摘要:

Automated On-line Preconcentration for Trace Metals Determination in Water Samples by Inductively Coupled Plasma Mass Spectrometry ANA PAULA PACKER, MARIA FERNANDA GINE� *, CARLOS E. S. MIRANDA AND BOAVENTURA F. DOS REIS Centro de Energia Nuclear na Agricultura, Universidade de Sa�o Paulo CENA-USP, C. Postal 96, 13400–970, Piracicaba SP, Brasil A preconcentration flow system using the cationic resin through both columns in sequence. Under these conditions, the eluants are merged stepwise forming an analyte zone AG50W-X8 for multielement determinations of trace constituents in water samples to work on-line with ICP-MS is downstream with minimal dispersion.The cationic resin AG50W-X8 was employed to preconcentrate Cd, Pb, Ni, Cu described. The flow network comprises two small resin columns inserted in parallel channels of the injector port. and Zn in water samples. Sample loading onto the columns occurs simultaneously and independently. Elution takes place intermittently through each EXPERIMENTAL column, releasing eluent fractions containing similar concentrations of the analytes, which are merged downstream.Apparatus The volume of eluent in the alternate aliquots sent through the The main flow system components were a peristaltic pump columns was adjusted to attain precise transient signals. with six channels (mp-13R Ismatec), an automatic injector Enrichment factors from 10 to 95 for sample loading times of described elsewhere,11 three-way solenoid valves (NResearch 0.5 to 5 min were attained.The eluted band produced a 161T031, Stow, MA, USA), Y-shaped connectors, Tygon pump- transient peak with width at half peak maximum of ing tubing and polyethylene tubing (0.86 mm id). An interface approximately 20 s, enough to scan several masses for card, PCL-711 (American Advantech, San Jose, CA, USA), multielement determinations by ICP-MS. The feasibility of the installed in the 486 microcomputer software written in proposed system was demonstrated by determination of QuickBASIC 4.5 was employed to control the solenoid valves Cd, Pb, Ni, Cu and Zn in the certified reference water and the pump synchronization device already described.12 sample SLRS-1. Accurate results were obtained using a The columns were constructed by drilling two separate holes preconcentration factor of 30 fold and a sample throughput of of 4 mm id and 10 mm long, giving a 126 ml internal volume, 20 samples per hour.in a central Perspex block. Two thin Perspex blocks were Keywords: Inductively coupled plasma mass spectrometry; attached to the central part using a polyester screen (400 mesh) preconcentration; water analysis; trace elements; flow system covering the column ends to avoid losses of resin. A thin rubber film was glued at the sides of the blocks which faced the screen. Polyethylene tubing was introduced into small The insertion of solid phase ion-exchangers in flow injection holes (0.9 mm id) in the external blocks to connect the columns (FI) systems coupled on-line to atomic spectrometers has been to the injector port.The three blocks were attached by screws widely described since the paper of Olsen et al.1 Most papers as described earlier.13 The wetted resin was loaded inside the reported the preconcentration of elements such as Cd, Pb, Ni, columns by pushing with a syringe from one end, the other Zn and Cu from dierent sample matrices using resins such as end being closed.The flow system was connected to the Chelex-100,1 AG50W-X8,2 I-8-HOQ3 and others.4–6 Direct concentric nebulizer (Meinhard T-2c, Meinhard, Santa A CA, determinations by ICP-MS at the ng l-1 level for most elements USA) of the ICP-MS instrument using 20 cm polyethylene is not feasible and enrichment factors higher than 10 were tubing (0.86 mm id) from the Y-shaped connector. The reported for Cd and Pb determination in some water samples.6 measurements were made on a VG Plasma Quad II Model Desirable characteristics for multielement on-line preconcen- PQII (VG Instruments, Stamford, CT, USA) under the follow- tration are an ecient ion-exchanger for several elements and ing conditions: plasma argon flow-rate, 12.0 l min-1; nebulizer the possibility of attaining high enrichment factors in a short argon flow-rate, 0.5 l min-1; auxiliary argon flow-rate, time.7 The eciency of the ion-exchange resin depends on its 1.2 l min-1; rf power, 1.35 kW; and reflected power, <3W.specificity, column packing and dimensions. Furthermore, for The sampler and skimmer nickel cone apertures were 1.1 and a flow preconcentration system the eciency is related to the 0.9 mm, respectively. A water cooled spray chamber at 10°C compromise between the resin particle size and the flow was used. parameters to avoid back-pressure.8 Another feature is to increase the enrichment factors by using long loading periods9 to the detriment of sample throughput.The simultaneous Reagents and Solutions sample loading of two columns in a flow system to increase the sample throughput was described in an earlier paper.10 The ion-exchange resin AG50W-X8 (200-400 mesh) from Bio- Rad Labs. (Richmond, CA, USA) was employed. Suprapur Sequential elution produced two repeated signals for each sample with a time delay close to that necessary for one HCl (Merck, Darmstadt, Germany) was used as eluent. The multielement stock solutions were from SPEX prepared for sample peak.Here, it is proposed that a flow system using two columns ICP-MS. Working solutions from 0.010 to 10.00 mg l-1 in 1% v/v HNO3 were prepared daily. High purity water and nitric inserted in parallel streams be coupled on-line with an ICP-MS instrument. The sample is loaded simultaneously onto the acid (sub-boiling distilled) were always used. The certified riverine reference water SLRS-1 National columns, as in the earlier proposal,10 but the elution occurs Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 (563–566) 563Research Council of Canada, (NRCC) Ottawa, Canada was pathway. Immediately, a set of electrical pulses is sent to valve V1 , switching it on/o. As a consequence, small aliquots of the analyzed to assess accuracy. eluent solution E are directed in sequence to each column. The eluted analytes are sequentially merged at junction Y while Procedure being carried to the nebulizer.Simultaneously with the injector movement, the acquisition procedure at the ICP-MS instru- Preconcentration of analytes was achieved by employing the ment is initiated, programming 35 s of uptake and 5 s of flow system shown in Fig. 1. The sample S is loaded indepenmeasurement. Quantification is made using peak area with dently and simultaneously into columns C1 and C2, while the blank subtraction. eluent solution E passes through the injector to the ICP.After a previously decided time delay, the central part of the injector is moved to the alternative position and the eluent solution is RESULTS AND DISCUSSION pumped through the columns in the reverse flow direction. The system was designed with symmetrical dimensions and The eect of the flow-rate on the ion-exchange process was determined experimentally for Cd and Pb by evaluating analyte flow-rates, with the aim of merging the eluted analytes from each column at the confluent point Y while they are being losses in the euents.A volume of 10 ml of a solution containing 200 mg l-1 of each element was pumped at flow- transported to the plasma. A more advanced flow system for preconcentration by rates from 2.0 to 10.0 ml min-1 and the euents were analyzed. In this way, a constant input of 2.0 mg of each element was intermittent analyte elution is depicted in Fig. 2. Three computer controlled solenoid valves are used in combination with introduced into the columns under dierent hydrodynamic conditions.Less than 1% of analytes were lost for flow-rates the injector port. The lines inside the valve symbols indicate the free flow pathway when the valves are switched o (solid) up to 6.0 ml min-1. For flow-rates between 6 and 10 ml min-1 a linear increase to 3% on losses was observed. In addition, or on (dashed). Initially, water is added instead of sample S and the eluent solution at E. The valves are switched on incrng the flow-rate increases the back pressure on the columns and leakage may occur.Thus, as a good compromise, simultaneously for 1 min to displace air bubbles from the system. The analytical process starts with all the valves o and the loading flow-rate was set at 4.0 ml min-1. After loading the sample, the elution parameters were the injector position as shown in Fig. 2. In this situation, the sample placed at S is recycled through V2 and V3 to the adjusted to attain the most ecient removal of the analytes from the resin.The elution was performed in reverse flow reservoirs R, while the eluent solution E is flowing to the ICP. The sample loading occurs when valves V2 and V3 are switched mode (direction of E contrary to that of S in Figs. 1 and 2) to prevent further dispersion due to dead volume and resin on together. After the sampling step, the injector is moved to the alternative position, placing the columns on the eluent squeezing inside the column.During elution, the isotopes were monitored separately in the single ion mode to obtain the respective transient signals. Nitric acid and solutions of less than 2.0 M hydrochloric acid failed to completely remove the elements from the resin; subsequently, 2.0 M HCl solution was employed throughout the experiments. The eect of the eluent flow-rate on the peak height at masses 114 for Cd and 208 for Pb are shown in Fig. 3. The single-ion monitoring allowed the observation of the elution profiles of the analytes under dierent flow conditions.As expected, the lower flow-rates led to higher peaks for the more strongly retained cation, in this case Pb. Increasing the flow-rate to over 1.1 ml min-1 resulted in the Pb peaks flattened and enlarged while those of Cd became higher and narrower up to 2.0 ml min-1. When both analytes were eluted at 1.5 ml min-1 the peak heights observed were similar, but considering the isotopic abundance of the Fig. 1 Schematic diagram of a dual column flow on-line preconcentration manifold. Black arrows represent sites where pumping is monitored isotopes (28.73 and 52.4%, respectively) this means applied for solutions of sample (S) and eluent (E). The three rectangles that the sensitivity for Cd was 1.82 times higher than for Pb. represent the injector. C1 and C2 are the ion-exchange columns For each element, peaks eluted at dierent flow-rates presented installed in the injector port shown in the sampling position.The big the same peak area, demonstrating good recovery. As a arrow below the injector indicates the movement of the central part. compromise, the elution flow-rate used in subsequent experi- Y is a connector. W indicates the waste. ments was 1.5 ml min-1. In addition, at this flow-rate the concentric nebulizer worked eciently. Fig. 2 Schematic diagram of a dual column flow on-line preconcen- Fig. 3 Eect of the elution flow-rate on the 114Cd and 208Pb signals.tration manifold with alternated elution. V1 to V3 are solenoid valves. R are reservoirs. The other components and symbols are the same as A 25 mg l-1 solution of Cd and Pb was loaded at 4.0 ml min-1 during 1 min through each column. in Fig. 1. 564 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12The resin columns used in the flow systems depicted in was made possible by switching on/o the solenoid valve with delays from 0.1/0.1 to 1.0/1.0 s.For the solenoid valves used Figs. 1 and 2 were constructed to be as similar as possible and filled with the same mass of resin. The similarity is irrelevant in this work, delays lower than 0.1 s were considered not precise.12 Peaks in Fig. 5 were obtained individually by using in this proposed method, once both eluted solutions are mixed at the junction, but the analytical path between the columns eluent aliquots from 2.5 to 25 ml to elute Cd (10.0 mg l-1) preconcentrated at the columns of the system shown in Fig. 2.and the confluent point was adjusted (to ca. 5 cm) to produce a perfect merging of the eluted analytes. In both systems, the For a better comparison, all the peaks were superimposed, indicating that diminishing the eluent aliquot volume and same eluent flow-rate of 1.5 ml min-1 was used. In the system shown in Fig. 1, after the confluent point the zone containing increasing the number of aliquots results in better sensitivity.In other words, at the confluent point improvement was the analytes was aected by dispersion while being carried to the nebulizer at a flow-rate of 3.0 ml min-1. The peak profiles observed in combining small slices of similar concentrations along the eluted zone. The dispersion is minimized when the obtained by using one or two columns in the flow system of Fig. 1 are shown in Fig. 4. The peak profile using two columns gradient concentrations of the added solutions are lessened and this also reflects more stable peaks (Table 1).Thus, by was enhanced by approximately 20% in peak height and 58% in peak area. Using the system of Fig. 2, switching on/o valve increasing the frequency of the solenoid valve action an improvement in sensitivity was attained (Fig. 5). The eective V1 resulted in alternate small aliquots of the eluent solution being directed through the columns and after the confluent volumeof eluent was approximately 0.5 ml, as can be calculated from the peak profiles in Figs. 4–6, considering the flow-rates point the flow-rate remained at 1.5 ml min-1. This flow pattern produces a stop-go situation inside each column, duplicating and the peak width. In an earlier paper, McLaren et al.14 reported that analytes were apparently eluted in less than 1 ml. the residence time of the eluent and improving the removal of analytes. The introduction into the columns of a defined Peak profiles obtained by eluting each column separately and by the intermittent approach using on/o time delays of volume of eluent (1.5 ml) in aliquots varying from 2.5 to 25 ml 0.1/0.1 s on valve V1 are depicted in Fig. 6. The two signals on the left are a result of sequential elution on the columns. Valve V1 (Fig. 2) sent the eluent through one column during 30 s and then to the other column. Clearly the columns were dierent (peaks A and B). The peak on the right (Fig. 6) was obtained by eluting both columns using time delays of 0.1/0.1 s on valve V1.The sample volume loaded onto the two columns was twice that of each column separately, which was reflected in the peak height enhancement. Considering the peak area, the results using the two columns approach produced a threefold enhancement of the peak area obtained for each column. The resulting signal demonstrated the real advantage of this configuration, in not only increasing the sensitivity but also Fig. 4 Peak profiles obtained by preconcentration using the system enlarging the peak, allowing the sequential scanning of analytes depicted in Fig. 1. Peaks A and B were obtained using one and two at the peak maximum for approximately 5 s. columns, respectively. A 10 mg l-1 solution of Cd was loaded at The enrichment factor defined by the ratio between the 4.0 ml min-1. The elution flow-rate was 1.5 ml min-1 . concentration of the analyte in the eluted solution and in the original solutions was calculated from the transient eluted Fig. 5 Peak profiles obtained by introducing the elution solution as Fig. 6 Signals obtained by preconcentration in each column separately and by using the system shown in Fig. 2. The sample loaded in alternating aliquots of dierent volumes. Peaks from left to right were obtained by switching on and o valve V1 in delays of 0.1/0.1; 0.2/0.2; each column was 10 mg l-1 of Pb at a flow-rate of 4.0 ml min-1. Peaks A and B were obtained by elution of each column at 1.5 ml min-1 . 0.3/0.3; 0.5/0.5; and 1.0/1.0 s during elution. A 10 mg l-1 solution of Cd was loaded at 4.0 ml min-1. The elution flow-rate was Peak C was obtained by alternated elution of both columns switching on/o valve V1 in 0.1/0.1 s. 1.5 ml min-1. Table 1 Precision of results obtained by alternate elution in aliquots generated by switching valve V1 on/o with dierent delays. Results in 104 counts s-1 obtained with a solution of 10.0 mg l-1 Cd V1 on/o/s 0.1/0.1 0.2/0.2 0.3/0.3 0.5/0.5 1.0/1.0 Volume/ml 2.5 5.0 7.5 12.5 25 104 counts s-1 9.43±0.15 8.60±0.16 7.76±0.16 7.3±0.16 6.13±0.14 RSD (%)* 1.59 1.86 2.06 2.19 2.28 * n=7.Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 565Table 2 Results for the certified reference material SLRS-l (riverine water)* Ni Cu ZN Cd Pb Isotope 60 65 66 114 208 Certified conc. mg l-1 1.07±0.06 3.58±0.30 1.34±0.20 0.015 ±0.002 0.106±0.011 This work conc. mg l-1 1.06±0.09 3.52±0.17 1.30±0.10 0.015 ±0.001 0.105±0.009 * Mean of the determinations and standard deviations of 10 replicates.continuous merging of the eluted analytes. The preconcentration in the flow system using two ion-exchange resin columns inserted in parallel was more ecient than that using a single column. By programming the sample loading time dierent enrichment factors are easily achieved. The determination of Ni, Cu, Zn, Cd and Pb in water samples was achieved by preconcentration using AG50W-X8 resin in a flow system coupled to an ICP-MS instrument. The system is appropriate for routine work as it allows throughput of 20 samples per hour.The proposed flow configuration produced transient signals large enough to determine five elements by ICP-MS using the peak hopping acquisition mode with good precision. Fig. 7 The dependence of enrichment factors on the sampling loading time. The curve was obtained using 10 mg l-1 of Pb at a flow-rate of Support from Fundac�a� o de Amparo a` Pesquisa do Estado 4.0 ml min-1 and eluting at 1.5 ml min-1.de Sa�o Paulo (FAPESP) and Conselho Nacional de Desenvolvimento Cientý�fico e Tecnolo�gico (CNPq) is greatly signal compared with the signal obtained by continuous nebul- appreciated. ization. Signal peak enhancements of 16–17 fold were attained for Cd and Pb when employing a sampling time of 60 s and REFERENCES an acquisition measurement time of 5 s. The preconcentration factor observed for dierent sample loading periods is pre- 1 Olsen, S., Pessenda, L.C. R., Ruzicka, J., and Hansen, E. H., sented in Fig. 7. The flexibility of programming dierent sam- Analyst, 1983, 108, 905. pling loading times to attain dierent enrichment factors is 2 Porta, V., Abollino, O., Mentasti, E., and Sarzanini, C., J. Anal. At. Spectrom., 1991, 6, 119. one of the advantages of the proposed method. 3 Beauchemin, D., and Berman, S. S., Anal. Chem., 1989, 61, 1857. For routine analysis of water samples, the proposed method 4 Fang, Z.L., Guo, T. Z., and Welz, B., T alanta, 1991, 38, 613. was used to determine the isotopes 58Ni, 63Cu, 68Zn, 114Cd and 5 Bloxham, M. J., Hill, S. J., and Worsfold, P. J., J. Anal. At. 208Pb. Simultaneous determination of the isotopes was carried Spectrom., 1994, 9, 935. out using the peak hopping mode. The dwell time for each 6 Porta, V., Sarzanini, C., Mentasti, E., and Abollino, O., Anal. isotope was adjusted according to its abundance, thus 13.54, Chim.Acta, 1992, 258, 237. 7 Fang, Z., Flow Injection Separation and Preconcentration, New 13.41, 18.58, 17.53 and 15.12 ms was used for the above York, 1993. mentioned isotopes, respectively. These values were calculated 8 Marshall, M. A., and Mottola, H. A., Anal. Chem., 1985, 57, 729. from the default 10.24 ms for a 100% abundance. Results for 9 Tyson, J. F., Spectrochim. Acta Rev., 1991, 14, 169. the analysis of the water reference material SLRS-1 are pre- 10 Fang, Z. L., Ruzicka, J., and Hansen, E. H., Anal. Chim. Acta, sented in Table 2. Agreement between the certified and found 1984, 164, 23. values was assessed by the paired t-test at 95% confidence 11 Bergamin, Fo, H., Reis, B. F., Jacintho, A. O., and Zagatto, E. A. G., Anal. Chim. Acta, 1980, 117, 81. level (tcalculated=0.46 and ttab=2.77). These data were obtained 12 Reis, B. F., Gine�, M. F., Zagatto, E. A. G., Lima, J. L. F. C., and using an enrichment factor of 30, consuming 16 ml of sample Lapa, R. A., Anal. Chim. Acta, 1994, 293, 129. and allowing a throughput of 20 samples per hour. This factor 13 Reis, B. F., Gine�, M. F., Santos, Fa, M. M., and Baccan, N., was necessary to determine low concentrations of Ni, Cu, Zn, J. Braz. Chem. Soc., 1992, 3, 80. Cd and Pb. 14 McLaren, J. W., Lam, J. W. H., Berman, S. S., Akatsuka, K., and Azeredo, M. A., J. Anal. At. Spectrom., 1993, 8, 279. CONCLUSIONS Paper 6/07251H Received October 24, 1996 The flow system with alternate introduction of eluent by Accepted January 29, 1997 the columns gave a better performance than that using the 566 Journal of Analytical Atomic Spectrometry, May

