matching of separately optimized analytical procedures in different instruments has to be resolved. FAAS, ICP-AES, HGAAS and ICP-MS instruments can easily be matched to the Øow system of HPLC and such methods are readily applicable to metallic speciation analysis; 12,13 however, the organic eluents can disturb the stability of the detector.12 The detection limits obtained with FAAS and ICP-AES are too high.ETAAS provides low detection limits, but suffers from pre-atomization losses. HGAAS also generates low detection limits for arsenic species; however, for arsenobetaine (AB), tetramethylarsonium ion (TMI) and arsenochlorine (AC), prior decomposition to a hydrideforming form is essential prior to HG.4,14 The Ar-ICP-MS has been widely used as an ion source in speciation elemental mass spectrometry, since ICP-MS provides spectral simplicity, multi-elemental analysis and low detection limits for the determined elements.15,16 However, spectral and non-spectral interferences are still a serious problem.The mono/polyatomic ions caused by the Ar plasma sustaining gas, such as 40Ar1Hz, 40Arz, 40Ar12Cz, 40Ar15Nz, 40Ar16Oz, 40Ar18Oz, 40Ar35Clz, 40Ar37Clz, 40Ar38Arz, 40Ar2 z, interfere with the determination of 39Kz, 40Caz, 52Crz, 55Mnz, 56Fez, 58Fez, 75Asz, 78Sez and 80Sez, respectively.16,17 To reduce the Ar-associated polyatomic ions for Ar-ICP, sources other than Ar or mixed gases with Ar and several types of MIP sources have been developed and studied.16±20 The gas-Øow system of a microwave- induced plasma (MIP) can be coupled directly to GC systems; therefore, MIP-AES has long been applied as an element-speciÆc GC detector.12 The introduction of liquid samples into the MIP discharge is, however, not free from difÆculties.The limited thermal energy and the relatively small volume of the MIP restrict the analyte Øow to be introduced.20 A high power nitrogen MIP-MS having a rectangular waveguide and an Okamoto cavity (2.45 GHz, maximum 1.5 kW; constructed by Hitachi, Ibaraki, Japan) is one of these sources which is sustained by nitrogen gas.18,19 The plasma is doughnut-shaped just like the Ar-ICP. It enables the introduction of sample aerosols directly into the center of the J.Anal. At. Spectrom., 1999, 14, 1853±1859 1853 This journal is # The Royal Society of Chemistry 1999plasma.14±19 Because the interference of Ar-associated polyatomic ions is replaced by that of nitrogen-related polyatomic ions, MIP-MS could determine arsenic without spectral interference.15±19 Surprisingly, relatively little work with a high power N2-MIP-MS has been reported for total elemental analysis.15,17,21,22 In previous papers, we reported the use of high power N2-MIP-MS for the isotope dilution analysis of selenium in clinical and marine samples.21,22 The methods are highly sensitive with a low detection limit, and are comparable to those of ICP-MS.23 Very few papers concerning the coupling of HPLC with N2-MIP-MS for elemental speciation analysis, especially for arsenic, have been published. This paper describes the feasibility of high power N2-MIPMS as an arsenic-speciÆc detector, for the determination of seven arsenic species by coupling with an HPLC system.Two types of sample introduction mode, i.e. combination of a Hitachi supplied conventional MIP-MS nebulizer with TeØon tubing and an ICP-MS (PMS-2000) Meinhard-type nebulizer with PEEK pressure tubing, were used. The results obtained by both sample introduction methods are compared in terms of analytical Ægures of merit for the seven arsenic species. Finally, the present HPLC-N2-MIP-MS (PMS-2000 nebulizer along with PEEK tubing) system was applied successfully to the determination of AB in NIES Candidate CRM-18 Freeze- Dried Human Urine.Attention was also paid to the detection limits and sensitivities of the arsenic compounds. Experimental Reagents and solutions NIES Candidate CRM-18 Freeze-Dried Human Urine was used as a reference material (National Institute for Environmental Studies, Environmental Chemistry Division, Tsukuba, Ibaraki, Japan). All solutions were prepared with Milli-Q (18.3 MV cm; Milli-QSP.TOC Reagent Water System, Nihon Millipore, Yonezawa, Japan) water.The mobile phase for cation-exchange chromatography was prepared