ISSN:0267-9477    

DOI:10.1039/a607251h

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Certification of the Rubidium Concentration in Water Materials for the International Measurement Evaluation Programme (IMEP) Using Isotope Dilution Inductively Coupled Plasma Mass Spectrometry ULF O� RNEMARK, PHILIP D. P. TAYLOR AND PAUL DE BIE` VRE Institute for Reference Materials and Measurements, European Commission-JRC, B-2440 Geel, Belgium This paper describes the certification of rubidium in synthetic inability of analytical chemists to assign a realistic uncertainty to their measurements.By oering SI-traceable values, and natural water samples for Round 6 of the Institute for Reference Materials and Measurements International accompanied by combined uncertainties, for toxic and essential elements in, e.g., environmental matrixes, the IMEP provides Measurement Evaluation Programme (IMEP). The analytical procedure is based on isotope dilution ICP-MS as the primary important support for field laboratories in monitoring the degree of reliability of their chemical measurements.method of measurement. Ion exchange chromatography was used to eliminate the isobaric interference on 87Rb from 87Sr. In recent years rubidium has received some attention owing to the possibility of its essential properties. Used in semicon- By using the adopted procedure, SI-traceable values with expanded uncertainties (U=2u c) of 2% were obtained for ductor technology, as medicament and catalyst, this element, however, lacks the industrial and biochemical importance rubidium at nmol kg-1 (low mg kg-1) levels.Results from 42 field laboratories participating in IMEP-6 are evaluated associated with lighter alkali metals. Mass spectrometric measurements of rubidium are found in geochronology6–9 (the against the certified values. Rb–Sr rock dating method) and its relative atomic mass has Keywords: Rubidium; isotope dilution ; inductively coupled been determined on an ‘absolute’ basis.10 Ecient separation plasma mass spectrometry ; traceability; water methods are required, in radiochemical applications and nuclear safeguards,11,12 to handle products of competitive The International Measurement Evaluation Programme nuclear reactions or for separation of daughter elements and, (IMEP), which has been in existence since 1988, aims at in mass spectrometry,6 to overcome the problem of the isobaric establishing representations of laboratory measurement results interference from 87Sr.The average concentrations of rubidium against SI-traceable values, thus oering measurement labora- in ocean water and rivers are 1.3 and 0.6 mmol l-1 (110 tories a tool for improving the reliability of chemical measure- and 50 mg l-1), respectively, whereas the strontium levels are ments as and where required.IMEP is run from IRMM approximately two orders of magnitude higher.13 Early data, (Institute for Reference Materials and Measurements), Geel, summarised by Smales and Salmon,14 showed large continental Belgium, in co-operation with NIST (National Institute of variations for rubidium concentrations, ranging from 0.4 Standards and Technology), Gaithersburg, MD, USA and to 40 mmol l-1 (35–5700 mg l-1).Groundwater contains under the auspices of IUPAC (the International Union of Pure <1.2 mmol l-1 (100 mg l-1)15 and as shown by Wyttenbach,16 and Applied Chemistry), EURACHEM (a Focus for Analytical the concentrations of rubidium in this and other types of Chemistry in Europe), EUROMET (the Association for freshwater may amount to only a few mg l-1 (low nmol l-1).European Institutes for Metrology) and CITAC (Co-operation Such levels are, however, readily determined using quadruon International Traceability in Analytical Chemistry). A more pole ICP-MS. detailed description of the programme can be found in refs. 1 In IMEP-6, (the complete evaluation of this round is schedand 2. uled to be published in 1997) certified or, in some cases, The concept of ‘traceability’ has been developed to ensure assigned values were obtained for rubidium and 13 other trace that an uninterrupted chain of links (traceability chain) leads elements in a synthetic and in a natural water, mainly using securely to a common base or primary standard of measure- isotope dilution mass spectrometry (IDMS).The present paper ment. In this chain, all links should have stated uncertainties.3 focuses on the certification of rubidium in the two water A EURACHEM document4 (based on ISO guidelines) explains materials involved.The analytical procedure is based on ID the approach of combined (total) uncertainty, uc, taking into in combination with ICP-MS. Cation exchange chromatogaccount all known sources of uncertainty and visualising these raphy is used to eliminate the isobaric interference on 87Rb in the form of an uncertainty budget. Whilst there are estab- from 87Sr. The various sources of uncertainty present in the lished systems for traceability to the realisation of base units, procedure are visualised in the form of an uncertainty budget.e.g., the kilogram and the metre in the SI system, this is still Results reported by 42 field laboratories are evaluated against not the case for amounts of substance measurements.5 If results the certified SI-traceable values. from chemical measurement cannot be compared on the basis of a consistent traceability system, the consequences in terms of cost can be immense because of superfluous or repeated EXPERIMENTAL measurements and loss of confidence in, e.g., international Instrumentation and Equipment trade and commerce, environmental regulations and medical diagnosis.The lack of comparability, presently existing in A VG PlasmaQuad 2+ ICP-MS instrument (VG Elemental, Winsford, UK) was used for measuring the isotope amount many chemical measurements, can partly be explained by the Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 (567–572) 567ratio n(87Rb)/n(85Rb) and for controlling the eciency of the aration involved sampling, stabilising with nitric acid, ultrafiltration, sterilisation and bottling in acid-cleaned poly- rubidium–strontium separation. The instrument is equipped with a secondary electron multiplier for which the dead time ethylene containers. The concentrations of strontium were approximately 300 and 100 mg l-1 in the synthetic and natural was experimentally determined to be 27±1 ns using the SRM 982 Lead Metal (NIST, Gaithersburg, USA).Instrumental water, respectively. The water samples were stored refrigerated (+4°C) protected by double layers of polyethylene. Prior to parameters were: sample aspiration rate, #1 ml min-1; forward power, 1400 W; plasma gas flow,#14 l min-1; intermedi- analysis, the bottles were conditioned in a class 100 environment where the gravimetric work was performed. In-house ate gas flow, #1.5 l min-1; nebulizer (V-groove type); aerosol carrier gas flow, #0.8 l min-1; data acquisition mode, peak calibrated weights were used and masses were corrected for buoyancy.jumping, three points per peak; dwell time, 10.24 ms; and acquisition time, 6×30 s. The rubidium concentration levels measured were kept below 10 mg l-1 (0.1 mmol l-1). The sensi- Sample Preparation tivity, as measured for 115In, was normally 3×104 s-1 mg-1 l. For most experimental work, and for protection of solutions Approximately 5 g of sample and 1 g of spike (containing and reagents, a class 100 (<100 0.5 mm particles ft-3) laminar #600 and 300 pmol, respectively, of rubidium) were accurately flow bench equippedwith anHEPA-filter was used (Vermeulen, weighed into a 50 ml quartz beaker.The nitric acid matrix L.J., Antwerp, Belgium). Polyethylene bottles were used for was removed by gentle evaporation to dryness. When the Milli-Q water and other solutions were stored in Teflon FEP beaker had cooled, the walls were rinsed with a few ml of bottles (Nalge, Rochester, NY, USA).Evaporation of water Milli-Q water followed by a second evaporation step. The samples was performed in quartz beakers placed on a ceramic residue was allowed to dissolve in 5 ml of Milli-Q water for at hot plate inside a glovebox to avoid contamination. Ion least 30 min. exchange columns were prepared by packing the resin in0 ml Poly-Prep plastic tubes (Bio-Rad, Nazareth, Belgium).Separation of Rb from Sr Solutions and water samples were transferred using 5 ml Eppendorf Combitips (Merck-Belgolabo, Overijse, Belgium) The sample solution and 5 ml of Milli-Q water, used to rinse and 30 ml quartz beakers. Bottles, pipette tips and beakers the beaker, were allowed to pass through a preconditioned were cleaned using 3% (m/m) hydrochloric acid (pro analysi ) column. The euent was discarded and any remains on the for at least 24 h, rinsed with Milli-Q water and dried under column walls were rinsed down with a few ml of Milli-Q water.clean air prior to use. Rubidium was eluted (flow rate #1 ml min-1) using 0.50 mol l-1 HCl and the fraction between 15 and 50 ml was collected for monitoring of the induced isotope amount ratio Chemicals, Reagents and Solutions n(87Rb)/n(85Rb) using ICP-MS. The NIST SRM 984 Rubidium Chloride was used to prepare calibration solutions of rubidium with natural isotopic com- RESULTS AND DISCUSSION position.This material has an isotope amount ratio n(87Rb)/n(85Rb)=0.385 71 (15) established by use of gravi- Separation of Rb from Sr metrically preparedsynthetic isotope mixtures. A spike solution A survey of the literature revealed that most ion exchange was obtained by dilution of the IRMM-618 isotopic reference separations of rubidium and strontium, including those in material: n(85Rb)/n(87Rb)=48.7852 (86); c (87Rb)=112.13 (17) radiochemical work, have been carried out on organic resin mmol kg-1 (IRMM, Geel). These solutions, approximately types.Many of the described procedures were, however, not 0.5 mol l-1 in nitric acid, were prepared gravimetrically. suitable for the work presented here since they also involved Hydrochloric and nitric acid, prepared by subboiling distilother analytes16,17 or were not designed for trace amounts.17,18 lation of analytical reagent grade acids (Merck, Darmstadt, The procedures described adopted impractical column dimen- Germany), and water, purified in a Millipore system (Milli-Q sions requiring hours to days and large amounts of acid for water), were used in all analytical work.elution.16,19 Correspondingly poor column eciency also results in substantial dilution of the sample16,17 and if sub- Ion Exchange Resin sequent decrease of volume is required, this may require considerable time unless microwave facilities are available. The The strongly acidic polystyrene–divinylbenzene resin, Bio-Rad use of low pressure6 or high performance6,20 chromatographic AG 50W-X8 (100–200 mesh, hydrogen form) was used.Fine systems, yielding sharper elution profiles and small fractions, particles were removed by decantation. The resin was leached are therefore attractive prior to use of ICP-MS instrumentation with 4 mol l-1 hydrochloric acid for at least 24 h, rinsed with with low consumption nebulisers or thermal ionisation mass and stored in Milli-Q water.Prior to use, a 1.5 ml resin bed spectrometric (TIMS) measurements. Stability of the ion (30×8 mm) was packed by dispensing the resin into a water exchanger is vital in trace analysis. The mg amounts of carbon- filled column. A small piece of quartz wool on top of the bed containing fragments, released from degrading organic resins, kept the resin in place. The resin was washed with 10 ml of may result in significant blank contributions.21 Inorganic ion 6 mol l-1 HCl followed by 10 ml of Milli-Q water.This washing exchangers have also proved to be impractical for this reason.22 step was repeated twice and the final rinsing with Milli-Q Examples of practical rubidium–strontium separations on inor- water was stopped when the pH of the euent had reached ganic ion exchangers can be found in refs. 22 and 23. #5.5. Columns can be prepared in advance but the condition- The fundamental exchange properties of resins such as Bio- ing with HCl should be carried out shortly before the separa- Rad Type 50 were investigated several decades ago.24–26 tion since the resin itself degrades slowly, even in water, with However, surprisingly few workers take advantage of estab- the possible release of interferents.lished equilibrium data to plan their separations.11 The exchange of protons for rubidium on a resin in the hydrogen Water Samples form is given by The two water samples, an artificial simulate of drinking water Rb+(aq)+R-H+=R-Rb++H+(aq) (1) and a natural water obtained from Clear Creek, CO, USA, were prepared under the responsibility of NIST.The prep- The concentration exchange constant, E, (selectivity coecient) 568 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12attributed to this reaction is defined as determined from a single column experiment, involves a strontium- selective resin.28 The levels of rubidium and strontium in the hydrochloric ERb/H=[Rb+]R[H+]aq [Rb+]aq[H+]R (2) and nitric acid used were comparable, which is why neither this latter acid19 nor other eluents8,9,18,22,29 were tested since Concentrations are expressed in mol l-1 for the aqueous phase initial experiments with HCl worked well.Larger dierences (subscript aq) and in mol kg-1 for the stationary phase in distribution coecients are obtained from eluents composed (subscript R). The distribution coecient, KD, is defined as the of HCl–acetone29 but the acetone concentrations exceed by far ratio of rubidium in the two phases those consistent with the plasma.Moderate dierences have been reported for strong acids17,24,25 and salt solutions17,27 as KD= [Rb+]R [Rb+]aq (3) elution agents for rubidium on this resin type. However, the concentrations needed for ecient elution preclude the use of KD is related to the capacity factor, k¾, defined in eqn. (4) salts if ICP-MS is used, and precipitation of SrSO4 could where Vr, Vm and w are the retention volume (centre of elution possibly change the flow characteristics and/or trap other peak), the column dead volume and the mass of the resin, analytes, if high levels of strontium are present.The evapor- respectively. ation step, although time consuming, ensures equilibration of spike and sample, helps destroy remaining dissolved organic k¾=KD w Vm =Vr-Vm Vm (4) constituents and brings the sample to a pH where rubidium is trapped as a narrow band on the column (Fig. 1). Eqn.(4) can be rearranged to give Vr=Vm+wKD (5) Isotope Dilution KD for the rubidium distribution can be expressed in terms of IDMS using stable isotopes can play an important role in the selectivity coecient, hydronium ion concentration and establishing traceability of chemical amount measurements to resin capacity, Q, (approximately equal to [H]R for analytical the SI system5 and has been recognised as a potential primary applications) yielding method of measurement.30 The reason is its inherent characteristics, such as the direct comparison of ratios of known and unknown number of entities (e.g., atoms) and less dependency KD= ERb/HQ [H+]aq (6) on matrix eects, and its metrological properties, i.e., a fully understood and transparent measurement procedure with a Introducing this expression for KD into eqn.(5) gives the clear relationship between what is measured and what is retention volume. An analogous expression is derived for the intended to be measured with the ratios expressed in the interfering element, i.e., strontium.E-values for rubidium and correct unit for amount of substance (mol). Consequently, if strontium were taken from the work by Bonner and Smith26 the measurement including the sample preparation is carried and recalculated for exchange against hydronium ions. As out correctly, the potential for accuracy is high. shown in Fig. 1, such predictions, although approximate, are With the general IDMS equation5 applied to rubidium, the very convenient for the planning of a separation.21 concentration, cx , of this element in the sample material can The predicted retention volumes obtained for 1 mol l-1 HCl be evaluated from eqn.(7): are 11 and 101 ml for rubidium and strontium, respectively. However, owing to the considerable band broadening the HCl cx=Ry-KRB(obs) KRB(obs)-Rx cyd my mx 1+Rx 1+Ry (7) concentration should not exceed 0.5–0.6 mol l-1 for baseline separation with the elution rate used here.With 0.50 mol l-1 HCl, the observed reproducible retention volume for rubidium which takes into account total mass discrimination eects was 30 ml (predicted 20 ml). Quantitative elution of strontium during the measurement (correction factor K) and the preprequires relatively large amounts of acid, and occasionally aration (dilution factor d) of the spike solution (concentration resulted in higher backgrounds when regenerated columns cy), where my, mx , Rx and Ry are the mass of diluted spike, were used.Therefore, the columns were discarded after each mass of sample, isotope amount ratio of spike and the natural separation. In this respect the procedure designed by To� rko� isotope amount ratio, respectively. Mixing of sample and spike and Szirtes,27 where strontium is eluted prior to rubidium resulting in a blend with isotope amount ratio RB , close to as an EDTA complex, is of interest in future IDMS work. unity provides optimum ICP-MS precisions.Another possible procedure, allowing both elements to be The total mass discrimination factor, K, was determined by measuring solutions of natural rubidium, having a similar concentration to that of the sample, several times during the analysis. Each blend, RB , was then corrected for the drift in mass discrimination observed during the measurement. For larger dierences in isotope amount ratio between sample and blend, or when higher accuracy is demanded, it is recommended that the mass discrimination factor is established using solutions with matching isotopic composition.A property of isotope dilution, often pointed out in the literature, is that recoveries in various steps of the procedure need not be quantitative after equilibration of spike and sample. However, there are good reasons for keeping the recoveries reproducible and close to 100%. (i ) Limitations in sample volume and badly performed separations, requiring partial collection of elution profiles, may result in unsatisfactory signal-to-noise ratios for elements present in low concen- Fig. 1 Predicted retention volumes (ml) for rubidium and strontium trations. (ii ) Partial collection of the elution profile is also on Bio-Rad AG 50W-X8. Data used in calculation: Vm 0.001 l; w critical with respect to isotope fractionation. Oi et al.31 investi- 0.00075 kg; (dry mass) Q 5.1 eqv. kg-1; ERb/H 2.49; ESr/2H 5.15 M/(mol kg-1). gated isotope fractionation of rubidium on various ion Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 569exchangers, including the strongly acidic, highly porous resin RbNO3 solution could not be reproduced in this laboratory. A discrepancy of-1.8% was observed. Through contacts with Asahi LS-6. These workers found changes in the isotope amount ratio of 0.2–0.3%. Such discrepancies could influence the manufacturer, it was revealed that the rubidium concentration had indeed only been assayed indirectly and that metal the result in high accuracy work if only part of the elution profile is used.(iii ) There are similar risks associated with the impurities of the order of 2% could not be excluded. The levels of rubidium and strontium in the reagents could spiking process as in normal recovery tests. Adding an isotope in a dierent chemical form from that present in the sample generally not be distinguished from the instrumental background. The procedural blank was approximately 1 nmol may lead to erroneous results if spike and sample do not react similarly.For rubidium, this is not likely to be a problem. (rubidium) and reproducible. The detection limit (3u procedure blank), starting from a sample volume of 5 ml, is estimated to However, there is no documented evidence of this. The availability of a certified spike material as described be 4 pmol or 0.8 nmol kg-1 of rubidium. Since rubidium is eciently preconcentrated on the ion exchanger above pH 3, here makes the IDMS process easily understood (eqn. 7). In principle only two ratios, Rx and RB , need to be measured. lower concentrations can, at least in theory, be measured by increasing the sample volume. A six- to seven-fold dilution is Alternatively a double isotope dilution experiment can be carried out,32,33 where the spike is first characterised against a obtained when a 5 ml sample is placed on the column. However, for low analyte concentrations using flow rates of primary assay reference material using reverse IDMS, followed by a determination of the unknown concentration in the 1 ml min-1, an increase in sample size to 35 ml does not constitute a problem.For most freshwater samples, this loading sample using normal IDMS. This requires the measurement of four ratios, Rx, Ry, and those of two blends RB1 and RB2, will only occupy a small fraction of the resin. but no correction factor has to be applied to the ratios and the concentration of the spike need not be accurately known.Contamination Study Special precautions were undertaken in the above procedure Analytical Procedure and its Figures of Merit to avoid contamination. In parallel experiments, accurate results without a significant increase of the blank level were, The recovery of rubidium in the above procedure was 102±7% however, obtained outside the clean bench. Also, metal- (n=7). This was confirmed for solutions of rubidium in nitric containing adjustable pipettes (avoided above for fear of acid, the IMEP-6 samples and for a matrix consisting of 23 corrosion products) and disposable tips (without previous elements derived from the Merck IV multi-element calibration cleaning) proved not to constitute any problem and would solution (Merck, Darmstadt, Germany).The recovery was the simplify routine work. From the point of view of contam- same for a 200–400 mesh size resin which, in addition, showed ination and low blank levels, closed HPLC systems have a slight sharpening of the elution profile.proved valuable in TIMS applications.6,20 The overall performance of the procedure was checked by analysing standard solutions of natural rubidium (NIST SRM 984 Rubidium Chloride) against the IRMM-618 spike. Certified Values and Estimation of Uncertainties Analyses on dierent occasions over a five month period showed an agreement of 0.10±0.30% (average and standard The results of the certification are 138.7 (28) and 23.46 (47) nmol kg-1 for the synthetic and natural water, respectively uncertainty, n=10) between the two materials atsimilar analytical concentrations (nmol kg-1 levels).The reproducibility in (Tables 1 and 2). The combined uncertainty, uc, for the analyte (cx ) expressed IDMS experiments on the same sample was 0.2–0.3%. Rubidium chloride, owing to its stoichiometric properties, in terms of the measured parameters in eqn. (1) was obtained using an approximate numerical method of dierentiation.4 is preferred to the more hygroscopic RbNO3 .8,34 While metrologically prepared solutions of the SRM 984 Rubidium The uncertainty of the verification of the absence of bottle to bottle variation was then added linearly to that uncertainty.Chloride were confirmed by measurements against the spike material, the nominal value of 1000±2 mg l-1 for a commercial The uncertainty from subtraction of the procedural blank was Table 1 Rubidium concentrations cx(Rb) as measured on individual samples of the IMEP-6 synthetic water and uncertainty budget.Uncertainties [correspondingly last digit(s) of the quoted results in parentheses] and relative uncertainty contributions are standard uncertainties u (y)=1s Relative standard Bottle No. Date (Typical) value uncertainty u(y) 6 26.1.95 138.56 180 26.1.95 138.47 190 26.1.95 139.15 283 26.1.95 138.72 100 9.2.95 139.43 321 9.2.95 137.80 Bottle to bottle variation for c(Rb)/nmol kg-1* 138.69 (56) 0.0041 Rx Natural isotope amount ratio 0.38571 (15) 0.0004 Ry Certified isotope amount ratio 48.7852 (86) 0.0006 K Mass discrimination factor 0.9571 (38) 0.0040 RB(obs) Isotope amount ratio in blend 2.3000 (70) 0.0031 Cyd Dilution of spike/mmol kg-1 0.65267 (52) 0.0008 my Mass of spike solution/g 1.0000 (3) 0.0003 mx Mass of sample solution/g 3.5000 (6) 0.0002 Coed uncertainty (uc) for cx(Rb) from eqn.(7) 0.0064 cx (Rb) Certified value† (average and expanded uncertainty, k=2) 138.7 (28) nmol kg-1 11.92 (24) mg kg-1 * Corrected for procedure blank.† Minimum uncertainty released with certified values was 2% which also covers for possible evaporation during shipping of sample material. 570 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12Table 2 Rubidium concentrations cx (Rb) as measured on individual samples of the IMEP-6 natural water and uncertainty budget. Uncertainties [correspondingly last digit(s) of the quoted results in parentheses] and relative uncertainty contributions are standard uncertainties u (y)=1s Relative standard Bottle No.Date (Typical) value uncertainty u(y) 153 26.1.95 23.69 81 9.2.95 23.50 117 9.2.95 23.26 166 9.2.95 23.46 217 9.2.95 23.43 258 9.2.95 23.40 Bottle to bottle variation for c(Rb)/nmol kg-1* 23.46 (15) 0.0062 Rx Natural isotope amount ratio 0.38571 (15) 0.0004 Ry Certified isotope amount ratio 48.7852 (86) 0.0006 K Mass discrimination factor 0.9571 (38) 0.0040 RB(obs) Isotope amount ratio in blend 3.400 (17) 0.0050 Cyd Dilution of spike/mmol kg-1 0.65267 (52) 0.0008 my Mass of spike solution/g 0.6000 (1) 0.0002 mx Mass of sample solution/g 7.5000 (15) 0.0002 Combined uncertainty (uc) for cx(Rb) from eqn.(7) 0.0078 cx (Rb) Certified value† (average and expanded uncertainty, k=2) 23.46 (47) nmol kg-1 2.008 (40) mg kg-1 * Corrected for procedure blank. † Minimum uncertainty released with certified values was 2% which also covers for possible evaporation during shipping of sample material.negligible. The detailed uncertainty budgets accompanying the oered) and in the international debate on degree of reliability and comparability of chemical measurements, depictions such certified values are shown in Tables 1 and 2. The expanded uncertainties, U, (=kuc) were approximately 1.5 and 2.0% (k= as Figs. 2 and 3 constitute a valuable source of information. There are few indications of gross errors by the participants 2) for the synthetic and natural water materials, respectively.for both water materials and most of the laboratories fall within ±30% of the certified values. This is an improvement Results from Field Laboratories Participating in IMEP-6 compared with IMEP-31,2 (1991–1992) where similar samples were analysed. Also noticeable since IMEP-3, is the substantial In Figs. 2 and 3, the certified ranges for rubidium in the two increase in the number of laboratories employing ICP-MS.water materials are displayed together with results reported However, most laboratories still fail in providing a realistic by 42 of the 180 participating laboratories (Chile and the USA uncertainty with their results. not included). The scale on the y-axis, ±50% around the SI-traceable value, is chosen for convenience and must not be interpreted as a ‘target value’ for uncertainties (which must be CONCLUSIONS set by authorities) for judgement of quality.A main feature of the IMEP is that participants can draw conclusions for Rubidium was certified at the nmol kg-1 concentration level to 2% expanded uncertainty (k=2) yielding values traceable themselves. However, for (inter)national authorities and legislators, as well as for accrediting bodies (to which IMEP is to the SI system. The work presented demonstrates the poten- Fig. 2 Certified range, U (k=2) against rubidium concentrations in the natural water reported by participating laboratories.Participants’ data arranged according to reported status of accreditation. The text boxes contain reported data which fall outside the ±50% range. Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 571Fig. 3 Certified range, U (k=2) against rubidium concentrations in the synthetic water reported by participating laboratories. Data arranged according to instrumental technique used by participants. The text boxes contain reported data which fall outside the ±50% range. 14 Smales, A. A., and Salmon, L., Analyst, 1955, 80, 37. tial for accuracy of isotope dilution ICP-MS. A simple ion 15 Freeze, R. A., and Cherry, J. A., in Groundwater, Englewood exchange procedure, based on predictions allows interfering Clis, 1979, Table 3.3, p. 85. strontium to be removed easily and rapidly. Results from a 16 Wyttenbach, A., Chimia, 1966, 20, 119. contamination study also indicate that accurate results are 17 Mehta, V.P., and Khopkar, S. M., Sep. Sci. T echnol., 1982, 17, 495. possible under such less rigorous working conditions. 18 Tsai, H.-T., Kato, T., and Oka, Y., Bull. Chem. Soc. Jpn., 1970, 43, 2823. 19 Tsuji, M., and Abe, M., Radioisotopes, 1984, 33, 218. The authors are indebted to B. Dyckmans and F. Hendrickx 20 Cassidy, R. M., and Chauvel, C., Chem. Geol., 1989, 74, 189. for assistance with the gravimetrical work, to S. Nelms for 21 O� rnemark, U., and Olin, A° ., T alanta, 1994, 41, 67.contribution to the contamination study and to L. Van Nevel 22 Mathew, J., and Tandon, S. N., J. Radioanal. Chem., 1975, 27, 315. and E. Poulsen for preparing two of the figures. 23 Jain, A. K., Agrawal, S., and Singh, R. P., J. Radioanal. Chem., 1980, 60, 111. 24 Strelow, F. W. E., Anal. Chem., 1960, 32, 1185. REFERENCES 25 Strelow, F. W. E., Rethemeyer, R., and Bothma, C. J. C., Anal. Chem., 1965, 37, 106. 1 Lamberty, A., Van Nevel, L., Moody, J.R., and De Bie`vre, P., 26 Bonner, O. D., and Smith, L. L., J. Phys. Chem., 1957, 61, 326. Accred. Qual. Assur., 1996, 1, 71. 27 To� rko�, J., and Szirtes, L., Radiochem. Radioanal. L ett., 1972, 2 Lamberty, A., Lapitajs, G., Van Nevel, L., Go�tz, A., Moody, J. R., 11, 147. Erdmann, D. E., and De Bie`vre, P., in Biological T race Element 28 EiChrom Industries, P.O. Box 3, Cupar, Fife KY7 7SE, UK and Research, ed. Schrauzer, G. N., Humana Press, 1994, p. 571. Nicholl, C., personal communication, IRMM, August 1996. 3 International Vocabulary on Basic and General T erms inMetrology, 29 Marhol, M., in Ion Exchangers in Analytical Chemistry: Wilson International Organisation for Standardisation, Geneva, and Wilson’s Comprehensive Analytical Chemistry, ed. Svehla, G., Switzerland, 1993, p. 47. Elsevier Science, New York, 1982, vol. 14, Table 8-8, p. 512. 4 Quantifying Uncertainty in Analytical Measurement, EURACHEM 30 Comite� Consultatif pour la Quantite� de Matie`re (CCQM), Report Secretariat, Teddington, Middlesex, UK, 1995. of the 1st Meeting, ed. by BIPM, Se`vres Cedex, France, 1995. 5 De Bie`vre, P., Fresenius’ J. Anal. Chem., 1994, 350, 277. 31 Oi, T., Ogino, H., Kawada, K., Hosoe, M., and Kakihana, H., in 6 Stray, H., Chem. Geol., 1992, 102, 129. New Developments in Ion Exchange Procedures: International 7 Ueda, M., Yamashita, S., and Hayakawa, T., Bull. Univ. Osaka Conference on Ion Exchange, eds. Abe, M., Kataoka, T., and Prefect., Ser. A, 1981, 30, 55. Suzuki, T., Kodansha, Tokyo, Japan, 1991, p. 299–304. 8 Pankhurst, R. J., and O’Nions, R. K., Chem. Geol., 1973, 12, 127. 32 Henrion, A., Fresenius’ J. Anal. Chem., 1994, 350, 657. 9 Eichho, H.-J., Appl. Spectrosc., 1960, 14, 74. 33 De Bie`vre, P., in T race Element Analysis in Biological Specimens: 10 Catanzaro, E. J., Murphy, T. J., Garner, E. L., and Shields, W. R., T echniques and Instrumentation in Analytical Chemistry, eds. J. Res. Natl. Bur. Stand., Sect. A, 1969, 73, 511. Herber, R. F. M., and Stoeppler, M., Elsevier, Amsterdam, 1994, 11 Vega-Rangel, E., Abou-Shakra, F., andWard, N. I., in Applications vol. 15, ch. 8. of Plasma Source Mass Spectrometry II, eds. Holland, G., and 34 Moody, J. R., Greenberg, R. R., Pratt, K. W., and Rains, T. C., Eaton, A. N., RoyalSociety of Chemistry, Cambridge, 1993, p. 103. Anal. Chem., 1988, 64, 1203A. 12 Zsinka, L., and Szirtes, L., Radiochem. Radioanal. L ett., 1969, 2, 257. Paper 6/07674B 13 Bue, J., Complexation Reactions in Aquatic Systems, An Received November 12, 1996 Analytical Approach, Ellis Horwood, Chichester, 1990, ch. 2, Fig. 2.15, p. 48. Accepted January 29, 1997 572 Journal of Analytical Atomic Spectrometry, May 1

ISSN:0267-9477    

DOI:10.1039/a607674b

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Determination of Trace Elements in Rice Flour by Isotope Dilution Inductively Coupled Plasma Mass Spectrometry CHANG J. PARK* AND JUNG K. SUH Korea Research Institute of Standards and Science, P.O. Box 102, Yusung, T aejon, 305–600, Korea Two rice flour reference materials (normal and elevated trace values for As, Cr, Hg and Pb. Under the national demand for rice flour reference materials, the Korea Research Institute of concentrations) were prepared from brown rice produced in Korea.As part of the certification process, trace elements such Standards and Science (KRISS) has initiated the development of rice flour reference materials for analytes at both normal as Cr, Fe, Cd and Pb were determined by an isotope dilution ICP-MS method. About 0.4 g of rice flour samples spiked with and elevated levels. Recently, there has been a breakthrough in the technique of appropriate amounts of enriched spike isotopes was decomposed in high-pressure microwave digestion bombs.For ICP-MS. Excellent work on the reduction of fundamental polyatomic ions under cool plasma conditions have been the determination of Cr and Fe, the ICP-MS instrument employed in the present work was operated under cool plasma reported.3–5 A new generation of commercial ICP-MS instruments can now measure K, Ca and Fe at the low ng l-1 level, conditions. The cool plasma was generated by inserting a copper shield between the load coil and the plasma, and by elements which were problematic owing to the spectral interferences by Ar+, ArH+ and ArO+.4–6 In the present work, an increasing the aerosol carrier gas flow rate up to 1.3 l min-1.An approximately 2 mg ml-1 Ca matrix in the digested analytical method is introduced to determine Cr, Fe, Cd and Pb in rice flour reference materials by the ID method. For the solution was observed to induce serious spectral interference on the determination of Fe; hence, the Ca matrix was separated determination of Cr and Fe, cool plasma conditions at a reduced plasma potential were employed.Owing to a serious from the analyte using a microcolumn loaded with silicaimmobilized 8-hydroxyquinoline. Analytical results for spectral interference caused by CaO+ and CaOH+ under the cool plasma conditions, Fe isotope ratios of microwave- National Institute of Standards and Technology Standard Reference Material 1568, Japan National Institute for digested sample solutions were measured after matrix separation with a microcolumn loaded with silica-immobilized Environmental Studies Certified Reference Material 10-a and Korea Research Institute of Standards and Science reference 8-hydroxyquinoline.materials are presented together with detection limits. Keywords: Inductively coupled plasma mass spectrometry; rice flour; microwave digestion ; matrix separation; silica EXPERIMENTAL immobilized 8-hydroxyquinoline; isotope dilution Instrumentation Rice is a staple food in most Asian countries.In Korea 200– The ICP-MS instrument employed in the present work was a 400 g of rice per person is consumed every day. Thus trace laboratory-constructed unit. Details of the instrument have element analysis of rice is important from both a nutritional been described previously,7 and instrument components and and a toxicological point of view. Japan experienced serious operating conditions are given in Table 1. For sample introduc- health problems (itai-itai disease) caused by the consumption tion, a concentric nebulizer (TR-30-C1, Meinhard, Santa Ana, of rice and drinking water contaminated with Cd.Under the CA, USA) and a Scott-type spray chamber were used with a Food Sanitation Law, sale of unpolished rice containing more peristaltic pump (Minipuls3, Gilson, Villiers-Le-Bel, France). than 1.0 mg g-1 of Cd is prohibited in Japan.1 Because of the The plasma was ignited with a copper shield (1.5×6.3 cm) importance of the accurate determination of Cd in rice, the inserted between the torch and the load coil to eliminate Japan National Institute for Environmental Studies (NIES) capacitive coupling of the plasma with the load coil, and then developed rice flour reference materials containing three levels the aerosol carrier gas flow rate was increased up to 1.3 l min-1 of Cd.In Korea, abandoned mines are one of the sources of at 900 W to suppress the background molecular ion intensity. river pollution and rice grown in the field near the abandoned The shield should not be grounded during ignition, otherwise mines has been reported to be contaminated with heavy it will burn out immediately upon ignition.metals.2 Since quality control of analytical results obtained utilizing rice flour reference materials is essentialto the accurate analysis of rice, there has been a growing national demand for such rice flour reference materials with certified values for both Preparation of Reference Material toxic and essential metals.At present, however, no reference material is available which provides certified values at an A batch (80 kg) of unpolished rice was collected from Nonsan, Korea. It was pulverized with a rotary mill at 18000 rev min-1 elevated level for all the important toxic elements such as As, Cd, Cr, Cu, Hg and Pb. The National Institute of Standards and then screened with a 0.5 mm sieve. One half of the rice flour was spiked with toxic elements such as As, Cd, Cr, Cu, and Technology (NIST) issued Standard Reference Material 1568 Rice Flour in January 1978, but it provides certified Hg and Pb to prepare a reference material with concentrations at an elevated level.The two levels (normal and elevated) of values for toxic elements at only the normal level. Furthermore NIST SRM 1568 is not particularly useful when Cr and Pb rice flour were dried in an oven at 80°C for 12 h and then mixed in a V-blender. The rice flour was packaged in 2 l glass are the analytes, for no certified values for Cr and Pb are given in the certificate.NIES rice flour reference materials are very bottles and irradiated by 60Co radiation for microbiological control. useful for the determination of Cd, but they also lack certified Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 (573–577) 573Table 1 Instrument components and operating conditions of HFand 0.5 ml of HClO4 were added to the digested solution to dissolve any siliceous material.The sample solutions were Component Operating Conditions then heated at 100°C to incipient dryness to expel fluoride ICP generator— and chloride. After adding 2 ml of HNO3 to the sample ICP-16, RF Plasma Products Forward power 900 W solutions, the solutions were again evaporated to near dryness Reflected power <2 W and then diluted to about 30 ml. Frequency 40.68 MH Plasma torch— Argon flow rates/l min-1 Column Preparation and Separation Procedure Precision glassblowing Outer plasma 15 Intermediate plasma 0.4 The column used in the present work is a 0.4 cm id×4 cm Aerosol carrier 1.28 for Fe, length microcolumn (Omega, Upchurch Scientific, Oak 1.15 for Cr and 1.0 for Harbor, WA, USA) supported by a polyethylene frit.After Cd and Pb precleaning the column in 3 M HNO3 and 3 M HCl, the column was loaded with a slurry of silica-immobilized 8-hydroxy- Sample introduction— Meinhard C-type Sample uptake rate quinoline.Prior to use, the entire column assembly was cleaned 0.7 ml min-1 by injecting 50 ml of 4 M HNO3 and then washed with de-ionized water. Sample solution aliquots of 10 ml were Interface— adjusted to a pH of around 5 with dropwise addition of Sampler orifice (nickel) 1 mm Sampling depth 12 mm ammonium hydroxide solution after adding 3 ml of 1 M Skimmer orifice (nickel) 0.7 mm ammonium acetate buer solution. A schematic diagram of Vacuum— the matrix separation apparatus is shown in Fig. 1, where Operating pressures/Torr Interface rotary (2020A, Alcatel) adsorption and elution are performed in two dierent but Interface 2.1 integrated flow systems employing a ten-port valve (Valcor, 2nd Turbo (5402CP, Alcatel) 2nd 5×10-4 Houston, TX, USA). When the valve was in the adsorption 3rd Turbo (5080CP, Alcatel) 3rd 1×10-6 position (dotted line), sample solution in the sample loop was carried through the ion chromatographic (IC) column, trap- Mass filter— 16 mm pole, 220 mm long ping analytes in the column.When the valve was switched to 300 W, 150QC, Extrel the measurement position (double line), the analytes were eluted in a countercurrent mode. Meanwhile, the sample loop Prefilter was refilled with a new aliquot of 10 ml of solution. 16 mm pole, 40 mm long, laboratory construction Isotope Dilution Detector— AF 562A, ETP Deflector +200 V All the analytical data reported here were quantified by the Bias -2.6 kV ID method, which is an especially useful method when a sample has to undergo some chemical pretreatment before analysis, because any loss of analyte during the pretreatment Reagents does not aect the final results.9 The ID method is based on Stock standard solutions of the elements determined were addition of a known amount of enriched isotope to a sample.prepared by dissolution of pure metals. Working standard After equilibration of the spike isotope with the analyte in the solutions were made by serial dilution of the stock solutions.sample, the altered isotope ratio is measured to calculate Silica-immobilized 8-hydroxyquinoline was prepared using analyte concentration by the ID equation given below.7 The Hill’s procedure.8 Silica gel (100–200 mesh, Fisher Scientific, two isotopes of each analyte and the enrichment of the spike Pittsburgh, PA, USA) was subjected to a thorough cleaning isotope employed for the ID are given in Table 2. procedure with aqua regia (HCl+HNO3, 3+1) before use.HNO3 , HF and ammonia solution were electronic grade Cs=Cp Wsp Ws Wp Wrsp AAsp-RBsp Asp-RrBsp RrBs-As RBs-As B purchased from Dongwoo Pure Chemicals (Iksan, Korea) and HClO4 was analytical-reagent grade fromWako Pure Chemical where Cs=analyte concentration in sample (mg g-1); Cp= Industries (Osaka, Japan). Ammonium acetate was first grade concentration of primary standard solution (mg g-1); Wp= from Duksan Pharmaceutical (Yongin, Korea) and was purified mass of primary standard solution in spike calibration solution with Chelex 100 prior to use.The enriched spike isotopes (g); Ws=mass of sample (g); Wsp=mass of spike solution in (53Cr, 57Fe and 111Cd) were purchased from US Services (Summit, NJ, USA). The 206Pb spike isotope (SRM 991) and isotopic standards (SRM 979 Chromium Nitrate and SRM 981 Lead Metal) for mass bias correction of Cr and Pb isotope ratios were obtained from NIST (Gaithersburg, MD, USA).The Fe isotopic reference material (IRMM-014) was purchased from the Institute for Reference Materials and Measurements (Geel, Belgium). Sample Dissolution The rice flour samples were decomposed with a Milestone microwave digestion system (MLS-1200 MEGA, Bergamo, Italy). To about 400 mg of the rice flour sample placed in a Teflon vessel, 3 ml of HNO3 were added together with appropriate amounts of the spike isotopes. The samples were digested inside a microwave oven with a digestion program of 250W Fig. 1 Schematic diagram of matrix separation apparatus. for 3 min followed by 600 W for 5 min. After cooling, 0.5 ml 574 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12Table 2 Isotopes used in ID method Abundance (%) Element Isotope Spike Natural Cr 52 1.85 83.8 53 98.00 9.5 Fe 56 2.40 91.7 57 95.20 2.2 Cd 111 96.31 12.8 112 1.95 24.1 Pb 206 99.90 24.1 208 0.02 52.4 spiked sample (g); W rsp=mass of spike solution in spike calibration solution (g); R=isotope ratio (A/B) of spiked Fig. 2 Eect of pH on Fe recovery eciency and Ca matrix sample; Rr=isotope ratio (A/B) of spike calibration solution; removal eciency. As=atomic fraction of isotope A in sample; Bs=atomic fraction of isotope B in sample; Asp=atomic fraction of isotope A in spike; and Bsp=atomic fraction of isotope B in spike. in the m/z range 52–57. The variation of 36ArO, 40ArO and 40ArOH ion intensities at 900W as the aerosol carrier gas flow rate is increased from 0.96 to 1.28 l min-1 is shown in Fig. 3. RESULTS AND DISCUSSION It can be seen from Fig. 3 that 36ArO ion intensity comes Matrix Separation and Eect of pH down to below 100 counts s-1 at an aerosol carrier gas flow rate of 1.15 l min-1, while the 40ArO ion intensity becomes the The final sample solutions contain a Ca matrix of about same at a flow rate of about 1.28 l min-1. It can be expected 2 mg ml-1, which would give only a little spectral interference that a higher rf power means increased energy input into the under the normal plasma operating conditions.10 However, plasma and hence will require a higher aerosol carrier gas flow under the cool plasma conditions used to suppress background rate to suppress the background polyatomic ions.The eect polyatomic ions, spectral interferences from CaO and CaOH of rf power on 40ArO intensity is shown in Fig. 4. were found to be very serious. The degree of spectral inter- Prior to the analysis of the rice flour samples by the ID ferences on 56Fe and 57Fe with their apparent concentrations method, isotope ratios of the analytes were measured to obtain are shown in Table 3.It is interesting to note that the ratio of mass bias correction factors. The mass spectra in the m/z range molecular ion intensities of CaOH+ to CaO+ is about 1000 49–57 from a both blank solution (solid line) and a 10 ng ml-1 for 1 mg ml-1 of Ca under the cool plasma conditions used in standard solution (dotted line) are shown in Fig. 5. The Cr the present work, while the ratio is below 0.4 at 10 mg ml-1 and Fe isotope ratios were measured in the peak-hopping Ca under the normal plasma conditions.10 Thus, for an accurate mode with three channels per m/z value. Dwell times were determination of Fe, complete elimination of the Ca matrix allocated so that isotopes of lower abundance had longer dwell from the sample solution is a necessity. times. Counts on each channel were accumulated for 300 Since the pH of a solution significantly aects the eciency sweeps.In the isotope ratio measurements, correction for dead of both analyte recovery and matrix removal, the eect of pH time was made by the equation11 given below: was studied to find an optimum pH of the sample solutions for Ca matrix separation. In Fig. 2 analyte recovery eciency m#n (1-nt) and Ca matrix removal eciency are plotted as a function of where m=observed count rate; n=true count rate; and t= pH.The Ca matrix removal eciency is almost 100% at a pH dead time. of up to 5.5, and then it goes down quickly at a pH above 7. Since the dead time of the ICP-MS instrument employed in In order to ensure Ca-free solutions, the sample solutions were the present work is about 20 ns, the error due to the dead time adjusted to pH values of about 5. This implies that the analyte eect would be only about 1%, even at a high count rate of recovery eciency was deliberately sacrificed for complete elimination of the Ca matrix.Also seen in Fig. 2 is the fact that a small change in pH at around 5 induces a large change in the analyte recovery eciency, which means that it is dicult to obtain quantitative recovery of the analytes. In ID analysis, however, the poor reproducibility of the analyte recovery eciency does not reduce the accuracy and precision of analyticaldata, thereby simplifying the separation procedure. Isotope Ratio Measurements For the determination of Cr and Fe by the ID method, it is essential to reduce the background molecular ions appearing Table 3 Spectral interference from Ca molecular ion Apparent concentration/ng ml-1 Ca matrix/ mgml-1 56Fe 57Fe 0.1 0.8 220 0.5 1.5 1300 Fig. 3 Variation of background ion intensity versus aerosol carrier 1.0 2.4 2500 gas flow rate at 900 W. Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 575Table 4 Isotope ratios measured from Fe and Cr isotopic reference materials for mass bias correction* IRMM-014 NIST SRM 979 Run 52Cr553Cr 54Fe557Fe 56Fe557Fe 1 8.821 2.360 40.71 2 8.800 2.366 40.69 3 8.827 2.391 41.06 4 8.821 2.392 41.15 5 8.846 2.369 40.80 Mean 8.823 2.375 40.88 RSD (%) 0.18 0.62 0.51 Certified ratio 8.8189 2.758 43.300 * Dwell time: 2 ms for 52Cr; 10 ms for 53Cr; 10 ms for 54Fe; 2 ms for 56Fe; and 20 ms for 57Fe.Fig. 4 Variation of 40ArO ion intensity versus aerosol carrier gas flow Analysis of Rice Flour rate at three rf powers.Three rice flour reference materials such as NIST SRM 1568, NIES CRM 10-a and KRISS reference materials have been analysed using the ID method. For the determination of Cr and Fe, the samples had to be digested further with HF and HClO4 after microwave digestion with HNO3. The determined concentrations together with certified or reference values are given in Table 5. Each value in Table 5 is an average of five measurements of separately-spiked and pretreated sample solutions.Rice flour reference materials contain some siliceous material, which is an integral part of the sample.1 Therefore, complete dissolution of the materials is required by using a mixture of HNO3, HClO4 and HF. Clear dierences between the two results with the dierent digestion procedures can be seen in Table 5. When the rice flour samples were digested with HNO3 alone, the Cr value for NIES 10-a was found to be almost one half of the reference value, and the determined concentrations of Fe for NIST SRM 1568 and NIES 10-a were slightly lower than the certified values. When the microwavedigested sample solutions were further digested with HF and HClO4 on a hot plate, the determined concentrations of Cr Fig. 5 ICP mass spectrum at m/z 49–57 (solid line=blank; broken and Fe were very close to the certified or reference values. The line=10 ng ml-1 Cr and Fe standard). reference values given for KRISS samples in Table 5 are neutron activation analysis results. For the determination of Cd and Pb, however, microwave digestion with HNO3 was 500 000.The Fe isotope ratios measured from IRMM-014 and adequate and the determined concentrations of Cd and Pb Cr isotope ratio from NIST SRM 979 are shown in Table 4. agree fairly well with the certified values. For Pb, no certified The Fe isotope ratios in Table 4 exhibit some mass discrimi- or reference value is given in Table 5 because it is not available nation against low mass and relatively poor repeatability in the NIST and NIES certificates.compared with that for the Cr isotope ratio, probably because of the cooler pasma conditions at the higher aerosol carrier Detection Limits gas flow rate. This requires more investigation to clarify the reason why the mass discrimination becomes more serious In Table 6, the detection limits and the corresponding blank concentrations are reported for the four analytes. These were under cooler plasma conditions.Table 5 Analysis of rice flour reference materials Determined concentration*/mg g-1 Digestion method Element NIST SRM 1568 NIES 10-a KRISS A KRISS B Digestion with HNO3 only Cr — 0.04±0.02 — 0.07±0.03 Fe 7.7±0.3 11.4±0.3 16.5±0.5 17.5±0.5 Cd 0.026±0.001 0.024±0.001 — 0.031±0.002 Pb 0.036±0.001 — — 0.027±0.002 Complete digestion with HNO3, HF and HClO4 Cr — 0.08±0.02 1.14±0.03 0.16±0.03 Fe 8.6±0.5 12.0±0.3 19.8±0.5 20.3±0.5 Certified or reference value Cr — 0.07 1.08±0.05† 0.16±0.05† Fe 8.7±0.6 12.7±0.7 20.8±0.9† 21.4±2.6† Cd 0.029±0.004 0.023±0.003 — — Pb — — — — * Mean and standard deviation of five determinations.† Neutron activation analysis values for KRISS samples. 576 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12Table 6 Detection limits of about 2 mg ml-1 dissolved in the rice flour sample solution induced a serious spectral interference problem particularly Concentration of blank*/ Detection limit/ onto 57Fe.Therefore, the Ca matrix in the sample solutions Element mg g-1 mg g-1 was eliminated by using a microcolumn loaded with silica- Cr 0.02±0.001 0.003 immobilized 8-hydroxyquinoline. Analyte recovery eciency Fe 0.45±0.013 0.069 was greatly aected by the pH of the sample solutions, but the Cd 0.002±0.0001 0.0003 ID method employed in the present work enabled accurate Pb 0.003±0.0001 0.0003 and precise determination of the analytes in spite of non- * Values given are mean±s, n=5.quantitative recovery. determined by calculating the analyte concentration that REFERENCES yielded a signal three times the standard deviation of the blank signal. The blank concentrations were measured by ID analysis 1 Okamoto, K., Sci. T ot. Environ., 1991, 107, 29. 2 Environmental Data Handbook, Korea National Institute of of blank solutions, which went through the same microwave Environmental Research, Seoul, Korea, 1996. digestion and matrix separation procedure as the sample. The 3 Jiang, S., Houk, R. S., and Stevens, M. A., Anal. Chem., 1988, Fe blank value shown in Table 6 is significantly higher than 60, 1217. other analytes. This is probably because the matrix separation 4 Sakata, K., and Kawabata, K., Spectrochim. Acta, Part B, 1994, procedure was carried out in a normal laboratory, whereas 49, 1027. the separation procedure needs to be done in a clean room to 5 Nonose, N. S., Matsuda, N., Fudagawa, N., and Kubota, M., Spectrochim. Acta, PartB, 1994, 49, 955. reduce the Fe blank. 6 Georgitis, S., and Stroh, A., paper presented at the 5th International Conference on Plasma Source Mass Spectrometry, CONCLUSION Durham, UK, 1996. 7 Park, C. J., Analyst, 1996, 121, 1311. Microwave digestion of rice flour samples with HNO3 alone 8 Hill, J. M., J. Chromatogr., 1973, 76, 455. was adequate for the accurate determination of Cd and Pb. 9 Beary, E. S., and Paulsen, P. J., Anal. Chem., 1993, 65, 1602. 10 Vanhoe, H., Goosens, J., Moens, L., and Dams, R., J. Anal. At. For the determination of Cr and Fe, however, it was necessary Spectrom., 1994, 9, 177. to dissolve the microwave-digested solutions further with a 11 Knoll, G. F., Radiation Detection and Measurement, John Wiley, mixture of HF and HClO4. Under the cool plasma conditions 1979, ch. 3, pp. 96–99. used to suppress the background molecular ions, a Ca matrix Paper 6/07393J Received October 30, 1996 Accepted January 23, 1997 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 577