by dissolving 1.58 g of pyridine (Merck, pro analysi; Merck, Darmstadt, Germany) in Milli-Q water and adjusting the pH of this solution to 2.5 by adding formic acid (#98%, Fluka, puriss, p.a., Fluka, Buchs, Switzerland); the solution was then made up to 1000 ml (20.0 mM) with Milli-Q water. Standard solutions (1000 mg l21 As) for the identiÆcation and quantiÆcation of arsenic compounds were prepared by dissolving 433.0 mg of NaAsO2 (Merck, pro analysi) [As(III)], 1.041 mg of Na2HAsO4?7H2O (Merck, pro analysi) [As(V)], 460.5 mg of dimethylarsinic acid (DMA), 466.6 mg of methylarsonic acid (MA), 594.2 mg of arsenobetaine bromide (AB), 554.2 mg of arsenocholine (AC), 453.9 mg of trimethylarsine oxide (TMAO) and 874.2 mg of tetramethylarsonium iodide (TMI) (Tri Chemical Laboratory, Yamanashi, Japan) in 250 ml of water.Calibration graphs for the HPLC-MIP-MS measurements were obtained by injecting aliquots (0.100 ml) of solutions containing 5.00, 10.0, 25.0, 50.0 or 100.0 mg l21 As and of As(III), As(V), MA, DMA, AB, AC, TMAO and TMI for the separations on the Supelcosil LC-SCX cation-exchange column. Nitrogen microwave-induced plasma mass spectrometry (N2-MIP-MS) A Hitachi P-6000 N2-MIP-MS, which is commercially available from Hitachi (Ibaraki, Japan) was used for this study.In Table 1 the operating conditions for N2-MIP-MS are displayed. The microwave power (2.45 GHz, maximum 1.5 kW) was produced by a magnetron (H3862: Hitachi) operated by a dc power supply and fed to a cavity known as the Okamoto cavity18,19 through a rectangular waveguide (WRJ-2). The cavity was cooled by circulating cold water (20 �C) from a refrigerator. To establish stable impedance matching between the plasma and the microwave power source, the height of the rectangular waveguide was reduced from its regular size (WJR- 2) of 54.2 to 8.0 mm by using a tapered waveguide.An annularshaped 15 electric Æeld, where the microwave discharge is maintained, was produced between an inner conductor and an outer cylindrical conductor terminated by a front plate. A quartz discharge tube of 10 mm id (1 mm thickness) was inserted into the inner conductor. The tube consisted of two concentric tubes, an inner and an outer. The inner tube was tulip-shaped, with a large outer diameter of 9 mm and a small inner diameter of 1 mm.The carrier gas (N2; 1.1 l min21) with the sample aerosol was fed through the central oriÆce (id 1.0 mm) and the plasma support gas (N2; 15 l min21) was fed into the cylindrical gap between the two quartz tubes, respectively. The plasma was ignited by a Tesla coil. A stable, annular (doughnut)-shaped, pink nitrogen plasma was produced. A Meinhard concentric-type nebulizer (Hitachi Electric, Ibaraki, Japan, Part No.P97M170, 300-8350) connected with TeØon sampling tubing (PTFE, Hitachi, Part No. 300-8868) and an ICP-MS concentric-type nebulizer (PMS-2000, Yokogawa Electric, Tokyo, Japan; Part No. K9250YW) with 30060.25 mm id PEEK (polyether ether ketone) capillary tubing without a desolvation system were used for the sample introduction systems. The exit of the HPLC column was connected directly to the nebulizer via the PEEK capillary or PTFE tubing. A Neslab refrigerating circulator was used to maintain the temperature of the glass spray chamber at 5 �C.The normal aqueous sample solution uptake rate was about 0.3 ml min21 when the Øow rate of the nitrogen gas (carrier) was 1.1±1.4 l min21. However, with HPLC connection Table 1 Operating conditions of nitrogen-MIP-MS N2-MIP-MS: Frequency 2.45 GHz Microwave power (a) Forward 1.3 kW (b) ReØected v20 W Plasma gas Øow/l min21 15 Carrier gas Øow/l min21 1.1 Peak point /mass 10 Dwell time 2.0 ms Number of sweeps 1500 Nebulizer (Meinhard) Concentric Temperature of spray chamber v5 �C Sampling cone (Pt) 0.8 mm oriÆce Skimmer cone (Pt) 0.4 mm oriÆce Sample uptake rate/l min21 1.5 Measuring parameters: Monitored signals: 75Asz, 77Sez, 40Ar37Clz m/z 75, 77 Total analysis time 600 s Vacuum pump: Stage 1: Mechanical pump DPF-6ZS, 750 l min21, Daiashinku, Japan Stage 2: Vapour pump DPF-6ZS, 1200 l s21, Kashiyama, Japan Stage 3: Vapour pump DPF-4ZS, 570 l s21, Kashiyama, Japan Both vapour pumps are Ætted with water-cooled bafØe valves Mass analyzer: Quadrupole analyzer: QMG420-4 Balzers, Liechtenstein; Mo rods, 200 mm length, 8 mm diameter, radiofrequency of 2.25 MHz Ion detector: Channeltron electron multiplier 4831 G, Dalileo, USA, mounted on a quadrupole analyzer with 90� ion deØection and off-axis Pulse counting: Pulse ampliÆer C3866, Hamamatsu Photonics, resolution 10 ns, maximum counting rate 107 counts s21. 