ISSN:0267-9477    

DOI:10.1039/a607393j

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Determination of Cadmium by Flow Injection Isotope Dilution Inductively Coupled Plasma Mass Spectrometry With Vapour Generation Sample Introduction TARN-JIUN HWANG AND SHIUH-JEN JIANG* Department of Chemistry, National Sun Yat-Sen University, Kaohsiung, 80424, T aiwan A simple and inexpensive laboratory-built vapour generator ion interferences could be significantly reduced. Furthermore, the sampler clogging problem that occurs when a high salt was used with isotope dilution-ICP-MS for the determination of cadmium in several reference materials. The application of content sample is analysed is alleviated.Isotope dilution (ID) techniques have been used in several vapour generation-ICP-MS alleviated the spectroscopic interferences in cadmium determination encountered when a previous ICP-MS applications.13,15,27–33 Since another isotope of the same element represents the ideal internal standard for conventional pneumatic nebulizer was used for sample introduction.The sensitivity, detection limit and repeatability that element, ID results are expected to be highly accurate even when the sample contains high concentrations of concomi- of the flow injection-ICP-MS system with a vapour generator were comparable to those of ICP-MS analysis with tant elements and/or losses occur during sample preparation or during sample introduction into the ICP. conventional pneumatic nebulization. The repeatability of the peak areas and isotope ratio determinations of seven Results obtained for the determination of cadmium by FI-ID-ICP-MS with a vapour generation sample introduction consecutive injections of a 1 ng ml-1 cadmium solution was 3.5 and 1.9%, respectively.The method has a detection limit of device are presented in this paper. The method was successfully applied to the determination of cadmium in several sediment, 0.026 ng ml-1 for cadmium. The method was applied to the determination of cadmium in NIST SRM 2704 and National urine and water reference materials.Research Council of Canada (NRCC) BCCS-1 and PACS-1 sediment reference materials, NIST SRM 2670 Toxic Metals EXPERIMENTAL in Freeze-Dried Urine and several water reference materials. ICP-MS Device and Conditions The concentrations of cadmium in the reference materials were quantified by the isotope dilution method. The precision An Elan 5000 ICP-MS instrument (Perkin-Elmer SCIEX, between sample replicates was better than 10% and the Thornhill, Ontario, Canada) was used.Samples were introanalysis results were within 5% of the certified values for most duced with a vapour generation sample introduction system of the determinations. and/or a cross-flow pneumatic nebulizer with standard Scotttype spray chamber. ICP conditions were selected that maxim- Keywords: Inductively coupled plasma mass spectrometry; ized the cadmium ion signal using an FI method. A simple FI isotope dilution; cadmium; flow injection ; vapour generation; system was used for all the FI work performed. It was biological and environmental samples assembled from a six-port injection valve (Rheodyne 5041) with a 100 ml sample loop.A solution of 10 ng ml-1 cadmium ICP-MS is a relatively new technique for trace multi-element in 0.1 mol l-1 HCl was loaded into the injection loop and and isotopic analysis.1,2 However, it still has some limitations. injected into the carrier of the FI system. The cadmium vapour Highly saline samples can cause both spectroscopic and generated was then transported to the ICP-MS instrument for non-spectroscopic interferences and problems from orifice cadmium determination.The sensitivity of the instrument clogging.3–12 tended to vary slightly from day to day. The operating The determination of cadmium in environmental and conditions used for ICP-MS are summarized in Table 1. biological systems is gaining increasing importance, mainly Data acquisition parameters used are listed in Table 1. The because of the toxic characteristics of the cadmium species.FI peaks were recorded in real time and stored on a hard disk ICP-MS has been satisfactorily applied to the determination with ‘graphic’ software. Peak areas of each isotope were then of cadmium in diverse samples.13–18 However, the determi- used for isotope ratio calculation. Version 2.2 of the Elan 5000 nation of cadmium in samples with high molybdenum and software was used.zirconium contents by ICP-MS is subject to interference by the metal oxide ions of molybdenum and zirconium. All the Vapour Generation System and Conditions useful isotopes of cadmium are overlapped by various isotopes of MoO molecular ions. Furthermore, 94ZrO and 96ZrO inter- A simple and inexpensive laboratory-built continuous-flow vapour generation sample introduction system was coupled fere with the determination of 110Cd and 112Cd, respectively. A possible approach to alleviating these metal oxide inter- with FI-ICP-MS for cadmium determination.A schematic diagram of the FI-vapour generation system is shown in Fig. 1. ferences is to separate molybdenum and zirconium chemically or chromatographically so that they do not reach the plasma.18 A better vapour generation eciency was obtained by placing about 120 glass beads (5 mm diameter) in the gas–liquid The vapour generation sample introduction technique has been used in several atomic spectroscopic applications for separator to increase the surface area for vapour evolution.A detailed description of the vapour generator was given in a cadmium determination.19–26 In the present work, a simple continuous-flow vapour generation system was employed as a previous paper.34 Vapour generated from the vapour generation system was delivered to the ICP-MS instrument sample introduction device for flow injection (FI)-ICP-MS analysis. With this sample introduction system, cadmium could for cadmium determination by means of Tygon tubing (60×0.8 cm id).be separated from the matrix and MoO+ and ZrO+ molecular Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 (579–584) 579Table 1 ICP-MS equipment and operating conditions Merck (Darmstadt, Germany). Trace metal grade HF, HCl and HNO3 were obtained from Fisher (Fair Lawn, NJ, USA). ICP-MS instrument Perkin-Elmer SCIEX ELAN 5000 Element standard solutions were obtained from Fisher.To prepare the solutions to be used as the carrier, suitable amounts Plasma conditions— of thiourea, cobalt standard solution and HCl were dissolved Rf power/W 1000 Plasma gas flow/l min-1 16.0 in pure water to the desired concentrations. The enriched Intermediate gas flow/l min-1 0.900 isotope 111CdO was purchased from Oak Ridge National Nebulizer gas flow/l min-1 0.875 Laboratory (Oak Ridge, TN, USA). A stock solution of approximately 500 mg ml-1 of cadmium was prepared by dis- Mass spectrometer settings— solution of an accurately weighed amount of the material in Bessel box lens/V 10.95 HNO3 and dilution to volume.The concentration of the spike Bessel box plate lens/V -72.50 Photon stop lens/V -10.05 solution was verified by reverse spike ID-ICP-MS. Einzel lenses 1 and 3/V 5.99 Resolution Normal Isotope monitored 112, 111 Dwell time/ms 50 Sweeps per reading 3 Sample Preparation Reading per replicate 60 The applicability of the method to real samples was demon- Number of replicate 1 strated by the analysis of NIST SRMs 2670 Toxic Metals in Points per spectral peak 3 Freeze-Dried Urine and 2704 Bualo River Sediment, National Signal processing— Research Council of Canada (NRCC) BCSS-1 and PACS-1 Peak signal Integrated (Marine Sediment Reference Materials for Trace Elements and Baseline times/ms 2000 Other Constituents), NRCC NASS-4 (Open Ocean Seawater Reference Material for Trace Metals), CASS-3 (Nearshore Vapour generation system— Seawater Reference Material for Trace Metals), SLRS-2 Sample volume/ml 100 Carrier solution 0.1 mol l-1 HCl, 2% m/v (Riverine Water Reference Material for Trace Metals) and thiourea, 2 mgml-1 Co SLEW-2 (Estuarine Water Reference Material for Trace Carrier flow rate/ml min-1 4.8 Metals).The Freeze-Dried Urine reference material was recon- Reductant solution 4% m/v NaBH4 in 0.1 mol l-1 stituted as described in the certificate. Aliquots (10 ml) of the NaOH reconstituted solution and appropriate amounts of enriched Reductant flow rate/ml min-1 2.4 isotope were transferred into closed Teflon PFA vessels.After 5.0 ml of HNO3 had been added, the solutions were heated inside a CEM MDS-2000 microwave digester (CEM, Matthews, NC, USA) to decompose the organic components. The microwave power was set at 80%, and the samples were heated for 20 min. The digests were then diluted to the desired volume with 0.1 mol l-1 HCl after cooling to room temperature.A blank was carried through the digestion procedure, as outlined above, to correct for any cadmium impurities in the reagents used for sample preparation. The sediment samples were dissolved according to the procedure described below. A 0.2 g amount of NIST SRM 2704 Bualo River Sediment, 0.2 g Fig. 1 Schematic diagram of the FI-vapour generation-ICP-MS of NRCC PACS-1 and 0.5 g of NRCC BCSS-1 Marine system. Sediments were weighed into closed Teflon PFA vessels and digested as described below.To each vessel, 3 ml of HNO3, The operating conditions for vapour generation were optim- 3 ml of HNO3, 3 ml of HF and appropriate amounts of ized using an FI method. A stock solution of cadmium at enriched isotopes were added. The mixtures were heated inside 10 ng ml-1 in 0.1 mol l-1 HCl was prepared. This stock a CEM MDS-2000 microwave digester to decompose the solution was then loaded into the injection loop and injected silicate constituent.The microwave power was set at 100%, into the vapour generation system. Several operating param- and the samples were heated for 20 min. After cooling, the eters aected the eciency of cadmium vapour formation. The digests were transferred onto a hot-plate to dry and to evapor- concentration of sodium tetrahydroborate (NaBH4), the con- ate the acid used for sample dissolution. The residues were centration of acid, thiourea and cobalt in the carrier solution, then diluted to the desired volume with 0.1 mol l-1 HCl.A and the volume of the mixing coil were studied to obtain the blank was carried through the digestion procedure to correct optimized conditions. for any analyte impurities in the reagents used for sample In order to compare the eciency of vapour generation with dissolution. With these sample dissolution procedures, the that of conventional pneumatic nebulization, ion signals of sediment and urine samples were completely dissolved and cadmium were obtained with conventional pneumatic nebuliz- equilibrated with the spike.After appropriate amounts of ation and with the FI sample introduction system. In this enriched isotope had been added, 25 ml aliquots of NASS-4, experiment, the gas–liquid separator was removed and a CASS-3, SLRS-2 and SLEW-2 water reference materials were conventional pneumatic nebulizer was used instead. The carrier diluted to 50 ml with 0.1 mol l-1 HCl and then injected into and NaBH4 solutions were replaced with distilled, de-ionized the FI-vapour generation-ICP-MS system for cadmium deter- water at the same flow rate.The signals obtained with the mination. The cadmium concentration in the sample was vapour generation system were then compared with those calculated by means of the equation described in a previous obtained with pneumatic nebulization. paper.13 Owing to the mass bias eect, the intensities obtained during isotope ratio determination of each solution were used Reagents to calculate the isotopic abundance of cadmium. Since the mass bias eect could be factored out during ID calculation, Analytical-reagent grade chemicals were used without further purification.NaBH4, thiourea and NaOH were obtained from the measured isotope ratio did not correct the mass bias eect. 580 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12Fig. 3 Eect of HCl concentration in the carrier on ion signal. Carrier Fig. 2 Eect of NaBH4 concentration in 0.1 mol l-1 NaOH on cad- solutions were mixtures of various concentrations of HCl, 1% thiourea mium signal. Carrier solution was a mixture of 0.1 mol l-1 HCl, 1% and 1 mgml-1 Co. Injected cadmium concentration was 10 ng ml-1 in thiourea and 1 mg ml-1 Co. Injected cadmium concentration was 0.1 mol l-1 HCl. NaBH4 concentration was 4% m/v in 0.1 mol l-1 10 ng ml-1 in 0.1 mol l-1 HCl. Carrier flow rate was 3.5 ml min-1 and NaOH. Carrier flow rate was 3.5 ml min-1 and NaBH4 solution flow NaBH4 solution flow rate was 1.8 ml min-1.All data are relative to rate was 1.8 ml min-1. the signal obtained with simple FI analysis and conventional pneumatic nebulization. Same data treatment method was used in Figs. 3–5. RESULTS AND DISCUSSION Selection of Vapour Generation Conditions The concentration of NaBH4 is critical in the determination of cadmium by vapour generation. Thus, the eect of NaBH4 concentration (% m/v) on the generation of cadmium vapour was investigated. The results are shown in Fig. 2. As the NaBH4 concentration increased, the peak area of cadmium slowly increased and reached a maximum at an NaBH4 concentration of 5% m/v. A high concentration of NaBH4 might generate too much H2 which would cause the plasma to become unstable, leading to a signal with poor reproducibility. Moreover, an orange plasma was observed when a high concentration of NaBH4 was introduced, which could be due to sodium vapour.In subsequent experiments, an NaBH4 Fig. 4 Eect of thiourea concentration in the carrier on ion signal. concentration of 4% m/v was used. Compared with conven- Carrier solutions were mixtures of various concentrations of thiourea, tional pneumatic nebulization, a 15-fold improvement in the 0.1 mol l-1 HCl and 1 mgml-1 Co. Injected cadmium concentration cadmium ion signal was obtained when 4% m/v NaBH4 was 10 ng ml-1 in 0.1 mol l-1 HCl. NaBH4 concentration was 4% m/v was used.in 0.1 mol l-1 NaOH. Carrier flow rate was 3.5 ml min-1 and NaBH4 The eciency of cadmium vapour generation is dependent solution flow rate was 1.8 ml min-1 . on the concentration of acid used. The eect of the concentration of HCl on the cadmium ion signal was investigated. Fig. 3 shows the area of the FI peak as a function of the concentration of HCl in the carrier. As can be seen, as the HCl concentration increased, the peak area of cadmium increased and reached a maximum at a HCl concentration of about 0.1 mol l-1.In subsequent experiments, 0.1 mol l-1 HCl was used. In the work of Guo and Guo,19,20 a solution containing thiourea and cobalt was added to the sample before reaction with NaBH4 solution. These workers suggested that thiourea and cobalt act as a catalyst of the cadmium vapour formation reaction. In the present work, a solution containing thiourea, cobalt and 0.1 mol l-1 HCl was used as the carrier. The eects of the concentrations of thiourea and cobalt on cadmium vapour formation were also studied.Fig. 4 shows the area of the FI peak as a function of thiourea concentration (% m/v). As the thiourea concentration increased, the peak area of cadmium increased gradually. The cadmium signal did not Fig. 5 Eect of cobalt concentration in the carrier on ion signal. change significantly when the thiourea concentration was Carrier solutions were mixtures of various concentrations of Co, greater than 2%.In subsequent experiments, 2% m/v thiourea 0.1 mol l-1 HCl and 2% m/vthiourea. Injected cadmium concentration was used. was 10 ng ml-1 in 0.1 mol l-1 HCl. NaBH4 concentration was 4% Fig. 5 shows the ion signal as a function of cobalt concen- m/v in 0.1 mol l-1 NaOH. Carrier flow rate was 3.5 ml min-1 and NaBH4 solution flow rate was 1.8 ml min-1. tration. The ion signal of cadmium increased rapidly with Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 581increasing cobalt concentration and reached a maximum when the cobalt concentration was about 2 mg ml-1. The cadmium signal decreased gradually when the cobalt concentration was greater than 2 mg ml-1.This could be due to the interference caused in the solution phase of the vapour generation process. A detailed description of the transition metal interference is given in the following section. In subsequent experiments, 2 mg ml-1 cobalt was used as the catalyst of the cadmium vapour formation reaction.Fig. 6 shows the eect of the flow rates of the carrier and NaBH4 solutions on the ion signal. The flow rate of the NaBH4 solution was kept at half that of the carrier solution. An increased carrier flow rate increased the ion signal slightly and reduced the analysis time. The maximum ion signal was obtained when a carrier flow rate of 4.8 ml min-1 was used. Fig. 7 Typical FI peaks for 1 ng ml-1 Cd. Operating conditions of However, an increased solution flow rate could increase the FI-vapour generation system are given in Table 1.amount of H2 generated, causing the ion signal to decrease when the carrier solution flow rate was greater than measurements was 3.5 and 1.9%, respectively, which is similar 4.8 ml min-1. Thus, a carrier flow rate of 4.8 ml min-1 and to the precision obtained in previous ICP-MS experiments NaBH4 solution flow rate of 2.4 ml min-1 were selected in using conventional pneumatic nebulization.13,15,18 The cali- subsequent experiments.bration graph based on peak area was linear (r=0.9991) for Although not illustrated here, it was found in other expericadmium in the concentration range tested (0.05–20 ng ml-1). ments that the volume of the mixing coils and the concentration The detection limit was calculated from the calibration graph of NaOH in the NaBH4 solution did not aect the cadmium and was based on the amount necessary to yield a net signal signal significantly.In fact, the ion signal of cadmium decreased equal to three times the standard deviation of the blank. The gradually with increasing mixing coil volume. In subsequent absolute detection limit was 2.6 pg, which corresponds to a experiments, except for the necessary connecting tubing, no relative value of 0.026 ng ml-1. Further improvement in the extra mixing coil was used. A summary of the optimum gas–liquid separator and the use of more purified reagents operating conditions of the vapour generation system is given should reduce the detection limit even further.in Table 1. Interference Studies Flow Injection Peaks and Detection Limit Transition metal ions decrease the eciency of vapour gener- Typical FI peaks (ICP-MS detection) for a solution containing ation in many atomic spectroscopic applications.19,26,35–37 In 1 ngml-1 of cadmium are shown in Fig. 7, from which it can this work, the interference caused by transition metal ions in be seen that the width of the FI peak is only about 8 s.The the solution phase of the cadmium vapour generation process peak width is narrower than that obtained with simple FI was also studied carefully. Sample solutions containing analysis. This could be due to the fast rate of cadmium vapour 10 ng ml-1 of cadmium and dierent concentrations of various generation. The backgrounds of various cadmium isotopes transition metals were analysed. The results are shown in were increased when vapour generation sample introduction Table 2.Recoveries were calculated by comparison with a was used, which could be due to the presence of cadmium cadmium standard without the interfering ion. As shown, impurities in the reagents used for vapour generation and to recoveries of cadmium decreased rapidly as the concentrations the better analyte transport eciency with vapour generation of Cu, Ni and Pb increased. Although no significant change sample introduction. Repeatability was determined using seven in the isotope ratio was observed, an increased concentration injections of a solution containing 1 ng ml-1 of cadmium.The of other transition metals also decreased the recovery of relative standard deviation of the peak areas and isotope ratio cadmium. Owing to these interferences, the external calibration method could not be used for cadmium determination. In subsequent analyses, the concentrations of cadmium in several reference samples were quantified by the ID method.Table 2 also shows that a stable 111Cd/112Cd ratio could be obtained when Mo and Zr were separated from the analyte with the vapour generation system. Determination of Cadmium in Urine and Water Samples In order to demonstrate the applicability of the system to real samples, several reference materials were analysed. A urine standard reference material with a moderate cadmium content (NIST SRM 2670) was used first to test the proposed FI-vapour generation-ID-ICP-MS method.A 100 ml injection of the digested urine sample was analysed for cadmium using the FI-vapour generation system. The concentration of cadmium present in the solution was quantified by the ID method Fig. 6 Eect of carrier solution flow rate on ion signal. Carrier and the result is presented in Table 3. Although the sensitivity solution was a mixture of 0.1 mol l-1 HCl, 2% m/v thiourea and of cadmium was depressed when a sample with a high salt 2 mg ml-1 Co.Injected cadmium concentration was 10 ng ml-1 in content was analysed, the cadmium concentration determined 0.1 mol l-1 HCl. NaBH4 concentration was 4% m/v in 0.1 mol l-1 by the ID method was in good agreement with the certified NaOH. NaBH4 solution flow rate was half that of the carrier. All data are relative to the first point. value. This experiment indicated that cadmium in urine could 582 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12Table 2 Eect of selected metal ions on the determination of cadmium by vapour generation-ICP-MS* (n=7) Interfering ion Concentration/ng ml-1 Recovery (%) 112Cd/111Cd† Counts of interferent‡ Cu 10 43.2 1.93±0.04 300 50 22.4 1.88±0.10 220 100 15.3 1.94±0.08 580 Zn 100 94.8 1.86±0.05 190 500 70.5 1.94±0.10 230 1000 44.3 1.84±0.16 470 Ni 500 36.4 1.91±0.13 900 1000 17.6 1.91±0.44 1120 2000 8.3 1.92±0.14 1750 Fe 500 81.7 1.89±0.09 500 1000 88.6 1.91±0.07 950 2000 74.4 1.93±0.05 1160 Pb 100 38.6 1.88±0.09 10 700 500 42.2 1.88±0.08 27 200 1000 18.2 1.90±0.18 45 100 95Mo 2000 77.7 1.89±0.04 420 5000 51.1 1.93±0.07 490 96Zr 2000 20.1 1.90±0.05 1500 5000 7.5 1.89±0.04 2500 * Cadmium concentration was 10 ng ml-1 .† Average of seven measurements±standard deviation. Expected value of 112Cd/111Cd is 1.89. ‡ Signal of the selected matrix ion. Table 3 FI-vapour generation-ID-ICP-MS determination of cad- Determination of Cadmium in Sediment Samples mium in NIST SRM 2670 Toxic Metals in Freeze-dried Urine* (n=7) A sample with high concentrations of Zr and Mo gives rise to Concentration/ng ml-1 the molecular ions 96ZrO+, 95MoO+ and 96MoO+, which interfere with the determination of 111Cd+ and 112Cd+, respect- Sample Found Certified ively.The extent of the interference is such that the direct Normal level 0.39±0.03 (0.4) determination of cadmium in a concentrated Zr and Mo Elevated level 86.9±1.9 88±3 sample is dicult. In a separate experiment, it was found that MoO+/Mo+ and ZrO+/Zr+ ratios were 0.5 and 16.7%, * Average of seven measurements±95% confidence limit.respectively, using pneumatic nebulization and normal operating conditions. As described above, the ratio of 111Cd/112Cd remained constant when 5 mg ml-1 spikes of Mo and Zr were added, which demonstrates that MoO+ and be readily determined by vapour generation-ICP-MS using ZrO+ interference is insignificant when vapour generation the FI procedure. sample introduction is used.In order to demonstrate the The concentrations of cadmium in several water reference eectiveness of the proposed method for alleviating the inter- materials (NRCC SLRS-2, SLEW-2, NASS-4 and CASS-3) ference from ZrO and MoO, three samples with high Zr and were also determined by FI-vapour generation-ICP-MS. The Mo contents (NIST SRM 2704 Bualo River Sediment, NRCC results are given in Table 4. Since the concentrations of cad- BCCS-1 and NRCC PACS-1 Marine Sediments) were analysed mium in the injected samples were below the detection limit for cadmium.The amount of cadmium present in each sample of the FI-vapour generation-ICP-MS system, no detectable was determined by the ID method. The metal oxide ion cadmium signal was obtained. However, when the samples interferences were removed by the vapour generation tech- were spiked with 5 ng ml-1 of cadmium, the results obtained were in good agreement with the expected values. nique.The results, given in Table 5, agree with the certified Table 4 Determination of cadmium in water reference materials by FI-vapour generation-ICP-MS* (n=7) Sample Concentration spiked/ng ml-1 Concentration found/ng ml-1 Recovery† (%) Concentration certified/ng ml-1 SLRS-2 0.0 ND‡ 0.028±0.004 5.0 4.81±0.32 96.1±6.3 SLEW-2 0.0 ND 0.019±0.002 5.0 5.00±0.12 100.0±2.3 CASS-3 0.0 ND 0.030±0.005 5.0 4.96±0.13 99.1±2.5 NASS-4 0.0 ND 0.016±0.003 5.0 4.98±0.15 99.5±2.9 * Values are means±95% confidence limit.† Recovery was calculated against the theoretical value (5 ng ml-1). ‡ ND=Not detectable. Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 583Table 5 Determination of cadmium in sediment reference materials 7 McLaren, J. W., Beauchemin, D., and Berman, S. S., Anal. Chem., 1987, 59, 610. by FI-vapour generation-ID-ICP-MS* (n=7) 8 Vaughan, M. A., and Horlick, G., Appl. Spectrosc., 1987, 41, 523. 9 Thompson, J. J., and Houk, R. S., Appl.Spectrosc., 1987, 41, 801. Concentration/mg g-1 10 Gregoire, D. C., Appl. Spectrosc., 1987, 41, 897. 11 Gray, A. L., and Williams, J. G., J. Anal. At. Spectrom., 1987, 2, 599. Sample Found Certified 12 Tan, S. H., and Horlick, G., J. Anal. At. Spectrom., 1987, 2, 745. NIST SRM 2704 Bualo River Sediment 3.52±0.16 3.45±0.22 13 Hwang, T.-J., and Jiang, S.-J., J. Anal. At. Spectrom., 1996, 11, 353. NRCC BCSS-1 Marine Sediment 0.25±0.03 0.25±0.04 14 Yang, H.-J., Huang, K.-S., Jiang, S.-J., Wu, C.-C., and Chou, NRCC PACS-1 Harbor Sediment 2.26±0.15 2.38±0.20 C.-H., Anal.Chim. Acta, 1993, 282, 437. 15 Lu, P.-L., Huang, K.-S., and Jiang, S.-J., Anal. Chim. Acta, 1993, * Average of seven measurements±95% confidence limit. 284, 181. 16 Ebdon, L., Fisher, A. S.,Worsfold, P. J., Crews, H., and Baxter, M., J. Anal. At. Spectrom., 1993, 8, 691. value. Hence, the proposed method is suitable for the determi- 17 Lam, J. W., and McLaren, J. W., J. Anal. At.Spectrom., 1990, nation of cadmium in matrices containing high concentrations 5, 419. of Mo and Zr. 18 Jiang, S.-J., Palmieri, M. D., Fritz, J. S., and Houk, R. S., Anal. Chim. Acta, 1987, 200, 559. 19 Guo, X., and Guo, X., Anal. Chim. Acta, 1995, 310, 377. CONCLUSION 20 Guo, X.-W., and Guo, X.-M., J. Anal. At. Spectrom., 1995, 10, 987. 21 Infante, H. G., Ferna�ndez Sa�nchez, M. L., and Sanz-Medel, A., The merits of coupling FI and ICP-MS with a vapour gener- J. Anal. At. Spectrom., 1996, 11, 571.ation technique for cadmium determination have been demon- 22 Guo, T., Liu, M., and Schrader, W., J. Anal. At. Spectrom., 1992, strated. Since molybdenum and zirconium were separated from 7, 667. 23 Valdes-Hevia y Temprano, M. C., Fernandez de la Campa, M. R., the analyte, MoO+ and ZrO+ molecular ion interferences and Sanz-Medel, A., J. Anal. At. Spectrom., 1993, 8, 847. were alleviated with this sample introduction system. Although 24 Ebdon, L., Goodall, P., Hill, S.J., Stockwell, P. B., and Thompson, the sensitivity of cadmium was depressed when a sample with K. C., J. Anal. At. Spectrom., 1993, 8, 723. a high salt content was analysed, the cadmium concentration 25 Cacho, J., Beltra�n, I., and Nerin, C., J. Anal. At. Spectrom., 1989, determined by the ID method was in good agreement with the 4, 661. certified value. The detection limit for cadmium obtained with 26 D’Ulivo, A., and Chen, Y., J. Anal. At. Spectrom., 1989, 4, 319. 27 Dean, J. R., Ebdon, L., and Massey, R., J. Anal. At. Spectrom., the proposed system is suciently low to allow the determi- 1987, 2, 369. nation of cadmium in many real samples without complicated 28 Gre�goire, D. C., and Lee, J., J. Anal. At. Spectrom., 1994, 9, 393. sample pre-treatment. Further improvement in the vapour 29 van Heuzen, A. A., Hoekstra, T., and van Wingerden, B., J. Anal. generation system (method, reagents and gas–liquid separator) At. Spectrom., 1989, 4, 483. should reduce the detection limit even further. 30 Beary, E. S., Brletic, K. A., Paulsen, P. J., and Moody, J. R., Analyst, 1987, 112, 441. 31 Liaw, M.-J., and Jiang, S.-J., J. Anal. At. Spectrom., 1996, 11, 555. This research was supported by a grant from the National 32 Bowins, R. J., and McNutt, R. H., J. Anal. At. Spectrom., 1994, Science Council of the Republic of China under Contract NSC 9, 1233. 86-2113-M-110-021. 33 Beauchemin, D., McLaren, J. W., Mykytiuk, A. P., and Berman, S. S., J. Anal. At. Spectrom., 1988, 3, 305. 34 Yang, H.-J., and Jiang, S.-J., J. Anal. At. Spectrom., 1995, 10, 963. 35 Brindle, I. D., Alarahi, H., Karshman, S., Le, X.-C., Zheng, S.-G., REFERENCES and Chen, H.-W., Analyst, 1992, 117, 407. 1 Houk, R. S., Anal. Chem., 1986, 58, 97A. 36 Welz, B., and S¢ ucmanova�, M., Analyst, 1993, 118, 1425. 2 Jarvis, K. E., Gray, A. L., and Houk, R. S., Handbook of 37 Chen, H.-W., Brindle, I. D., and Zheng, S.-G., Analyst, 1992, Inductively Coupled Plasma Mass Spectrometry, Blackie, 117, 1603. Glasgow, 1992. 3 Boomer, D. W., and Powell, M. J., Anal. Chem., 1987, 59, 2810. Paper 6/07469C 4 Olivares, J. A., and Houk, R. S., Anal. Chem., 1986, 58, 20. Received November 4, 1996 5 Vaughan, M. A., and Horlick, G., Appl. Spectrosc., 1986, 40, 434. Accepted January 21, 1997 6 Tan, S. H., and Horlick, G., Appl. Spectrosc., 1986, 40, 445. 584 Journal of Analytical Atomic Spectrometry, May 199

ISSN:0267-9477    

DOI:10.1039/a607469c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Investigations Into the Role of Modifiers for Entrapment of Hydrides in Flow Injection Hydride Generation Electrothermal Atomic Absorption Spectrometry as Exemplified by the Determination of Germanium BO HILLIGSØE, JENS E. T. ANDERSEN AND ELO H. HANSEN* Department of Chemistry, T echnical University of Denmark, Building 207, DK-2800 L yngby, Denmark. E-mail: ehh@kemi.dtu.dk Pd-conditioned graphite tubes, placed in the furnace of an that the temperature does not exceed 2300 °C. Yet, for certain AAS instrument, were used for entrapment of germane as hydride-forming elements, such as Ge, which is the analyte of generated in an associated flow-injection system.Two dierent interest here, an atomization temperature of more than 2300 °C approaches were tested with the ultimate aim of allowing is required because lower temperatures will result in lower multiple determinations, i.e., introduction of the modifier as a reproducibility and carry-over of analyte between individual Pd salt solution, and use of electrolytically Pd pre-coated samples.1 tubes.While the former treatment resulted in satisfactory With its longer analytical lifetime, Ir thus appears to be the analytical performance, although requiring separate option of choice. However, Pd yields much higher signals (of regeneration of the tubes prior to each sampling sequence, the the order of five times), which indicates that Pd leads to much electrolytically pre-coated tubes did not function satisfactorily. better entrapment and/or atomization eciency.Furthermore, This in turn led to a closer investigation as to the function of in order to implement an eective automation of an Pd in entrapping hydrides and releasing the analyte. Based on FI–HG–ETAAS system, the modifier should be able to withan evaluation of the behaviour of the two types of tubes, stand at least 50 in situ entrapments to be of practical use. supplemented by scanning tunnelling microscopy studies of Shuttler et al.4 tried to solve this problem by using a combitubes with and without modifier and treated in dierent ways, nation of Ir and Pd for the determination of As.Although it is shown that heating at 2600 °C results in a physical– they were successful in executing more than 300 determinations chemical change of the modifier in the electrolytically coated with a single pre-treatment, it was subsequently shown that tubes; however, this change leads only to a loss of the ability the added modifying eect was due solely to the presence of of the modifier to entrap the hydrides, while the stabilizing Ir.Also, attempts have been made to precipitate the modifier eect is maintained. on the inside of the graphite tube by vaporization,5,6 which essentially should result in a much larger deposition of modifier Keywords: Flow injection ; hydride determination ; palladium than is possible by liquid injection. Although the experimental modifier; electrolytically pre-coated ; graphite tube; germane; results with this procedure were encouraging, they also revealed scanning tunnelling microscopy; role of modifier that the method led to extensive carry-over between samples and, therefore, it was inapplicable for practical purposes. Recently, Bulska and Jedral7 have presented an investigation Recently, work at this laboratory has been conducted on the in which they used Rh and Pd as modifiers for liquid samples.determination of hydride-forming elements by means of flow However, instead of using injected liquid modifier solutions, injection (FI) hydride generation (HG) in conjunction with these workers applied graphite tubes that were electrolytically electrothermal atomic absorption spectrometry (ETAAS) with pre-coated with the metals. The analytical lifetime was greatly the aim of achieving lower limits of detection for the individual enhanced by this approach. Thus, it was reported that the analytes.1 To facilitate an ecient entrapment of the generated Pd-coated tubes could be used for 80 firings for the determi- hydride species, it is inherently necessary to condition the nation of As and Se and 60 firings for Si, using an atomization graphite tubes used with a suitable modifier.A number of temperature of 2600 °C. modifiers have been suggested in the literature, such as Pd, Pt, Stimulated by these results, we contacted the Polish research Rh, Ir or Zn,2,3 although Pd and Ir appear to be the most group and persuaded them to condition a number of graphite ecient modifiers,4 and, therefore, have been used most extentubes electrolytically to allow comparative studies for sively.The modifier is typically introduced into the graphite FI–HG–ETAAS assays of germane. However, the use of the tube by injecting a suitable amount of a salt solution of the electrolytically treated tubes was not successful for the entrap- metal ion, such as Pd(NO3)2. The solution is then dried at a ment of hydrides, which in turn gave rise to a more detailed moderate temperature, whereupon it is treated thermally at investigation into the conditions/mechanisms of hydride collec- around 400–800°C, the object being to reduce the modifier,2 tion.This was eected partly by comparing the performance which is then ready for entrapping the hydrides. of graphite tubes conditioned with modifiers in dierent ways, The individual modifiers dier significantly in analytical and partly by examining the untreated and treated graphite lifetime, that is, the number of determinations that can be tubes by scanning tunnelling microscopy (STM).eected before the modifying eect starts to decrease. Thus, it Set in the context of current theories of the role of the has been reported that the hydride-collection lifetime for Pd modifier, this paper attempts, on the basis of the experimental is merely one determination, while it has been claimed that Ir should allow up to 300 determinations; provided, however, results obtained supplemented by the microscopic examination Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 (585–588) 585of the graphite tubes, to explain the underlying mechanisms RESULTS AND DISCUSSION that might be responsible for the role of the modifier in Experiments with the Pd-modified Graphite Tubes entrapping and atomizing the hydride-forming elements.In order to ensure that the electrolytically Pd pre-coated tubes provided by Bulska and Jedral were identical with those EXPERIMENTAL previously used by the Polish group, a number of experiments Apparatus were initially conducted. Thus, the pre-coated tubes were tested by determining Se introduced as liquid samples. Applying an The FI–HG–ETAAS system used was similar to that described ashing temperature of 1000 °C and an atomization temperature in detail previously,1 consisting of a Perkin-Elmer (Norwalk, of 2200°C, a characteristic mass, m0, of 28 pg per 0.0044 A s CT, USA) Model 2100 atomic absorption spectrometer with was obtained, with an RSD of 2.9% (n=20).Using the same an HGA-700 graphite furnace, combined with a Perkin-Elmer temperature programme and a tube without modifier, the FIAS-200 flow injection unit (controlled by in-house written sensitivitywas about 70% lower, thus confirming the modifying software) for the generation of germane.The gas–liquid separ- eect of the Pd-electroplated tube. ator (Zhaofa, Shenyang, China) had a dead volume of 4.5 ml. Having successfully reproduced some of the results reported The Ge hollow cathode lamp (S&J Juniper, Essex, UK) was by Bulska and Jedral,7 the ability of the electrolytically Pd operated at 10 mA with the wavelength set at 265.2 nm and pre-coated tubes to entrap hydrides was then investigated. For the spectral slit-width at 0.2 nm. Deuterium background cor- this purpose two series of experiments were performed, that is, rection was used throughout. Argon was used as purge and a reference series with graphite tubes conditioned with ‘manu- carrier gas.Unless otherwise stated, the entrapment tempera- ally’ injected modifier, and a series using the Pd-electroplated ture was 400°C and the atomization temperature was 2400 °C, tubes. In both series, a sample volume of 100 ml with a using a 1 s ramp. concentration of 10 mg l-1 of Ge was used.The reference The STM measurements were performed at room tempera- experiments yielded satisfactory results, with a value for m0 of ture in air using a commercially available apparatus (Danish 26 pg per 0.0044 A s.1 Following this, a similar series of experi- Micro Engineering Rasterscope 3000), allowing, via the tunnel- ments was performed using a Pd-electroplated graphite tube. ling current drawn by the scanning tip used, surface areas from The first atomization gave rise to a sensitivity comparable to 10×10 nm to 1000×1000 nm to be imaged.Any influence of that obtained using manually injected Pd, m0 being 30 pg per mechanical vibration on the scanning was damped by placing 0.0044 A s. The appearance of the signal was delayed compared the sample cell on a heavy platform hanging by rubber wires with the reference experiments. Furthermore, the peak shape attached to the ceiling.8 was broader, a dierence that is probably explained by the excessive amount of Pd present.Surprisingly, the following Graphite Tubes measurements using the Pd-electroplated tube did not give Pyrolytic graphite tubes (without platforms; Perkin-Elmer; rise to any signals. Variation of the temperatures of entrapment Part No. P/N B300–0643.H3) were employed for all measure- and atomization did not lead to any improvement, nor did the ments. Conditioning of the graphite tubes was eected by tubes show capacities for the determination of other hydrideinjecting, prior to each sampling cycle, 10 ml of a 1000 mg l-1 forming elements (As and Se were tested). Finally, the atomiz- PdII modifier solution [Pd(NO3)2, Perkin-Elmer/Merck]. The ation time was increased from 4 to 12 s, but to no avail.computer program of the FI–HG–ETAAS system then ensured Since the electrolytically Pd pre-coated tube had been shown that the modifier was duly dried and prepared before the to exert modifying capabilities when analysing liquid samples, entrapment of the hydrides.1 it was decided to ascertain whether Pd was still present in the The experiments with the electrolytically Pd pre-coated graphite tube after the first atomization.This was done by graphite tubes were performed with tubes prepared in Poland making atomic absorption measurements with a Pd hollow by the research group of Bulska and Jedral according to the cathode lamp. At this juncture the graphite tube had been procedure published previously by these workers,7 that is, the used for 20 firings at dierent atomization temperatures tubes, the exterior surfaces of which were pre-covered with between 2200 and 2600 °C.Ten determinations were performed, Teflon tape, were each exposed to electrolysis for 50 min at a giving rise to a Pd signal that gradually decreased from 4.5 to current of 3 mA at 50°C, using an electrolyte solution contain- 3.0 A s with the number of firings, thus confirming the presence ing 2 g Pd per 100 ml (as Na2[Pd(NO3)4]).After this treat- of Pd. ment, the tape was removed from the tubes, which were rinsed The reason for the lack of signal must, therefore, be due to in distilled water and dried at room temperature. The tubes the graphite tube as such. Two explanations are possible: (a) were then ready for use. the hydride does not become entrapped and, therefore, it For the STM investigations, the individual graphite tubes escapes detection; or (b) the hydride is held so strongly by the were carefully cut in half along the axial length, thereby modifier that it does not become atomized (cf.the comments exposing the interior cavity. in the introduction, where it was mentioned that excessive amounts of modifier actually lead to slow release of analyte and excessive carry-over). The fact that Ge was shown to be Reagents entrapped and released during the first measurement cycle All chemicals were of analytical-reagent grade and de-ionized using Pd-electroplated tubes makes possibility (b) unlikely.water (18 MV cm-1), obtained with a Milli-Q water purifier This conclusion is supported by the fact that increasing the (Millipore, Bedford, MA, USA), was used for the preparation atomization time has no eect. of reagents and standards. The reductant for generating the Hence, the explanation for the lack of signal when using the hydrides was a 2.30% (m/v) sodium tetrahydroborate (Fluka, Pd-electroplated tubes must be that Ge does not become Buchs, Switzerland) solution in 0.05 mol l-1 NaOH (prepared entrapped in the graphite tube.Before describing the ensuing daily). The carrier in the FI system consisted of 0.21 mol l-1 STM investigations, it might be advantageous to examine the HCl, as made by dilution of 32% HCl (Merck, Darmstadt, current interpretations of the role of Pd as a modifier. Germany). Standard solutions of germanium(IV) were prepared by two-stage dilution with 0.21 mol l-1 HCl of a 1000 mg l-1 Current Theories of the Role of the Modifying Eects of Pd stock solution (Teknolab, Drøbak, Norway).A number of workers have investigated Pd in order to establish All glassware was soaked for at least 24 h in 1 mol l-1 nitric acid, and rinsed in de-ionized water before use. its eect as a modifier for the analysis of liquid samples,9–11 586 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12and for samples introduced as hydrides.2,12 Concerning the components and ensure a uniform atomization.For samples introduced as hydrides, the function of the modifier is dierent, liquid samples, the prevailing theory is that the analyte and modifier interact during the pre-atomization steps of the the primary aim being an ecient entrapment of the analyte. Evidently, the stabilizing eect of the modifier is of limited furnace programme, either by direct interaction or by migration during heating.11 At higher temperatures during the atomiz- interest if the analyte does not become entrapped in the graphite tube.ation, the stabilization is probably caused by the analyte being embedded in molten Pd. As shown by Majidi and Robertson,13 Majidi and Robertson13 demonstrated that Pd migrates below the graphite surface as the temperature is increased. The this occurs, at least in part, below the graphite surface. Sturgeon et al.,2 who were the first to investigate the ability present work showed that the Pd-electroplated tube lost its ability to entrap hydrides after a single atomization cycle, that of Pd to entrap hydrides, have put forward a hypothesis for the reaction mechanism based on a catalytic eect, suggesting is, after heating of the tube at 2600 °C.A possible explanation of the loss of trapping ability could, therefore, be that the that it is due to the high anity of Pd for hydrogen. Thus, it is noted that Pd is capable of adsorbing an amount of hydrogen modifier migrates into the graphite, thus making it unable to entrap the hydride.In order to broaden the basis for such a corresponding to 900–3000 times the volume of the metal. Furthermore, these workers investigated the entrapment conclusion, investigations by means of STM were performed. capacity of Pd by leading a gaseous stream of nitrogen containing arsine over a piece of polished metallic Pd. Ensuing STM Investigations examination of the surface showed that it was covered by a black film containing As.Similarly, a black film was observed In order to reveal if the topography of the graphite tubes changed as a function of the dierent Pd treatments to which by exposing the polishedPd surface to pure hydrogen. Chaudry et al.12 reported that the collection eciency of germane they were subjected, four graphite tubes were cut open and investigated by STM. The examined tubes and their treatment strongly depended on the contact time between the hydride and the modifier. prior to cutting were as follows: (a) Untreated ‘reference’ graphite tube, which was burned It should be emphasized that when determining analytes introduced as liquid or gaseous samples, respectively, the at 2600 °C for 10 s.(b) Graphite tube electrochemically coated with Pd. After modifier plays dierent roles. For the liquid samples, the aim is to stabilize the analyte already present in the tube, thus coating, the tube was heated to 2400 °C for 5 s.(c) Graphite tube electrochemically coated with Pd. The allowing the temperature pre-treatment to take place at temperatures suciently high to vaporize interfering matrix tube had been fired 50 times. (a) (b) (c) (d) 100 nm 100 nm 100 nm 100 nm 100 nm 100 nm 100 nm 100 nm 37.6 nm 58.1 nm 52.3 nm 106.5 nm Fig. 1 STM images of graphite tube surfaces: (a) surface of an unused, normally conditioned tube; (b) tube surface electrochemically coated with Pd, sequentially heated at 2400 °C for 5 s; (c) surface of tube electrochemically coated with Pd.The image was recorded after the tube had been fired 50 times; (d) tube surface with manually injected Pd(NO3)2 modifier. The tube was heated at 120 °C. For details, see text. Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 587(d) Graphite tube with the modifier injected ‘manually’. Welz et al.15 have shown that manually injectedPd maintains its modifying characteristics for more than 50 determinations After the injection of the liquid solution of Pd(NO3)2, the tube was heated to 90°C, followed by heating at 120 °C, correspond- when analysing liquid samples of mercury, where the maximum temperature of the graphite furnace is about 1500°C.ing to the treatment to which the tube is subjected in real applications. Furthermore, Bulska and Jedral7 showed that electrolytic application of Pd stabilized the modifier thermally, allowing A large number of STM images of the tubes were recorded (more than 100).A representative selection is shown in 50–100 determinations, the atomization temperature being 2200–2600°C, results which, in part, have been reproduced in Fig. 1(a)–(d). As can be seen, the reference tube (tube a) is characterized by an even and smooth surface. On the other the present work. Hence, it can be concluded that the heating at 2600 °C results only in a loss of the ability of the modifier hand, tube b, i.e., with electrolytically applied Pd, exhibits an uneven roughened surface.Tube c, which was subjected to 50 to entrap the hydrides, while the stabilizing eect is maintained. The mechanism of the modifying characteristics of Pd, both firings, appears in a state that is intermediate between that of tubes a and b, i.e., the surface is characterized by being uneven, in connection with solute samples and hydrides, is still not fully understood, making it dicult to explain why electrolyti- but the extent of the unevenness is greatly reduced.The surface of tube d may be characterized as a thin-film of cally deposited Pd cannot entrap hydrides upon heating to 2600°C. The most likely explanation is that the modifier is poorly conducting material with a number of cavities of high conductivity. not available to the hydride. This is supported by the findings of Majidi and Robertson,13 who showed that the modifier The graphite tubes used are obtained from the supplier coated with pyrolytic graphite.The pyrolytic graphite, which migrates away from the tube surface as the temperature increases. The STM images suggest that the electrolytically is deposited by leading methane through the tube at 2000 °C, forms a brittle, layered surface on the graphite.14 A possible coated graphite tubes are degraded as a consequence of repetitive firings. Again, this is supported by an experiment in explanation of the frayed structure seen on the images of tube b [Fig. 1(b)] might be that when Pd is precipitated at which Pd was manually injected into an electrolytically coated graphite tube that had previously been subjected to 50 firings. low overpotentials on the surface, finely grained Pd particles grow across the surface and disrupt the smooth appearance. The determination had a sensitivity that was considerably lower than that obtained by use of graphite tubes that had When the tube is subsequently subjected to elevated temperatures, either a coalescence of Pd particles occurs, a migration not been electrochemically treated.While STM has proved to be a valuable tool in gaining a of Pd particles into the bulk occurs, or the Pd particles evaporate. In any case, the surface will regain its smooth better insight into the role of Pd in hydride generation procedures, it would, however, be beneficial to supplement this appearance as seen in Fig. 1(c). In connection with the measurements on tube d [Fig. 1(d)] approach with investigations by other means of analysis, thereby permitting a better characterization of the form of Pd it proved very dicult to obtain reproducible images. Thus, the images showed large planar surfaces with deep cavities and on the tube surface. gorges. The planar surfaces do not represent the graphite The authors thank Drs. E. Bulska and W. Jedral of the surface, but are areas where the STM tip had diculties in University of Warsaw for preparing the electrolytically Pd ‘contacting’ the surface, because it conducted the current badly, pre-coated graphite tubes.yielding imaging diculties. The planes can be interpreted as the dried Pd(NO3)2, which forms a poorly conducting film on top of the graphite tube, while the cavities in the planar surface REFERENCES represent areas where the tip again is able to draw a tunnelling 1 Hilligsøe, B., and Hansen, E. H., Fresenius’ J. Anal. Chem., (in current and hence be in contact with the underlying graphite the press). surface.The images of tube d are thus an indication that 2 Sturgeon, R. E., Willie, S. N., Sproule, G. I., Robinson, P. T., and Pd(NO3)2 does not decompose at 120 °C. The applicability of Berman, S., Spectrochim. Acta, Part B, 1989, 44, 667. 3 Yan, X., and Ni, Z., J. Anal. At. Spectrom., 1991, 6, 483. this information is, however, limited in relation to the present 4 Shuttler, I. L., Feurstein, M., and Schlemmer, G., J. Anal. At. problem, since the temperature is subsequently elevated to Spectrom., 1992, 7, 1299. 400 °C, which in turn can lead to reduction of Pd. 5 Schlemmer, G., personal communication. 6 Hilligsøe, B., MSc Thesis (in Danish), Technical University of Denmark, 1995. 7 Bulska, E., and Jedral, W., J. Anal. At. Spectrom., 1995, 10, 49. CONCLUSION 8 Andersen, J. E. T., Bech-Nielsen, G., and Møller, P., J. Appl. When Pd is deposited electrolytically, it is the reduced form Electrochem., 1996, 26, 161. 9 Qiao, H., and Jackson, K., Spectrochim. Acta, Part B, 1991, that is precipitated.7 Since Sturgeon et al.2 have shown that 46, 1841. reduced Pd is capable of entrapping hydrides, the electrochemi- 10 Carnrick, G., Schlemmer, G., and Slavin, W., Am. L ab., 1991, cally applied Pd should function as a modifier in connection February, 120. with in situ entrapment. On the other hand, the results obtained 11 Chen, G., and Jackson, K. W., Spectrochim. Acta, Part B, 1996, demonstrate that it is not possible to extend the analytical 51, 1505. lifetime of Pd beyond a single atomization cycle. 12 Chaudry, M. M., Ure, A. M., Cooksey, B. G., Littlejohn, J., and Halls, D. J., Anal. Proc., 1991, 28, 44. On the basis of the present work, it can be concluded that 13 Majidi, V., and Robertson, D., Spectrochim. Acta, Part B, 1991, the heating of the Pd-electroplated tubes to 2600 °C entails 46, 1723. some physical–chemical changes/modifications of the applied 14 Welz, B., Atomic Absorption Spectrometry, VCH, Weinheim, 2nd modifier, which result in the loss of trapping ability. There are edn., 1985. two possible explanations for this loss. Firstly, the heating 15 Welz, B., Schlemmer, G., and Mudakavi, J. R., J. Anal. At. could lead to the formation of Pd species, e.g., the forma- Spectrom., 1992, 7, 499. tion of palladium carbides, lacking modifying abilities. Alternatively, one could imagine that the modifier is not Paper 6/07478B available for the hydride, so that the analyte cannot become Received November 4, 1996 Accepted February 5, 1997 entrapped. 588 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12