1854 J. Anal. At. Spectrom., 1999, 14, 1853±1859using 20 mM pyridine as mobile phase at a Øow rate of 1.0± 1.5 ml min,21 the plasma was stable for 6 h.A more detailed description of the microwave system, plasma ignition system, sample introduction system, MS instrument and interface system has been published previously.15,18,19 Chromatography The HPLC system consisted of a Perkin-Elmer Model Series 410 B10 solvent delivery unit (Perkin-Elmer, Norwalk, CT, USA) and a Rheodyne 9725 six-port injection valve (Cotati, CA, USA) with a 100 ml injection loop.The separations were performed on a Supelcosil LC-SCX (Supelco, Bellefonte, PA, USA) cation-exchange column (25 cm64.6 mm id, 5 mm silicabased particles with propylsulfonic acid exchange sites). The column was equilibrated by passing at least 100 ml (Øow rate 1.0 ml min21) of the mobile phase through the column before injection of the arsenic compounds. The exit of the column was connected directly to the nebulizers with PEEK capillary or PTFE tubing. The ion signals at m/z 75 (75As) and m/z 77 (40Ar37Cl) were recorded with the time-resolved analysis software# Version of Hitachi.For quantiÆcation, the chromatograms were exported, peak areas and peak heights determined, and the concentrations calculated with external calibration graphs and with the standard additions technique. Urine preparation Freeze-Dried Human Urine (NIES Candidate CRM-18) served as a reference material for the application study. The reference value for the total arsenic was 134°6 mg l21 when it was dissolved in 9.57 ml of water, but this is not a certiÆed value.The urine was prepared from 10.0 l of urine collected from nonexposed male clerical workers of the NIES in late July 1996. The collected urine was immediately Æltered through a 5 mm membrane Ælter and stored in a freezer in one lot until further treatment. Frozen urine was thawed 2 weeks after collection and Æltered again through a 0.45 mm membrane Ælter.Ten grams of the Æltered urine (10.0°0.05 g) were dispensed into 930 borosilicate bottles, which had been pre-cleaned with nitric acid and numbered. The weight of the dispensed urine was recorded for every 50 bottles for calculation of weight loss due to subsequent freeze-drying. All of the bottles except for 50 were freeze-dried in one batch at Wako Pure Chemicals, Mie, Japan. Freeze-dried urine was stored at 4 �C at NIES. The weight loss of the urine due to freeze-drying was calculated to be 9.57°0.02 g (n~19).A more detailed description of the preparation of the urine samples was given previously.24 A 9.57 ml volume of pure water was added to each bottle for reconstitution. The bottles were swirled gently to dissolve the material completely. The reconstituted urine was stored at 4 �C prior to analysis. Results and discussion Chromatographic separation of arsenic compounds Among the many methods which have been applied to the chromatographic separation of arsenic compounds, a common and reliable method is ion-exchange chromatography.The polarity of cationic and anionic arsenic compounds makes them amenable to ion-exchange HPLC. A cation-exchange chromatographic system was used (LC-SCX), because this provides the possibility of applying a low organic content (20 mM pyridine) of the eluent and is compatible with the limited loading of the MIP source. The normal aqueous sample solution uptake rate in this N2-MIP is very low and about 0.3 ml min21 when the Øow rates of the nitrogen gas (carrier) are 1.1±1.4 l min21.15 However, with the HPLC connection using 20 mM pyridine as mobileØow rate of 1.5 ml min21, the plasma is stable and the count variation is about 5±8% during 6 h of continuous operation.The high power N2-MIP produces an annular (doughnut)-shaped plasma15±19 that has superior operational stability and a higher tolerance for the injection of highly aqueous aerosol samples.18 This plasma is more robust than the Ar-ICP-MS and is not extinguished even if an air sample is injected.Hence, a high mobile phase Øow rate (1.0±1.5 ml min21) is compatible with the MIP system. Fig. 1 and 2 show the retention times of anionic and cationic arsenic compounds in LC-SCX using 20 mM pyridine at pH 2.50. The AB, AC, TMAO, TMI, MA, DMA and As(V) are well separated. However, the signals for arsenite and MA overlap, making their simultaneous determination impossible with the present chromatographic system.Further details of the separation methods have been described previously.1,4,25 Optimization of experimental parameters In an attempt to obtain a maximum line to background intensity ratio and stability for the seven arsenic compounds, various operating parameters were studied and optimized individually with the on-line chromatographic system using a mixture of the seven arsenic compounds, while the other parameters were kept at their optimum values.The parameters investigated were microwave power to the plasma, and nitrogen Øow rates of the plasma and carrier gas for both sample introduction systems. It was observed that on increasing the microwave power the ion signals of all the arsenic compounds (seven) increased up to 1.3 kW (Fig. 3). A further increase of power was not possible. Hence, a 1.3 kW microwave power was used during the measurements. The effect of microwave power on the elemental ion signals is correlated with the ionization potential.18,19,26 A higher Fig. 1 Chromatogram obtained with a solution of dimethylarsinic acid (DMA), methylarsonic acid (MA) and arsenate [As(V)] (10 ng As from each species) in distilled water on a Supelcosil LC-SCX cationexchange column with N2-MIP-MS (for optimized conditions see Table 1; mobile phase, 20 mM pyridine at pH 2.50; injection volume, 0.100 ml; Øow rate, 1.5 ml min21). Fig. 2 Chromatogram obtained with a solution of arsenobetaine (AB), trimethylarsine oxide (TMAO), arsenocholine (AC) and tetramethylarsonium iodide (TMI) (10 ng As from each species) in distilled water on a Supelcosil LC-SCX cation-exchange column with N2-MIP-MS detection (conditions as in Fig. 1). J. Anal. At. Spectrom., 1999, 14, 1853±1859 1855microwave power is necessary for elements with high ionization potentials.26,27 For elements with high ionization potentials such as As (9.8 eV), the intensity is enhanced on increasing the forward microwave power,26 which agrees with our experimental Ændings.The annular-shaped plasma is inØuenced by the microwave power.18 The stable region of the plasma increases with increasing microwave power.18 This is because the plasma pressure and the plasma diameter increase with increasing microwave power. The tail Øame of the plasma also increases with increasing power.18 A long plasma is expected to lead to longer analytical residence times of the analytes in the plasma.18 Consequently, greater decomposition and higher ionization of the arsenic compounds occur, which ultimately increases the central analyte-rich region. Hence, the ion signals increase on increasing the microwave power.As the counts increase with increasing power, further increase in power may increase the concentration of 75Asz in the plasma.18,19,26 It is desirable for the power to be as high as possible to increase the sensitivity. Hence, further work will be directed towards lowering the detection limits of arsenic compounds by increasing the microwave power.The plasma and carrier gas Øow rates were altered to optimize the ion signals of the arsenic compounds, keeping the microwave power at 1.3 kW. On increasing the plasma Øow from 13 to 19 l min21, the ion signals increase up to 15 l min21 and reach a maximum at this value (Fig. 4). A further increase of plasma Øow rate does not cause the signals to vary signiÆcantly. This is probably as a result of the high viscosity of the plasma, which is not able to accommodate the increased gas Øow without some spatial distribution to provide an unimpeded path for the plasma gas.18 The visual hole of the annularshaped plasma produced by the carrier gas Øow decreases with an increase in the plasma Øow rate, as has been observed by Okamoto.18 The increase in ion signals with plasma