ISSN:0267-9477    

DOI:10.1039/a607478b

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Evaluation of the Gas–Liquid Separation Efficiency of a Tubular Membrane and Determination of Arsenic Species in Groundwater by Liquid Chromatography Coupled With Hydride Generation Atomic Absorption Spectrometry FU-HSIANG KO, SHUN-LONG CHEN AND MO-HSIUNG YANG* Department of Nuclear Science, National T sing Hua University, Hsinchu 30043, T aiwan An experimental approach for evaluating the hydride membrane into a basic solution, but also determined the separation eciency of a tubular microporous cyanide concentration in precious metal ores with a spectropoly( tetrafluoroethylene) (PTFE) membrane by flow injection photometric method.Motomizu et al.3 also used a PTFE hydride generation atomic absorption spectrometry (HGAAS) membrane to separate carbon dioxide gas from solution and is described. A method is also proposed for determining arsenic then determined the carbonate concentration in river water species in groundwater by coupling ion-pair chromatography via a spectrophotometric method.Moreover, Pacey et al.4 (IPC) with HGAAS. The evaluation was undertaken with a designed a dual phase gas diusion cell for the permeation of combined gas–liquid separator (GLS), consisting of a tubular arsine gas, and investigated the interference eect of various PTFE membrane and a conventional U-shaped glass device. ions on the arsenic signal in hydride generation atomic absorp- Two important controlling variables, viz., the length of the tion spectrometry (HGAAS).In addition to the PTFE memmicroporous PTFE tubing and the ratio of the tubing id to the brane, the feasibility of using silicone as a membrane has also wall thickness, were investigated. The results showed that the been explored to determine selenium in the digestate of sediabsolute separation eciency of the PTFE membrane can be ments and arsenic in saline waters via inductively coupled eectively evaluated with the combined GLS. In addition, plasma mass spectrometry (ICP–MS).5,6 arsenic speciation in groundwater was also investigated by The relevance of using a microporous PTFE phase separator IPC–HGAAS.The chromatographic separation of arsenic to determine hydride forming elements by employing a flow species with this analytical system was examined in terms of injection system lies in the rapid signal response, small zone the eects of pH and concentration of eluent. The absolute dispersion and eective separation of interferences from the detection limits achieved with a 100 ml injection can be as low co-existing ions during the gas evolution and/or detection as 0.07, 0.11, 0.12 and 0.30 ng for arsenite (AsIII ), processes.7 Despite the above-mentioned advantages of microdimethylarsinic acid, monomethylarsonic acid and arsenate porous membranes, the controlling parameters and the extent (AsV), respectively. A commercially available standard (NIST of their eects on the separation eciency require further SRM 1643b Trace Elements in Water) and the spike additions clarification.method were used to confirm the analytical reliability of the Investigation of the controlling parameters in terms of method. The proposed method has already been routinely membrane separation eciency of the gas evolved from the applied to the speciation of arsenic in groundwater samples. liquid phase has received only limited attention. Among these studies, Yamamoto et al.8 optimized the size of the gas–liquid Keywords: Gas–liquid separator; poly(tetrafluoroethylene) membrane; groundwater sample; arsenic; speciation; ion-pair separator (GLS) by altering the length of the microporous reversed-phase liquid chromatography; hydride generation PTFE tubing and applied the optimized conditions to deteratomic absorption spectrometry mine inorganic arsenic species via flow injection HGAAS.Barnes and co-workers7,9 evaluated the controlling variables of a microporous PTFE membrane separator based on a The use of a microporous poly(tetrafluoroethylene) (PTFE) permeation model for hydride generation and detection by membrane, with its hydrophobic properties, as an interface for ICP-AES.From the above studies, they established an empiri- gas–liquid separation not only prevents the penetration of the cal equation that related the volume of hydride passing through aqueous solution, but also allows the gas evolved from the the tube wall with various physical parameters, e.g., pore size, solution to permeate freely through the surface pores.1–4 porosity, surface area of the tubing, the permeability of the Membrane applications have been commercialized in recent hydride and the ratio of the tubing id to the wall thickness.years for on-line de-gassing purposes in the eluent reservoir of Despite such studies on the controlling variables in order to high-performance liquid chromatography (HPLC) systems to improve the gas separation eciency, the development of a replace the conventional o-line de-gassing method (i.e., sparg- method to evaluate the overall separation eciency of an ing, heating, vacuum and sonication).In addition, the use of a analyte gas in a specific GLS is still of interest, because a GLS membrane with prior gas–liquid separation for analytical used in routine analysis may not always achieve its maximum purposes has received extensive attention as a means of pre- separation eciency due to the short length of the tubular venting the interference eect from the liquid solution.1–4 membrane or degradation of the membrane after prolonged Ohura et al.1 determined the ethanol concentration in alcoholic use. This inability may lead to incomplete diusion of the beverages with a potentiometric method by permeating the analyte gas through the membrane wall and, ultimately, cause ethanol vapour at an elevated temperature through the memthe analytical precision and accuracy to deteriorate.Therefore, brane pores. Marion et al.2 not only established a method that allowed gaseous hydrogen cyanide to diuse across a PTFE a method should be established to evaluate the overall recov- Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 (589–595) 589ery of the analyte gas, thereby ensuring that the membrane Instrumentation separation system can be used for quantitative analysis. Fig. 1 depicts the on-line system used for evaluating the hydride In view of the toxicological eect of arsenic both on environ- permeation eciency of the GLS.The system consisted of four mental and biological systems, analytical techniques must be main parts: a flow injection device, a hydride generator, a GLS developed to determine not only total arsenic but also indi- and an atomic absorption spectrometer. AsIII solution was vidual arsenic species in the sample down to extremely low injected into the flow injection system, consisting of a dual- concentration levels.For this purpose, both on-line and piston solvent-delivery pump (Model 501; Waters, Milford, o-line separation and determination systems have been MA, USA) equipped with a Waters U6K sample injection extensively investigated. This work investigates the on-line valve fitted with a 0.1 ml stainless-steel sample loop. separation of arsenic species in groundwater by ion-pair A continuous hydride generation system was used for gener- reversed-phase liquid chromatography and the subsequent ating the volatile arsenic species. The euent from the flow determination of the separated species.system was led through two tee connectors (A and B) to a This study attempts not only to establish a method for Teflon mixing loop of length 1 m (0.8 mm id, 1.5 mm od). The quantitatively evaluating a specific microporous PTFE mem- hydride generation reagents were delivered from the reagent brane-based GLS for hydride separation eciency, but also to reservoirs to the mixing loop with pressurized argon gas establish an on-line analytical system for determining arsenic controlled by pressure regulators (Model 3104; Omnifit, NY, species, including AsIII, dimethylarsinic acid (DMAA), mono- USA).With this system, the reagents can be delivered more methylarsonic acid (MMAA) and AsV in groundwater samples. smoothly and, ultimately, cause a lower fluctuation of the A combined GLS in which a tubular microporous PTFE detected signal than that obtained by using a peristaltic pump membrane is coupled with a conventional U-shaped separator for reagent delivery.was initially constructed to evaluate the hydride separation After mixing, the sample stream was directed to the GLS, eciency of the tubular membrane. Two important parameters, which consisted of a tubular microporous PTFE membrane viz., the length of the tubular microporous PTFE membrane and a U-shaped glass device.A constant flow of argon carrier and the ratio of the tubing id to the wall thickness, were then gas was achieved by means of a regulating system consisting evaluated, after which the arsenic speciation of groundwater of a bellows pressure valve (Model 5330B; Kofloc, Kojima, was investigated by coupling ion-pair reversed-phase liquid Japan), a mass flow controller (Model 2203; Kofloc) and a chromatography with HGAAS. In addition, a conventional U- flow meter (Model RK1600R; Kofloc).The separated hydride shaped GLS, instead of the microporous PTFE membrane from the GLS was then transferred via a Tygon tube into a separator, was utilized to ensure quantitative recovery of the quartz tube aligned in the lightpath of a flame atomic absorp- analytes and the long-term stability of the analysis. The tion spectrometer (Model 4000; Perkin-Elmer, Norwalk, CT, possible reaction mechanism in the column separation process USA). was also investigated.Regarding the tubular membrane GLS, two dierent modes were used. One mode was constructed with two concentric tubes between which the argon carrier gas passed to the AAS detector [Fig. 2(a)]. The outer tube typically consisted of a EXPERIMENTAL 27.5 cm×10 mm od glass tube with gas inlet and outlet ports located approximately 3 cm from each end, on opposite sides Reagents and Vessels of the tube. The inner tube consisted of microporous PTFE All reagents used were of analytical-reagent or higher grade tubing (Gore Tex, Okayama, Japan) firmly fitted at the ends (Merck, Darmstadt, Germany).Water was treated by reverse of the outer tube with a septum in a flange-type connection. osmosis and mixed-bed ion exchange (Milli-RO 10 PLUS, Tubing of various inner diameters between 1 and 3 mm, and Millipore, Bedford, MA, USA), followed by double distillation with a fixed pore size of 2.0 mm, a porosity of 50% and a in a quartz still equipped with a quartz immersion heater permeability of 0.05 cm3 cm-2 s-1, was used.(Heraeus, Destamat, Germany). A 2.5% m/v solution of sodium A laboratory-built system, in which a U-shaped glass device tetrahydroborate (Riedel-de Ha�en, Hannover, Germany) stabil- was coupled with microporous PTFE tubing, was constructed ized with 0.8% m/v sodium hydroxide was prepared daily. to evaluate the overall separation eciency of this concentric Stock standard solutions (1000 ppm) of AsIII, AsV, DMAA GLS [Fig. 2(b)].In this combined GLS, the tubular PTFE and MMAA were prepared by dissolving sodium arsenite, membrane assembly was connected to the U-shaped glass GLS disodium hydrogenarsenate, sodium dimethylarsinate and disodium monomethylarsonate (Chem Service, West Chester, PA, USA), respectively, in 0.2% v/v sulfuric acid and were stored in Pyrex bottles at 4°C until required.10 Under these conditions, the stock solutions remained stable for at least 1 year. Standard solutions of each species were freshly prepared on a daily basis from the stock solutions.For separation of the arsenic species on a reversed-phase column (NG1 and NS1; Dionex, Sunnyvale, CA, USA), the ion pairing reagent tetrabutylammonium phosphate was used as the mobile phase at a concentration of 0.15 mmol l-1. The pH of the mobile phase was adjusted to 5.78 with hydrochloric acid. PTFE and poly(propylene) containers were used throughout. The vessels were cleaned by immersion in 20% v/v nitric acid overnight and steamed successively with nitric acid and water vapour before use.High-density polyethylene bottles were used as containers for collecting groundwater. The bottles were cleaned by immersion in 20% v/v nitric acid overnight and washed repeatedly Fig. 1 Schematic diagram of the proposed hyphenated system. with water. 590 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12Sampling and Analytical Procedure Groundwater samples were collected from deep wells in a district of Putai, Taiwan, where blackfoot disease is endemic.In the sampling step, the well water was allowed to run through the pumping pipe for at least 10 min prior to sample collection. A 100 ml volume of the collected water was filtered through a 0.45 mm Millipore membrane filter immediately after collection. The water samples were stored in situ at 4°C, and analysed within 24 h. A 0.1 ml volume of water sample was loaded in the sample loop and then injected into the guard and separation columns packed with the polystyrene–divinylbenzene-based resin.The arsenic species were eluted at a flow rate of 1 ml min-1 with 0.15 mmol l-1 tetrabutylammonium phosphate of pH 5.78 as the mobile phase, using a high-pressure chromatography pump, and detected by HGAAS. RESULTS AND DISCUSSION Evaluation of Membrane Eciency Unlike the conventional glass GLS with its free flow of hydride, Fig. 2 Schematic diagram of (a) tubular PTFE membrane GLS and the hydride and gas in the tubular microporous PTFE phase (b) the combined system in which a tubular PTFE membrane is separator must diuse through the tubing wall.To evaluate coupled with a U-shaped glass GLS. the diusion eciency of a membrane system, Barnes and Wang9 established experimental conditions for using micropo- with a slight inclination to facilitate the flow of the solution rous PTFE tubing as a GLS in both continuous and flow stream to the glass separator.With this combined GLS, both injection hydride generation inductively coupled plasma atomic the analyte gas (which permeated across the porous tubing) emission spectrometry. Based on the permeation model, they and the undiused analyte gas (which remained in the sample proposed an empirical relationship between the emission signal stream) could be equally detected by the AAS instrument. and factors such as tubing area, porosity, length and wall Table 1 lists the instrumental parameters of the on-line thickness.The volume of hydride and other gases passing system for evaluating the hydride permeation eciency of the through the microporous PTFE tubing wall (V ) is proportional membrane GLS. This on-line system can be further applied to to the pore size (j), porosity (p), surface area (pDL ) of the arsenic speciation by simply inserting a column system between tubing of diameter (D) and length (L ), the permeability of the the flow injection device and the hydride generation system. hydride (a), sample introduction time (t) and the integrated The column system consisted of a guard column (5 mm resin flow rate over the length of the tube (F), and is also inversely beads, 5 cm×4.6 mm id, NG1; Dionex) and a separation proportional to the tubing wall thickness (d), as represented column (5 mm resin beads, 25 cm×4.6 mm id, NS1; Dionex).in the following equation:9 V=kpajpDtFL /d Data Acquisition where k denotes the proportionality constant.For a fixed The signals from the AAS instrument were recorded in real- length and equal sampling times, the ratio of the microporous time with ‘CHEM-LAB’ software developed by Scientific tubing id to the wall thickness (D/d) is considered to be the Information Service (Taipei, Taiwan), and stored on the hard controlling variable in this phase separation system.9 disk of a PC. The signals were then smoothed by a five-point Having already examined the respective controlling variables Savitzky–Golay method.11 Peak area was calculated by auto- for the gas separation eciency, the development of a method matically integrating the total areas above the baseline using to evaluate the overall separation eciency of the analyte gas points on both sides of each peak with this software. in a specific PTFE membrane-based GLS as in Fig. 2(a) was also considered to be of interest. For this purpose, Fig. b) depicts the combined GLS devised in this work by coupling a Table 1 Instrumental parameters of the on-line system tubular microporous PTFE membrane to a conventional UHydride generation— shaped GLS.With this combined system, the hydride evolved Argon pressure in NaBH4 solution bottle 1.05 atm from the sample stream which permeates across the membrane NaBH4 solution flow rate 1.2 ml min-1 is carried by argon gas into the conventional U-shaped GLS, Argon pressure in HCl solution bottle 1.3 atm while the undiused analyte gas still remaining in the sample HCl solution flow rate 6.2 ml min-1 stream can be further brought into the U-shaped GLS, separ- Argon flow rate through the GLS 175 ml min-1 ated and eventually detected.The integrated signal recorded Atomic absorption spectrometer— from the combined GLS system can function as an indicator Wavelength 193.7 nm of the total recovery of the analyte (i.e., 100%) in the sample; Bandpass 0.7 nm meanwhile, that from the PTFE membrane represents only Lamp power (electrodeless discharge lamp) 8 W Signal read interval 0.1 s those analytes diusing across the membrane. The absolute Background correction (deuterium lamp) On separation eciency of the membrane separator system can, therefore, be deduced from the signal ratio obtained from Combined gas–liquid separator— Tubular microporous PTFE membrane Inner diameter, Fig. 2(a) with respect to Fig. 2(b). 1–3mm Various parameters can aect the separation eciency of a Pore size, 2.0 mm tubular microporous PTFE membrane for volatile hydrides, Porosity, 50% in accordance with the equation of Barnes and co-workers.7,9 U-shaped glass device Shown in Fig. 2(b) By using the combined GLS system, the respective factors and Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 591the extent of their eect on the membrane separation eciency range tested (curve a), the separation eciency of the tubular membrane GLS (curve b) increases linearly with an increase can be quantitatively evaluated.In this study, two important in the ratio of tubing id to wall thickness from 2.5 to 6.0. As controlling parameters were evaluated, viz., the length of the curve b indicates, increasing the ratio beyond 6 apparently tubular membrane and the ratio of the tubing id to the wall causes the separation eciency to increase further, thereby thickness. achieving a maximum value. This maximum value, however, Fig. 3 summarizes the eects of the length of the tubing on is not quantitative because the tube length used in the test was the absolute separation eciency for arsenic hydride.A tubular only 25 cm, i.e., shorter than that required to achieve quantitat- membrane of id 3 mm and an outer glass tube of diameter ive recovery as in Fig. 3. From the above result, it can be 10 mm were used. Experiments were performed by injecting inferred that increasing the ratio can provide a larger gas– 50 ng of arsenite, followed by measurement with the instrumenliquid separation volume to allow the arsenic hydride to tal parameters listed in Table 1.According to Fig. 3, the permeate through the wall, subsequently causing a higher absolute separation eciency increases linearly from about separation eciency. 