gas Øow rate up to 15 l min21 is due to an increase in the stable plasma region, which Ænally increases the production of ions. A 1.1 l min21 carrier Øow rate (0.6±1.3 l min21) provides the maximum optimized signals for the seven arsenic species when peak area measurement is considered. A further increase/ decrease results in a decrease in total counts, as has been observed previously for total arsenic determination.15,26 It should be noted that the maximum ion signal value may be obtained as a compromise between the increment of aerosol into the plasma and the decrease of temperature in the central part of the plasma.This is recognized by the observation of a change in the color of the central part of the plasma with an increase in the nebulizer gas Øow rate. The color of the central part of the nitrogen plasma changes to a pale and whitish pink on increasing the nebulizer gas Øow rate above 1.1 l min21. Hence, the ion signals decrease above a 1.1 l min21 carrier Øow. Below 1.1 l min21, the carrier Øow decreases the aerosol production, transportation and injection velocity necessary to produce a central analyte-rich region, which ultimately leads to a decrease of the ion signals.Details of the optimization of the carrier Øow have been given by several workers.15,19,26 Effect of nebulizer and tubing The ICP-MS nebulizer with PEEK tubing increases the ion signals by at least 17±30% for anionic species and 21±25% for cationic species (Table 2) and is more efÆcient than the in-built instrumental nebulizer with PTFE tubing.The PEEK capillary tubing used additionally increases the counts (ion signals) and gives a better baseline separation of the arsenic compounds compared with the PTFE tubing (Table 2). By using the PEEK tubing, the peak heights are also enlarged (35±78%; Table 2) and the peaks are very sharp. In particular, the PTFE tubing has a larger diameter than the PEEK tubing. Hence, after separation, the analytes are diluted and diffused during transportation. Further, the PTFE tubing increases the dead volume more than the PEEK tubing when both are connected individually to the HPLC column and the nebulizer.As a result, the PTFE tubing reduces the separation efÆciencies and lowers the ion signals of the arsenic compounds. The ICP-MS nebulizer with PEEK tubing enhances aerosol production and its transmission efÆciency to the MIP, thereby increasing the sensitivity. Thus, it was used for all further measurements. Figures of merit The performance of the hyphenated technique was characterized by the linearity of the calibration graphs, repeatability, reproducibility and limits of detection (based on 2s of the blanks) within a series, which were calculated for three replicate measurements and are listed in Table 3.The intensities for the HPLC-N2-MIP-MS signals (on-line injection of a mixture of the seven arsenic compounds) given, indicate that a dynamic range of at least two orders of magnitude (Fig. 5) is necessary to cover the response range caused by the arsenic compounds present in the urine.Calibration graphs, obtained from the areas of the signals (three replicates) in chromatograms with standard solutions [As(V), MA, AB, AC, DMA and TMI at concentrations of 5.00, 10.0, 25.0, 50.0 and 100 mg l21 As; TMAO at concentrations of 10.0, 25.0, 50.0 and 100 mg l21 As)], are linear (Fig. 5). Repeatability is good for all the species; with respect to peak areas better than 9.2% and to peak heights 3.97% (Table 3; Fig. 6).Day-to-day reproducibility for the seven arsenic compounds is better than 11.4% for peak areas. We also analyzed and calculated the total counts and the detection limits of the investigated arsenic compounds using on-line injection of a mixture of the seven arsenic compounds with HPLC-ICP-MS. The total counts of the seven (individual) arsenic compounds found by using the HPLC-N2-MIP-MS and HPLC-ICP-MS systems are presented in Table 2. The counts of the seven arsenic compounds obtained with the HPLC-N2-MIP-MS system are lower than those with Fig. 3 Dependence of the ion signals of the seven arsenic compounds [As(V), MA, DMA, AB, AC, TMAO and TMI] on the microwave power using HPLC-N2-MIP-MS (conditions as in Fig. 1). Fig. 4 Dependence of the ion signals of the seven arsenic compounds [As(V), MA, DMA, AB, AC, TMAO and TMI] on the plasma gas Øow rate using HPLC-N2-MIP-MS (conditions as in Fig. 1). 1856 J. Anal. At. Spectrom., 1999, 14, 1853±1859the HPLC-ICP-MS system (Table 2).The detection limits for As(V), MA, DMA, AB, TMAO, AC and TMI are 0.69, 0.32, 0.86, 0.18, 6.33, 0.41 and 0.37 mg l21 As, respectively. Hence, the detection limit obtained for As(V) by using HPLC-N2-MIPMS is the same as that obtained with HPLC-Ar-ICPMS. However, for DMA (2), MA (3), TMAO (3), AC (3), AB (4) and TMI (5 times), the detection limits are about 2±5 times higher than those obtained with HPLC-Ar-ICP-MS (Table 2).The repeatability, reproducibility and correlation coefÆcient obtained with HPLC-high power-N2-MIP-MS are comparable to those obtained with HPLC-ICP-MS.28 The detection limit of arsenic is about 1±2 orders of magnitude lower than those reported for other N2-MIP systems.15,18,19 The main reasons for the lower detection limit may be the difference in the microwave power (500 versus 1300 W) used, and the difference in the shape (annular) of the plasma formed. However, owing to the high ionization potential of arsenic the detection limit is higher than that for Ar-ICP-MS.18,19 The higher detection limit of arsenic in N2-MIP-MS is because of the interference of the high concentration of 30NOz in the nitrogen plasma.18,19 It has been reported that the 30NOz content decreases on increasing the microwave power.18,19 Hence, further investigation will be aimed at lowering the detection limit of arsenic by increasing the microwave power or by using helium as the plasma support gas.Spectroscopic interference An interference study with chloride ion (500±10 000 mg l21) was carried out to observe whether there was any enhancement of the arsenic signal (m/z 75) by 40Ar35Cl.z The typical background signals at m/z 5±100 for Milli-Q water, 20 mM pyridine and 10 000 mg ml21 chloride are displayed in Fig. 7. Background counts of 290±300 were found with Milli-Q water at m/z 75. With the pyridine mobile phase the background signal was increased to 500±520 counts. No detectable arsenic signal (counts) at m/z 75 was observed when 500, 1000, 5000 and 10 000 mg l21 chloride were injected onto the column (Fig. 7). Hence, 40Ar35Clz interference due to chloride is absent, which agrees well with previous Ændings.26,27 This may be because the polyatomic species which are generated from the chloride matrix are about two orders of magnitude less intense than those of the Ar-ICP ion source.27 The increase of background counts in the pyridine mobile phase is owing to arsenic contamination, either from the pyridine or from the formic acid, used for mobile phase preparation.Analysis of real samples To validate the developed HPLC-MIP-MS method, NIES CRM-18 Freeze-Dried Human Urine was analyzed for the certiÆcation of arsenic compounds, particularly AB (Fig. 8). The peaks were identiÆed by comparison of the retention times (Fig. 1 and 2) with those of authentic standards and were conÆrmed by spiking with the standard arsenic compounds.Arsenate, MA, DMA and AB were detected, and conÆrmed by HPLC-ICP-MS using anion,1 cation,25 and reversedphase 6,24,29 columns with different mobile phases described previously.1,6,24,25,29 The AB concentration, estimated using the standard additions technique, was 78.1°1 mg l21, in agreement with the HPLC-ICP-MS value (72.6°8 mg l21). The AB is present as a major arsenic species (58% of the total arsenic) in urine followed by DMA, As(V) and MA.The AB, As(V), MA and DMA concentrations were also calculated using external calibration graphs, but the concentration of AB found was higher (90.8°7 mg l21) with respect to the standard additions method, indicating signal enhancement due to the urine matrix. The sum of all four arsenic species (arsenate, 8.92°1.1 mg l21; MA, 11.2°0.5 mg l21; and DMA, 41.5°0.7 mg l21) is consistent with the total arsenic content (134°6 mg l21). Moreover, Table 2 Total countsa of different arsenic species with two different nebulizing systems and with HPLC-ICP-MS MIP-MS nebulizerzPTFE tubing connecting the nebulizer and the HPLC column ICP-MS nebulizerzPEEK tubing connecting the nebulizer and the HPLC