55% at a length of 7 cm to 100% at about 27 cm, and then With this established combined GLS, the overall eciency levels o up to a length of 50 cm, which was the maximum of the membrane separator system can be eectively evaluated.tested in this work. This finding suggests that the diusion This experimental design also confirms the validity of con- eciency of arsenic hydride through the membrane is, to a trolling the parameters of the empirical relationship proposed certain extent, directly proportional to the length of the by Barnes and Wang.9 Moreover, it can provide further membrane tubing. advantages to ensure quantitative recovery of the analyte gas The separation eciency was further evaluated by varying and the long-term signal stability of the GLS, regardless of the ratio of the microporous tubing id to the wall thickness, possible deterioration of the membrane that might occur after which is thought to be the most influential factor of this the membrane system had been used for a prolonged period membrane-based GLS system.The tubular membranes used of time. had various id values (1, 2 and 3 mm) and two dierent wall thicknesses, viz., 0.4 mm for a tubing id of 1 and 2 mm, and 0.5 mm for a tubing id of 3 mm.Other parameters, e.g., Optimization of Hydride Generation Conditions for Various Arsenic Species membrane tubing length (25 cm), pore size (2.0 mm), porosity (50%) and permeability (0.05 cm3 cm-2 s-1) were identical for Arsenic speciation can generally be achieved by chromato- all the membranes. Experiments were performed by injecting graphic separation and subsequent determination of the a constant amount of 50 ng of arsenite into the hyphenated respective arsenic species by HGAAS.The use of a PTFE system, and applying the experimental parameters listed in membrane as a GLS is favourable owing to its characteristic Table 1. Fig. 4 presents the absolute separation eciency versus features of a rapid signal response and small zone dispersion, the ratio of tubing id to wall thickness subsequently obtained. whereas the conventional U-shaped glass GLS, although poss- As can be seen, in contrast to the quantitative recovery of ibly suering from the disadvantage of peak broadening, can analyte gas obtained by the combined system over the entire ensure good separation eciency of the analyte gas.The separation eciency of the GLS would be of particular concern under circumstances, such as those found here, where good separation of the respective analytes by the chromatographic system used can be achieved. For achieving quantitative gas– liquid separation with this on-line analytical system, the conventional U-shaped glass GLS and the combined GLS are equally applicable.However, use of the conventional GLS is more favourable from the perspective of routine analysis because of the simplicity of the system. Before determining arsenic species in groundwater by HPLC in conjunction with HGAAS, the hydride generation conditions for the respective arsenic species, viz., AsIII, DMAA, MMAA and AsV, should be investigated by using the established Fig. 3 Eect of the length of the PTFE membrane tubing on the hyphenated system. This is necessary because the hydride separation eciency. Tubular PTFE membrane: id, 3 mm; wall thick- generation eciency of arsenic species relies not only on the ness, 0.5 mm; pore size, 2.0 mm; porosity, 50%; and permeability, experimental variables employed but also on the characteristic 0.05 cm3 cm-2 s-1. Sample injection: 50 ng arsenite in 100 ml.properties of the respective arsenic species. A compromise solution obtained from the individual results can be employed for the optimum hydride generation of various arsenic species. Among the prominent factors aecting the eciency of hydride generation are the concentrations of hydrochloric acid, sodium tetrahydroborate and sodium hydroxide, and the flow rate of the carrier gas. The feasibility of optimizing these variables by means of a univariate searching method was investigated.Fig. 5 summarizes the results. Fig. 5(a) predicts the eect of the hydrochloric acid concentration on the signal intensity for the respective arsenic species. As can be seen, the signal intensities of AsIII, MMAA and AsV increase with increasing HCl concentration. As the HCl concentration approaches 4% v/v, the signal intensities of the respective species reach their maximum levels and then remain Fig. 4 Eect of the ratio of tubular membrane id to wall thickness constant up to 10% v/v.The DMAA signal intensity, however, on the separation eciency. (a) Combined GLS; (b) tubular membrane behaves dierently, abruptly increasing to a maximum value GLS. PTFE membrane: length, 25 cm; pore size, 2.0 mm; porosity 50%; at about 1% v/v and then rapidly decreasing. A previous and permeability, 0.05 cm3 cm-2 s-1. Sample injection: 50 ng arsenite in 100 ml. investigation revealed a similar trend in signal changes with 592 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12In a recent report, Hu et al.24 loaded a reversed-phase ODS column with ‘strong/strong’ charged zwitterionic species and then used pure water as the mobile phase for the separation of inorganic solutes and organic zwitterionic solutes. In this study, ion-pair reversed-phase HPLC using TBA phosphate as the mobile phase was employed for the speciation of arsenic. The influence of several parameters such as pH and the ionpairing reagent concentration was also examined in order to achieve eective chromatographic separation of AsIII, DMAA, MMAA and AsV.The fact that the four arsenic compounds contain an acidic group in their chemical structure explains why their apparent charge relies on the pH of the mobile phase.25 Varying the Fig. 5 Eect of changing the concentrations of HCl, NaBH4 and apparent charge for each arsenic species versus pH can serve NaOH and carrier gas flow rate on the signal response of arsenic.A as a useful index to the elution order in ion-pair reversed- 50 ng amount of each arsenic species was used in all instances. %, phase chromatography. By considering the acid dissociation AsIII ; &, DMAA; #, MMAA; and $, AsV. (a) Hydrochloric acid constants of AsIII, DMAA, MMAA and AsV (see Table 2), the concentration; (b) sodium tetrahydroborate concentration; (c) sodium elution order can be predicted from their respective apparent hydroxide concentration; and (d) carrier gas flow rate.charge and the hydrophobicity of each species. Table 3 summarizes the eect of the pH of the mobile phase HCl concentration.12 A compromise HCl concentration of 3% on the retention time and capacity factor for the four dierent v/v was chosen for the four arsenic species. arsenic species. As can be seen, the chromatographic behav- Fig. 5(b) depicts the eect of the sodium tetrahydroborate iours in terms of capacity factor and retention time depends concentration on the signal responses of the arsenic species.on the acid dissociation constant of each arsenic species. To The responses of all the species generally increase with increas- illustrate this, the data obtained at pH 5.78 will be used in the ing sodium tetrahydroborate concentration, and then gradually following discussion. AsIII and DMAA exist predominantly in decrease. By considering the response of the four species, the a neutral form as anticipated from their respective dissociation optimum compromise concentration of sodium tetrahydrobo- constants and are, therefore, predicted to exhibit similar chromrate solution was determined to be 2.5% m/v. atographic behaviour, mainly because it is not possible for Fig. 5(c) demonstrates the eect of the sodium hydroxide these species to form ion pairs with the TBA cation at this concentration, which was added to stabilize the sodium tetra- specific pH. However, as Table 3 reveals, the experimental data hydroborate solution, on the signal intensities for various are not as expected; they exhibit marked dierences in the arsenic species.The responses of the arsenic signals to the chromatographic behaviour between these two arsenic species. sodium hydroxide concentration were not obvious and, there- Hence, AsIII is eluted at the void time of the chromatographic fore, a 0.8% m/v sodium hydroxide solution was chosen. system (retention time 2.7 min), whereas DMAA is eluted later According to a literature report,7 the carrier gas flow rate is with a retention time of 5.3 min and a capacity factor of 1.0.a prominent operating parameter aecting sensitivity in the This dierence in elution order might be accounted for on the hydride generation method. Fig. 5(d) depicts the dependence basis that AsIII, which exists in a neutral form at this specific of the arsenic signal on the carrier gas flow rate. As can be pH, retains its polar character and is, therefore, not adsorbed seen, the signal response reaches a plateau in the flow rate on the hydrophobic surface of the column packing.In contrast, range 100–175 ml min-1, and subsequently decreases fairly DMAA with its methyl groups on the arsenic atom should be rapidly up to 450 ml min-1. A possible explanation for this more lipophilic and, therefore, more easily adsorbed on the phenomenon is that a higher argon flow rate may dilute the hydrophobic surface. Regarding the chromatographic behavarsine produced and also shorten the gas residence time in the iour of AsV and MMAA, as expected, both species should be optical path, thereby causing the signal response to decrease.Therefore, an optimum flow rate of 175 ml min-1 was selected Table 2 Acid dissociation constants of various arsenic species in order to achieve high sensitivity and precision of measurement. Species Formula pK1 pK2 pK3 H3AsO3 AsIII 9.3 C2H7AsO2 DMAA 6.2 Separation of Arsenic Species by Ion-pair Reversed-phase CH5AsO3 MMAA 2.6 8.2 Chromatography H3AsO4 AsV 2.3 6.8 11.3 Before applying the hyphenated system (consisting of HPLC and a U-shaped glass GLS coupled with HGAAS) to determine Table 3 Eect of pH of the mobile phase on the retention time (tR ) the arsenic species, the important parameters for achieving and capacity factor (k¾) for various arsenic species* eective separation by ion-pair reversed-phase liquid chromatography were investigated.The determination of inorganic AsIII DMAA MMAA AsV and organic species of arsenic by using liquid chromatography or ion chromatography coupled with detection techniques such tR/ tR / tR/ tR / pH min k¾ min k¾ min k¾ min k¾ as AAS,13–16 ICP-AES,17,18 ICP-MS19,20 and hydride generation ICP-MS21 has received extensive attention.In studies 2.67 2.7 0 2.7 0 3.0 0.1 8.0 2.0 using ion-pair reversed-phase HPLC, Francesconi et al.22 3.95 2.7 0 3.3 0.2 7.5 1.8 8.6 2.2 5.78 2.7 0 5.3 1.0 8.4 2.1 18.5 5.9 resolved MMAA, DMAA, arsenobetaine and arsenocholine 7.05 2.7 0 7.2 1.7 9.4 2.5 25.2 8.3 on a PRP-1 column with an acidic mobile phase (50 mmol l-1 9.97 5.4 1.0 6.7 1.5 30.2 10.2 52.3 18.4 sodium heptanesulfonate in 2.5% aqueous acid); Morin et al.23 resolved mixtures of arsenite, arsenate, MMAA, DMAA, * Column: Dionex, NS 1; mobile phase: 0.15 mmol l-1 tetrabutylam- arsenobetaine and arsenocholine on an octadecyl-bonded silica monium phosphate; flow rate: 1.0 ml min-1; sample size: 100 ml.column using water as the mobile phase (pH 7.3) and the Capacity factor=(elution time of retained species-elution time of unretained species)/(elution time of unretained species). tetrabutylammonium (TBA) cation as an ion-pairing reagent. Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 593Table 4 Eect of the concentration of tetrabutylammonium phos- strongly retained on the reversed-phase surface of the column phate on the retention time (tR) and capacity factor (k¾) for various packing due to the formation of ion-pair complexes of the arsenic species* anionic arsenic species at this pH (i.e., 5.78) with the ionpairing reagent (TBA cation). However, Table 3 reveals a AsIII DMAA MMAA AsV notable dierence in the chromatographic behaviour between Concentration/ tR/ tR/ tR/ tR/ AsV and MMAA.On the basis of the close resemblance in mmol l-1 min k¾ min k¾ min k¾ min k¾ their apparent charges25 at this specific pH, the dierence in chromatographic behaviour may be attributed to the dierence 0 2.7 0 13.8 4.1 16.8 5.2 100.4 36.2 0.15 2.7 0 5.3 1.0 8.4 2.1 18.5 5.9 in their molecular lipophilic property. 0.5 2.7 0 4.1 0.5 6.1 1.3 9.6 2.6 In addition to the chromatographic behaviour of the various 2 2.7 0 3.2 0.2 4.3 0.6 5.3 1.0 arsenic species at pH 5.78 discussed above, other data obtained 5 2.7 0 2.9 0.1 3.7 0.4 4.7 0.7 at dierent pH values (listed in Table 3) can be similarly explained.As Table 3 suggests, the four arsenic species can be * Column: Dionex, NS 1; mobile phase: tetrabutylammonium phos- eectively separated in the pH range 5–7 with this chromato- phate (pH 5.78); flow rate: 1.0 ml min-1; sample size: 100 ml. graphic separation system. Fig. 6(a) depicts a typical chromatogram obtained by injecting a mixed arsenic standard solution (5 ng of each species) and elution with a mobile phase of TBA phosphate in the mobile phase would cause an increase in the concentration of phosphate anions, which could compete 0.15 mmol l-1 TBA phosphate of pH 5.78.Fig. 6(a) indicates that the four arsenic species can be eciently resolved in a with the anionic arsenic species for the TBA counter cations and, subsequently, tend to attenuate the capacity factor of the sequence assumed to be related to both the apparent charge and the hydrophobicity of each species. analyte ions.26 Interestingly, it was also found that the four arsenic species The eect of the concentration of ion-pairing reagent on the separation of the arsenic species was also investigated.Table 4 can be eciently separated on a column previously loaded with TBA phosphate solution, even when using pure water as depicts the results obtained by elution with a mobile phase of pH 5.78 containing TBA cations at concentrations from 0 eluent. This finding apparently correlates with that of Hu et al.,24 who used pure water as the mobile phase for separating (pure water) to 5 mmol l-1.As can be seen, the retention time of AsIII remains constant at 2.7 min, i.e., the void time of the inorganic solutes on an ODS column previously loaded with ‘strong/strong’ charged zwitterionic species. They suggested chromatographic system, irrespective of the concentration of TBA cations, whereas those for MMAA, DMAA and AsV that the analyte anions and their counter cations combine to produce ‘ion-pairing-like’ forms and are separated by the significantly decrease with increasing concentration of ionpairing reagent.Morin et al.23 observed a similar trend, with simultaneous electrostatic attraction and repulsion interactions between the zwitterionic charged stationary phase, analyte both the capacity factor and retention time decrease with increasing concentration of TBA cations for arsenic species in anion, and its counter cation. As to the reason why arsenic species can be separated on a reversed-phase column, pre- reversed-phase liquid chromatography. They attributed this trend to an increase in the ionic strength caused by an increase viously loaded with TBA phosphate solution, by elution with pure water, it is assumed that the TBA cations adsorbed on in the concentration of the mobile phase.Another possible explanation might be that an increase in the concentration of the surface of the column packing may act as an anion exchanger. To verify this assumption, the column was washed by elution with methanol or acetonitrile, which are conventionally used organic modifiers for mobile phases, after which the separation of the arsenic species was performed in the normal way.Interestingly, the column, after treatment with methanol or acetonitrile, could no longer eect the separation of the arsenic species with pure water as the mobile phase. This finding suggests that the TBA cations initially adsorbed were removed from the column surface by washing with methanol or acetonitrile and, subsequently, caused the column to lose its characteristic anionic-exchange property.Our results also demonstrated that if the column was again treated with the ion-pairing reagent, i.e., the TBA phosphate solution, the anion-exchange property of the column was restored. The following question then arises: Why should the four arsenic species, which may be present either in the anionic or undissociated neutral form depending on their acid dissociation constants (Table 2), be eectively separated with the use of water as the mobile phase.It is anticipated that the separation should not be solely due to an ion-exchange process. Some type of interaction, similiar to the simultaneous electrostatic attraction and repulsion interactions, may also be involved between the charged stationary phase and the analyte species as proposed by Hu et al.24 However, this aspect was not investigated further. Fig. 6 Chromatograms obtained by the proposed method for the According to Table 4, the separation of arsenic species with analysis of an arsenic standard, a groundwater sample and a spiked pure water, although having a certain feasibility, is not of groundwater sample.(a) Mixed arsenic standard solution (5 ng of each practical applicability owing to the long elution time (retention species); (b) groundwater sample (diluted 2-fold) from an area where time 100.4 min) required for the separation of AsV.In order to Blackfoot disease is endemic; (c) groundwater sample of (b) after improve the elution eciency of the four arsenic species, it spiking with a mixed arsenic standard. Column: Dionex NS 1; mobile was found that the addition of sodium chloride to the pure phase: 0.15 mmol l-1 TBA phosphate solution (pH 5.78); flow rate: 1.0 ml min-1; sample size: 100 ml. water eluent was eective. However, in this work, a mobile 594 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12phase containing 0.15 mmol l-1 TBA phosphate was chosen arsenic concentration was 715±39 ng ml-1 (n=5), of which AsIII and AsV constituted 595±38 and 120±8 ngml-1, respect- mainly because of the relatively short time required for the separation and the high resolution for the speciation of the ively; no methyl arsenicals (MMAA and DMAA) were found. This finding implies that the DMAA and MMAA concen- four arsenic species. trations in this groundwater sample are below the MDL of 1 ngml-1. Analysis of Groundwater Samples The analytical performance of the proposed method, i.e., ion- The authors thank the National Science Council, Republic of pair reversed-phase HPLC coupled with a U-shaped GLS and China, for financially supporting this research under Contract HGAAS, was evaluated in terms of detection sensitivity and No.NSC84–2621-M007–007 ZA. analytical reliability. The method detection limits (MDLs), defined as three times the standard deviation of the lowest concentration (n=7) that can be found in the chromatogram, REFERENCES were found to be 0.07, 0.11, 0.12 and 0.3 ng for AsIII, DMAA, 1 Ohura, H., Imato, T., Asano, Y., Yamasaki, S., and Ishibashi, N., MMAA and AsV, respectively, by injecting a 0.1 ml sample Anal.Sci., 1990, 6, 541. into the analytical system. The MDLs achievable with this 2 Marion, P., Rouillier, M. C., Blet, V., and Pons, M. N., Anal. hyphenated method are comparable to the best values reported Chim.Acta, 1990, 238, 117. 3 Motomizu, S., Toe, K., Kuwaki, T., and Oshima, M., Anal. previously13–16 for similar quartz-tube flame AAS detection Chem., 1987, 59, 2930. schemes. 4 Pacey, G. E., Straka, M. R., and Gord, J. R., Anal. Chem., 1986, The analytical reliability of the method was also evaluated. 58, 502. Because no water samples with known concentrations of 5 McCurdy, E. J., Lange, J. D., and Haygarth, P. M., Sci. T otal various arsenic species are available, the accuracy of the Environ., 1993, 135, 131.analytical results may alternatively be evaluated by using a 6 Creed, J. T., Magnuson, M. L., Brockho, C. A., Chamberlain, I., and Sivaganesan, M., J. Anal. At. Spectrom., 1996, 11, 505. standard water sample with a known total arsenic concen- 7 Wang, X., and Barnes, R. M., J. Anal. At. Spectrom., 1988, 3, 1091. tration and the spike additions method. Analysis of NIST 8 Yamamoto, M., Takada, K., Kumamaru, T., Yasuda, M., SRM 1643b Trace Elements in Water revealed that only AsIII, Yokoyama, S., and Yamamoto, Y., Anal.Chem., 1987, 59, 2446. with a value of 48.4±2.2 ng ml-1 (n=5), could be determined, 9 Barnes, R. M., and Wang, X., J. Anal. At. Spectrom., 1988, 3, 1083. which correlates well with the ‘suggested value’ for the total 10 Cheam, V., and Agemian, H., Analyst, 1980, 105, 737. arsenic concentration of 49 ng ml-1. From this correlation, it 11 Savitzky, A., and Golay, M. J. E., Anal.Chem., 1964, 36, 1627. 12 Anderson, R. K., Thompson, M., and Culbard, E., Analyst, 1986, can be inferred that the arsenic present in the standard water 111, 1143. sample should exist solely as AsIII. 13 Ve�lez, D., Yba�nez, N., and Montoro, R., J. Anal. At. Spectrom., This proposed method was further verified by checking the 1996, 11, 271. recovery of the respective arsenic ies added to a ground- 14 Woolson, E. A., and Aharonson, N., J. Assoc. O. Anal. Chem., water sample.Fig. 6(b) shows the chromatogram of a ground- 1980, 63, 523. water sample with a high arsenic content collected from the 15 Ricci, G. R., Shepard, L. S., Coloves, G., and Hester, N. E., Anal. Chem., 1981, 53, 610. south-western part of Taiwan. As can be seen, except for a 16 Chana, B. S., and Smith, J., Anal. Chim. Acta, 1987, 197, 177. very large AsIII peak and a relatively small AsV peak, no 17 Spall, W. D., Lynn, J. G., Andersen, J. L., Valdez, J. G., and perceivable peaks of DMAA and MMAA appear. This finding Gurley, L. R., Anal. Chem., 1986, 58, 1340. suggests that the DMAA and MMAA concentrations in this 18 LaFerniere, K. E., Fassel, V. A., and Eckels, D. E., Anal. Chem., water sample are below the detection limit of the proposed 1987, 59, 879. method. Consequently, a groundwater sample was analysed 19 Branch, S., Ebdon, L., and O’Neill, P., J. Anal. At. Spectrom., 1994, 9, 33. after adding AsIII and AsV in amounts that might be expected 20 Shibata, Y., and Morita, M., Anal. Sci., 1989, 5, 107. to be present in the sample andDMAA andMMAA at ng ml-1 21 Magnuson, M. L., Creed, J. T., and Brockho, C. A., J. Anal. At. concentration levels. Fig. 6(c) depicts the chromatogram of the Spectrom., 1996, 11, 893. groundwater sample after the addition of the various arsenic 22 Francesconi, K. A., Micks, P., Stockton, R. A., and Irgolic, K. J., species. An increase in the height of the peaks corresponding Chemosphere, 1985, 14, 1443. exactly to the position of these standard arsenic species as 23 Morin, P., Amran, M. B., Lakkis, M. D., and Leroy, M., Chromatographia, 1992, 33, 581. shown in Fig. 6(a) was observed. The recovery of each added 24 Hu, W., Takeuchi, T., and Haraguch, H., Anal. Chem., 1993, arsenic species ranged from 94 to 110%, indicating the feasibil- 65, 2204. ity of applying the method to groundwater analysis. 25 Morin, P., Amran, M. B., Favier, S., Heimburger, R., and The proposed method has already been applied to the Leroy, M., Fresenius’ J. Anal. Chem., 1991, 339, 504. speciation of arsenic in groundwaters as part of an environmen- 26 Iskandarani, Z., and Pietrzyk, D. J., Anal. Chem., 1982, 54, 1065. tal surveillance programme. Groundwater samples collected 27 Tseng, W.-P., Environ. Health. Perspect, 1977, 19, 109. 28 Chen, C.-J., and Wang, C.-J., Cancer Res., 1990, 50, 5470. from wells in an area on the south-western coast of Taiwan where Blackfoot disease, which is a peripheral vascular disease, has long been endemic,27,28 were analysed for total dissolved Paper 6/06099D Received September 4, 1996 arsenic and for the distribution of AsIII, DMAA, MMAA and AsV. For a typical well water sample analysed, the total soluble Accepted February 17, 1997 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 595