column HPLC-ICP-MS with PEEK tubing Arsenic species (100 mg l21) Area Height Area Height Area Height AB 8500°590 2210°131 10 320°73.5 (25)b 3950°89 (78) 50 600°136 8870°56 TMAO 1970°175 132°20.2 2390°60 (21) 179°2.83 (35) 7460° 499 230°34 AC 9910°850 1920°180 11 960°160 (20) 2950°114 (50) 39 900°3060 3650°240 TMI 9420°580 1640°118 11 560°468 (23) 2270°3.54 (38) 62 700°4410 4330°340 As(V) 12 000°120 2470°35.4 14 660°813 (23) 5860°24.7 (38) 18 900°76 2510°254 MA 15 610°570 2780°169 18 300°687 (17) 4270°76 (53) 45 200°2060 4700°223 DMA 14 140°610 1310°83 18 370°574 (30) 1870°60 (42) 32 600°1280 1740°55 aAverage of three determinations.bPercentage increased value is given in parentheses. Table 3 Detection limits,a repeatability,b reproducibilityc and correlation coefÆcients (R2)d after chromatographic separation of the arsenic compounds Repeatability (%) Reproducibility(%) R2 Analyte Detection limit/mg l21 (% RSD) Area Height Area Height Area Height As(V) 0.68 (5.2) 5.6 2.4 11 12 0.9978 0.9982 MA 0.95 (1.9) 9.2 1.8 8.3 5.2 0.9997 1.000 DMA 2.01 (2.8) 5.8 3.2 9.2 6.1 0.9992 0.9999 AB 0.92 (5.6) 0.7 2.3 6.5 6.2 0.9996 0.9964 TMAO 22.1 (1.6) 2.6 1.6 11 5.3 0.9982 0.9984 AC 1.31 (4.2) 3.8 6.0 7.8 5.2 0.9998 1.000 TMI 1.75 (1.4) 4.1 2.6 11 6.3 0.9999 0.9990 aDetection limits were determined as the elemental concentrations giving a signal twice the standard deviation (n~5) of the blank (Milli-Q water injected with 20 mM pyridine mobile phase).bRepeatability was determined for peak areas and peak heights by calculating the relative standard deviation (RSD of three successive analyses; concentration of each analyte was 25.0 mg l21). cReproducibility was determined for peak areas and peak heights by calculating the RSD of three analyses performed on three different days; the concentration of each analyte was 100 mg l21.dThe concentration range was 5±100 mg l21 for As(V), MA, AB, AC, TMI and DMA, and 10±100 mg l21 for TMAO. J. Anal. At. Spectrom., 1999, 14, 1853±1859 1857the veriÆcation of the anionic arsenic compounds requires the use of other methods and different chromatographic systems with the standard additions technique, before Ænal certiÆcation.Conclusions High power nitrogen-MIP-MS coupled with HPLC was examined as an element-speciÆc detector for its applicability to the determination of cationic and anionic arsenic compounds. The MIP is a promising alternative ion source for mass spectrometry for elemental speciation analysis. The detection limits, repeatability and reproducibility of the system are adequate for the determination of arsenic compounds. The background species found in the nitrogen plasma during the nebulization of aqueous solutions are not as complex as those found with the argon-ICP27 and are absent above m/z 45.18,19 Additionally, a high-purity nitrogen gas supply minimizes argon contamination and is easily obtained from liquid nitrogen boil-off.Moreover, the gas running cost of the N2- MIP-MS is lower than that of the Ar-ICP-MS. On the other hand, the high power nitrogen-MIP is more robust than the Ar- ICP. Hence, the high power N2-MIP-MS coupled with HPLC can be used as an element-speciÆc detector for elemental speciation and successfully used for arsenic species analysis. Limited information is available on elemental speciation using HPLC-N2-MIP-MS. To our knowledge, this is the Ærst report of the coupling of HPLC with MIP-MS for the determination of arsenic compounds.This coupling system may also be applicable to other elemental speciation analysis. Determination of arsenic in urine by ICP-MS is complicated by the formation of the argon chloride (40Ar35Clz) molecular ion, which overlaps with monoisotopic As at m/z 75.Several researchers have endeavored to eliminate or correct 40Ar35Clz formation.30±32 With N2-MIP-MS, the argon-related polyatomic interference is eliminated. No detectable interference in the m/z 70±80 region is observed. The chloride interference as 40Ar35Clz, arising from the urine matrix, in ICP-MS has been overcome. Hence, N2-MIP-MS should be a very useful detector for the determination of arsenic compounds in samples with high chloride concentrations (10 000 mg l21). The procedure is very promising and reliable and was successfully applied to the determination of AB in urine.The sample used here has been distributed to many organizations and will be issued as a certiÆed reference material by our institute, for a collaborative study to certify the arsenic species, especially AB. Although the accuracy of the AB value obtained was conÆrmed by HPLC-ICP-MS, the accuracy of the values for the anionic arsenic species in urine has not yet been established as collaborative work is still in progress.Acknowledgements The authors gratefully acknowledge Dr. Sukti Hazra for helpful comments during the preparation of the manuscript, Mr. Minoru Yoneda for data transformation and JISTEC and STA, Japan, for Ænancial support. Amit Chatterjee also acknowledges Mrs. Shizuko Kinoshita for her technical support and help. Fig. 5 Calibration graphs and variation of the ion signals with concentration for the seven arsenic compounds [As(V), MA, DMA, AB, AC, TMAO and TMI] using HPLC-N2-MIP-MS (conditions as in Fig. 1). Fig. 6 Dependence of the RSD (%) on the concentration of the different arsenic compounds [As(V), MA, DMA, AB, AC, TMAO and TMI] in HPLC-N2-MIP-MS (conditions as in Fig. 1). Fig. 7 Background mass spectra in the range m/z 5±100 when Milli-Q water, 20 mM pyridine and 10 000 mg l21 chloride solutions are introduced into the optimized plasma directly (conditions as in Table 1).Fig. 8 Chromatogram obtained for the reconstituted NIES Candidate CRM-18 Freeze-Dried Human Urine on a Supelcosil LC-SCX cationexchange column with N2-MIP-MS; 50 ml urine injected (conditions as in Fig. 1). 1858 J. Anal. At. Spectrom., 1999, 14, 1853±1859References 1 A. Chatterjee, Sci. Total Environ., 1999, 228, 25 and references cited therein. 2 X. C. Le and M. Ma, Anal. Chem., 1998, 70, 1926. 3 J. Mattusch and R.Wennrich, Anal. Chem., 1998, 70, 3649. 4 E. H. Larsen, G. Pritzl and S. H. Hansen, J. Anal. At. Spectrom., 1993, 8, 1075. 5 X. C. Le and M. Ma, J. Chromatogr. A, 1997, 764, 55. 6 Y. Shibata and M. Morita, Anal. Sci., 1989, 5, 107. 7 S. A. Pergantis, W. Winnik and D. Betowski, J. Anal. At. Spectrom., 1997, 12, 531. 8 H. Greschonig, M. G. Schmid and G. Gubitz, Fresenius' J. Anal. Chem., 1998, 362, 218. 9 A. Chatterjee and A. L. Mukherjee, Sci. Total Environ., 1999, 225, 249. 10 A. Chatterjee, D. Das, B. K. Mandal, T. R. Chowdhury, G. Samanta and D. Chakraborti, Analyst, 1995, 120, 643. 11 A. Chatterjee, D. Das and D. Chakraborti, Environ. Pollut., 1993, 80, 57. 12 Gy. Heltai, T. Jozsa, K. Poresich, I. Fekete and Zs. Jarr, Fresenius' J. Anal. Chem., 1999, 365, 487. 13 R. I. Botto, J. Anal. At. Spectrom., 1993, 8, 51. 14 X.-C. Le, W. R. Cullen and K. Reimer, Appl. Organomet. Chem., 1992, 6, 161. 15 K. Oishi, T. Okumoto, T. Iino and M. Koga, Spectrochim. Acta, Part B, 1994, 49, 901. 16 M. Ohata, T. Ichinose and N. Furuta, Anal. Chem., 1998, 70, 2726, and references cited therein. 17 M. Ohata and N. Furuta, J. Anal. At. Spectrom., 1997, 12, 341. 18 Y. Okamoto, Anal. Sci., 1991, 7, 283. 19 Y. Okamoto, J. Anal. At. Spectrom., 1994, 9, 745. 20 F. Leis and J. A. C. Broekaert, Spectrochim. Acta, Part B, 1984, 39, 1459. 21 J. Yoshinaga, T. Shirasaki, K. Oishi and M. Morita, Anal. Chem., 1995, 67, 568. 22 T. Nakahara and N. Takeuchi, Anal. Sci., 1997, 13, 13. 23 T. Shirasaki, J. Yoshinaga, M. Morita, T. Okumoto and K. Oishi, Tohoku J. Exp. Med., 1996, 81. 24 J. Yoshinaga, Y. Shibata, T. Horiguchi and M. Morita, Accredit. Qual. Assur., 1997, 2, 154. 25 W. Goessler, A. Rudorfer, E. A. Mackey, P. R. Becker and K. J. Irgolic, Appl. Organomet. Chem., 1998, 12, 491. 26 W. L. Shen, T. M. Davidson, J. T. Creed and J. A. Caruso, Appl. Spectrosc., 1990, 44, 1003. 27 W. L. Shen, T. M. Davidson, J. T. Creed and J. A. Caruso, Appl. Spectrosc., 1990, 44, 1011. 28 A. Chatterjee, Y. Shibata, J. Yoshinaga and M. Morita, Anal. Chem., submitted for publication. 29 Y. Shibata and M. Morita, Anal. Chem., 1989, 61, 2118. 30 C. J. Amarasiriwardena, N. Lupoli, V. Potula, S. Korrick and W. Hu, Analyst, 1998, 123, 441. 31 F. Laborda, M. J. Baxter, H. M. Crews and J. Dennis, J. Anal. At. Spectrom., 1994, 9, 727. 32 M. J. Campbell, C. Demesmay and M. Olle�, J. Anal. At. Spectrom., 1994, 9, 1379. Paper 9/05078G J. Anal. At. Spectrom., 1999, 14, 1853±1859 18