ISSN:0267-9477    

DOI:10.1039/a606099d

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Speciation of Methyl- and Inorganic Mercury in Biological Tissues Using Ethylation and Gas Chromatography With Furnace Atomization Plasma Emission Spectrometric Detection MARIA S. JIMENEZ† AND RALPH E. STURGEON* Institute for National Measurement Standards, National Research Council of Canada, Ottawa, K1A 0R9, Canada A sensitive and interference-free method for the quantification of methylmercury chloride using GC-ECD. This approach is more definitive and widely practiced as it provides a positive of inorganic and methylmercury species in biological tissues is presented using purge-and-trap injection–GC–AES.Samples identification of individual species. Numerous problems associated with work-up of such polar compounds on packed were solubilized with tetramethylammonium hydroxide and the ionic species were purged from aqueous solution after columns has fostered development of derivatization techniques to convert mercury species to their non-polar dialkyl analogues ethylation with sodium tetraethylborate. The species were preconcentrated on Tenax-TA and thermally desorbed onto an which can then be successfully and reliably quantified.Thus, Rapsomanikis et al.8 and Bulska et al.9 utilized ethylating and isothermal (90 °C) GC column packed with 15% OV-3 on Chromasorb W. The separated species were eluted in He to a butylating agents, respectively, to accomplish this goal. Derivatization with sodium tetraethylborate is particularly FAPES source for detection by AES at 253.6 nm.Absolute detection limits of 1 and 7 pg for inorganic and attractive. This reagent is supported in aqueous media and there is no need for secondary extraction steps as the volatile methylmercury, respectively, can be obtained, corresponding to concentration LODs of 0.2 and 1.4 ng g-1, respectively, in ethylated mercury species can be purged from the host matrix via a stream of inert gas and transferred to the chromatographic solid tissue samples. Precision of determination is better than 10% RSD.The accuracy of the technique was validated by the column directly, or following initial trapping on a cryogenic or other adsorbent support. This approach has been widely analysis of National Research Council of Canada CRMs DORM-2, DOLT-2 and TORT-2, certified for mercury used in conjunction with AAS,10,11 AFS,12 AES13 and ICPMS14 detection. Recently, ethylation and solid-phase micro- species content. extraction of the products onto a silica fibre coated with Keywords: Speciation ; mercury; methylmercury ; f urnace poly(dimethylsiloxane) has been successfully utilized with atomization plasma emission spectrometry; sodium GC–MS detection.15 tetraethylborate; chromatography–atomic emission Use of sodium tetraethylborate as an in situ derivatizing spectrometry ; reference materials agent is clearly not limited to mercury speciation but finds application with a number of elements.16 Full advantage of Despite low crustal abundance, mercury and its compounds such a potential multi-element derivatization scheme can only are considered to be ubiquitous global pollutants.They are be realized through coupling to multi-element detection, such present as trace contaminants in all environmental compart- as ICP-MS and MIP-AES systems. The latter are particularly ments, a consequence of both natural and anthropogenic attractive due to small capital investment, low operating costs activities.It is well-established that mercury is biomethylated and picogram-range detection power. An alternative AES in the environment1 and subsequently bioconcentrated in the source that should prove useful for this purpose is FAPES, an food chain.2 The toxicity of organomercury compounds, especi- rf He plasma sustained at atmospheric pressure within a ally the methylated form, exceeds that of inorganic and metallic conventional graphite furnace used for AAS.17,18 This source mercury because their lipophilicity enhances transport across can be configured with a graphite or metallic centre electrode19 cellular membranes.Chemical speciation information is thus and a graphite or ceramic outer tube20 operating at powers relevant, not only to understanding the biogeochemical cycling from 50 to 200 W and from ambient to elevated temperature. of mercury, but also as a prerequisite for toxicity investigations. As a consequence, the FAPES source exhibits extended lifetime It has been recognized for many years that virtually all of the in comparison with MIPs in that there is no torch erosion to total mercury in fish consists of methylmercury.3,4 As a conse- be concerned with and deposition of GC euents can be quence, considerable eort has been invested in the develop- minimized by operation at elevated temperature (1000–2000 K ment of techniques for the separation and identification of continuous).As absolute LODs achieved with this source are individual mercury species; these have recently been concisely in the low picogram range for many elements,21 it is expedient reviewed by Emteborg5 and Puc and Weber.6 Frequently, this to examine its capabilities as an element-specific detector for involves some form of selective extraction followed by chroma- mercury.This study was undertaken in an eort to characterize tography with element-specific detection. The most common the performance of a GC–FAPES system for the speciation of approaches have utilized the operationally defined method- mercury in biological materials.ology of Magos,7 which is based on a selective reduction of inorganic mercury and total mercury using dierent reagents followed by cold vapour atomic absorption (AAS), emission EXPERIMENTAL (AES) or fluorescence (AFS) detection,or variations of the clas- Apparatus sical Westo�o� procedure4 for the species-specific determination A schematic diagram illustrating the generation–chromatography system, similar to that used by Liang et al.,12 is presented † On leave from Department of Analytical Chemistry, University of Zaragoza, Zaragoza 50009, Spain.in Fig. 1. The custom-made Pyrex cell used for derivatization Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 (597–601) 597has been detailed earlier22 and is fitted with a 29/32 tapered top, fine glass frit bottom and side-arm for introduction of the ethylating agent.A1%m/v solution of sodium tetraethylborate (Strem Chemicals, Newburyport, MA, USA) was delivered from a 100 ml Teflon separating funnel to the side-arm of the reaction cell via a length of Tygon tubing fitted to a Rabbit peristaltic pump (Rainin, Boston, MA, USA). Aliquots of sample, placed in the reaction cell, were supported on the porous glass frit. Volatile mercury species formed were sparged from the solution by the action of a 250 ml min-1 flow of He through the frit.The generation cell was silanized before use by rinsing with a dilute methanolic solution of dimethyldichlorosilane (Petrarch Systems, Bristol, PA, USA) followed by air drying. The volatile analytes were subsequently transferred to a trapping column consisting of a 16 cm length of 6 mm od×4 mm id quartz tubing packed with 100 mg of 20–35 mesh Fig. 2 Schematic diagram of the GC–FAPES interface. Tenax-TA (Alltech Associates, Deerfield, IL, USA) secured in place with silanized quartz wool. The column was wound with Nichrome heating wire and covered with a layer of insulating securely press-fitted into the Macor.One end was capped with fibre to promote uniform heating. The leads were attached to the rf centre pin while the other served as the threaded a variac. A thermocouple, set in a dummy trap, was used to connection for a 4 cm×2 mm od high-purity nickel tube. This establish the optimum heating rate and temperature for analyte tube served as the powered centre electrode in the FAPES desorption from this trap.Mercury species desorbed from this source and was capped at one end and transversely drilled heated trap were transferred to the head of a 75 cm long, 6 mm along its length with numerous small holes to permit uniform od×4 mm id quartz tube serving as the chromatographic escape of GC euent gas into the source. In this manner, theumn. The column was packed with 60–80 mesh 15% OV-3 eluted mercury species were directly transported to the most on Chromasorb W,AWDMCS (Chromatographic Specialities, energetic portion of the FAPES discharge for optimum Brockville, Ontario, Canada) held in place by silanized glass excitation.20 The FAPES workhead was also supplied with a wool plugs at each end.The quartz tube was silanized prior 1 l min-1 flow of Ar to the exterior of the tube (to prevent to packing and was placed in the oven of a Varian Aerograph oxidation if high temperature is used) and a 250 ml min-1 flow gas chromatograph (Varian Instruments, Walnut Creek, CA, of He to the usual ‘internal’ purge gas line which aided in the USA) held at 90 °C.The column was conditioned prior to use support of the rf plasma. The furnace tube was of standard by following the manufacturers’ suggestions. Mercury species, dimensions and made from pyrolytic graphite coated electro- eluted in a 135 ml min-1 flow of He from the GC column, graphite but did not contain an injection hole (SGL Carbon, were transported via 1/8 in Teflon tubing to the FAPES source.Ringsdor-Werke, Bonn, Germany). Thus, gas entering the A septum port was also located in this line to permit injection internal volume of the tube (via the GC column through the of mercury vapour into the system for calibration purposes. centre electrode as well as by the left-hand internal purge gas The conventional FAPES source and detection system have line) exited via an opened right-hand internal purge gas line.been described earlier21 and were operated at 40 MHz using All gas flows were regulated with the use of ball-float an RF10L generator and AM5 matchbox (RF Power Products, rotameters (Cole-Parmer, Vernon Hills, IL, USA) and all Voorhees, NJ, USA). The workhead is based on a Perkintransfer lines and conduits consisted of 1/4 or 1/8 in Teflon Elmer (Norwalk, CT, USA) HGA-2200 furnace. In the present tubing, interconnected with Teflon unions and T-pieces.PTFE study, the centre electrode assembly was modified, as illustrated stopcocks in glass bodies were used for all valve applications. in Fig. 2, to permit the introduction of GC euent directly Helium was purified during use by passage through 10 cm through the centre electrode. In order to accomplish this, a columns of activated coconut charcoal. gas-tight side-arm containing a 1 mm id channel in Macor machinable glass ceramic (Wesgo/Duramic, Fairfield,NJ, USA) was supported in a brass body which threaded into a modified Reagents and Standards and extended N-type rf connector.This connector consisted of a brass body fitted with a Macor dielectric containing a 2 mm Stock standard solutions (1000 mg l-1) ofHgII and methylmercury (MeHg+) were prepared by dissolution of mercury(II) od high-purity nickel tube (Aldrich, Milwaukee, WI, USA) chloride(Aldrich Gold Label) in de-ionized, distilled 18MV cm high-purity water (DDW) obtained from a NanoPure system (Barnstead/Thermolyne, Boston, MA, USA) and by dissolution of methylmercury chloride (Alfa Aesar, Johnson Matthey, Ward Hill, MA, USA) in a small volume of propan-2-ol followed by dilution to volume in dilute hydrochloric acid (prepared in-house by sub-boiling distillation of feedstock).These solutions were kept in pre-cleaned Pyrex bottles under refrigerated storage, as suggested by Lansens et al.23 Working standards were prepared by serial dilution of the stocks with DDW.A gas phase Hg0 standard was taken by airtight gas syringe (Hamilton, Reno, NV, USA) from the headspace of a glass bulb containing a few millilitres of metallic mercury equilibrated to room temperature. The mass of Hg0 used was calculated from vapour pressure data, assuming ideal gas behaviour. Sodium tetraethylborate (NaBEt4) (Strem Chemicals) solu- Fig. 1 Schematic diagram of the derivatization–GC–FAPES system. tions were prepared daily at a concentration of 1% m/v and 598 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12kept refrigerated when not in use. No attempt was made to comparison of the response from generated species spikes with the signal from Hg0 introduced via a septum directly into the filter the solutions. A 1 M ammonium acetate buer of pH 5 was prepared by euent from the GC column. reaction of high-purity ammonia solution and acetic acid and stored in a pre-cleaned poly(propylene) bottle.This solution was further purified by addition of several grams of acid-rinsed RESULTS AND DISCUSSION thiol-based Sumichelate Q-10R resin (Sumitomo, Osaka, Performance Optimization Japan) to the bottle followed by periodic agitation over several days. Optimization of experimental parameters was achieved through the response from both injections of Hg0 and gener- A 25% tetramethylammonium hydroxide solution in methanol (TMAH) was obtained from Aldrich and used as received. ation/chromatography of the ethyl analogues of HgII and MeHg+ from aqueous solutions.An antifoaming agent (defoamer product No. 4528R158), obtained in the form of a household carpet cleaning additive As no information was available concerning the performance characteristics of the FAPES source for detection of mercury, from a local hardware store (Home Hardware Stores, Burford, Ontario, Canada), was used for the analysis of some samples. it was of interest to characterize first the response from injections of 20 ml volumes of mercury-saturated He (containing He and Ar were of laboratory grade (Air Products, Ottawa, Ontario, Canada); He was purified of mercury vapour by prior 375 pg Hg0).Selection of the optimum operating power for the FAPES source was based on the maximum S/B achieved. passage through columns of activated coconut charcoal. National Research Council of Canada CRMs TORT-2 Integrated response increased more than 10-fold when the forward power was increased from 30 to 50W but further (Lobster Hepatopancreas), DORM-2 (Dogfish Muscle) and DOLT-2 (Dogfish Liver) were selected for analysis as these increasesto 70Wenhanced the signals by only 20%.A forward power of 50 W was thus selected for use. As expected, the flow materials are certified for species content of mercury. rate of He transporting the Hg0 to the source substantially alters the response, with an optimum for both peak height and Procedures area occurring at 250 ml min-1.Addition of any internal He purge gas directly into the furnace decreases the response, due Nominal 250 mg sub-samples of reference material biological tissues were placed in pre-cleaned screw-capped poly(propy- to analyte dilution and reduced residence time in the excitation volume. For a nominal spectral bandwidth of 0.10 nm, an lene) bottles and 4 ml of TMAH added. The slurry was permitted to stand for a few hours and sucient DDW added LOD of 0.4 pg was obtained from the integrated response and 0.6 pg from the peak height response.These figures are based to bring the volume to 25.0 ml (mass basis). Blanks were processed through an identical procedure. on a 3sblank criterion and essentially correspond to baseline noise fluctuations. The precision of replicate measurements A 5 ml volume of DDW was added to the generation cell along with 1 ml of acetate buer to ensure that the pH of the (n=10) at the 375 pg level was approximately 4.5% RSD and is probably more a reflection of the reproducibility of the reaction solution was 5.Aliquots of digested reference material were added (500 ml TORT-2 solution; 150 ml DOLT-2 solution; syringe sampling/delivery process than the capabilities of the source/detection system. The linear dynamic range extended 500 ml DORM-2 solution for HgII and 200 ml of 1+9 diluted DORM-2 solution for MeHg+) and the cell was connected to more than four orders of magnitude to 9 ng.These figures of merit compare well with other sources currently used for the system. A flow of 250 ml min-1 He sparge gas was initiated through the bottom frit via valve ‘a’ and 100 ml of 1% m/v detection of mercury, surpassing those based on ECD and comparable to those based on MIP-AES and cold vapour- NaBEt4 were metered into the reaction solution via the sidearm of the cell using the peristaltic pump. A total reaction/ AFS techniques.5,24 The source was operated for nearly 500 h during this study and no deterioration of the nickel centre sparge time of 15 min was used during which the volatile products formed were transferred through valve ‘b’ and col- electrode was evident.FAPES thus possesses the required attributes for consideration as an element-specific detector for lected on the Tenax-TA column at room temperature. The sparge gas was passed through valve ‘d’ to vent into a receiver chromatography. Optimum ethylation conditions for HgII and MeHg+ dier flask containing a solution of KMnO4.Valve ‘a’ was subsequently closed and valve ‘c’ was opened such that a flow of little from one literature source to another.8,12,13,25 As such, a pH of 5 was selected for convenience as this is readily attained He at 250 ml min-1 passed through the Tenax-TA column via ‘b’ to waste for a period of 5 min in order to remove any water with addition of acetate buer. The amount of 1% m/v NaBEt4 reagent added to the generation cell was found to have no vapour.Valve ‘c’ was subsequently closed and, in combination with those of ‘d’ and ‘e’, a stream of He at a flow rate of influence on the signal intensities from either species in the range 50–150 ml, with the result that 100 ml was selected for 135 ml min-1 carried the mercury species desorbed from the Tenax-TA column onto the head of the chromatographic use. The lengthy reaction time for ethylation suggested by Bloom and co-workers12,25 was found to be unnecessary with column.In order to eect this, the Tenax column was heated to 200 °C in 30 s. The chromatographic column was main- the present system. Ceulemans and Adams13 concluded that the rate of formation of the ethylated mercury species is the tained at a constant temperature of 90 °C. Elution of analyte species into the FAPES source was permitted to proceed for limiting step in the sparging process and adopted a 10 min purge time with their system; this is in agreement with recent 10 min, after which the generation cell was cleaned and the entire process repeated.Calibration was achieved by both the data of Cai and Bayona.15 In the present study, this ‘reaction time’12,25 was omitted and asingle 15 min He sparge, commenc- method of spike additions to the samples and through direct derivatization of species spikes in bueredDDW in the absence ing with the introduction of the ethylating agent, was found to produce the same response as when the mixture was of a sample matrix.Digestion/generation blanks were also run through the system. The FAPES source was operated at a permitted to react for 15 min prior to sparging. Liang et al.12 noted that Tenax-TA adsorbent was superior forward power of 50 W (0 W reflected) and an auxiliary internal He flow of 250 ml min-1. Emission signals were to Carbotrap for the initial collection of ethylated analyte species, its performance being less subject to small variations recorded at the 253.6 nm Hg(I) resonance line using both peak height and area evaluation.Optimization of all experimental in procedure. Decomposition of organomercury compounds was also less likely to occur with this material and a useful parameters was undertaken. The overall eciency of the system was estimated through lifetime in excess of 300 cycles could be expected. Evidence of Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12 599species decomposition in this system, characterized by the Concentration LODs, based on solubilization of 0.25 g samples in 25 ml with a 500 ml sub-sample, are calculated to be 1.4 and appearance of spurious peaks in the chromatograms, was never noted.Variation of the He carrier gas/GC flow rate from 60 0.2 ng g-1 for MeHg+ and HgII, respectively. The reproducibility of eight replicate measurements of 2 ng to 180 ml min-1 had almost no influence on response. A flow rate of 135 ml min-1 was (arbitrarily) selected for use, as the spikes of each species added to the cell, expressed as the RSD, was 3.7 and 3.1% for MeHg+ and HgII, respectively, based on total elution time could be kept to 7 min with complete resolution of peaks.peak area measurements. Peak height reproducibility was about 3-fold inferior for each species. These performance data Interconversion/decomposition of species during the desorption process did not occur.12,25 Care was taken to ensure that are comparable to those reported by Fischer et al.11 The absolute eciency of the generation–GC–detection the Tenax-TA column was heated to 200 °C in a 30 s period.Fig. 3 presents a chromatogram obtained for the ethylation of system was estimated by comparison of the response from known masses of MeHg+ and HgII, which underwent ethyl- 3 ng spikes of both MeHg+ and HgII in buered DDW. No evidence for any additional or spurious analyte peaks is ation, with that from the direct injection of known masses of Hg0 vapour into the FAPES source under identical conditions present.The peaks noted prior to that for MeHg+ are probably the result of perturbations to the baseline stability of the of forward power and gas flow rates. On average, absolute eciencies of 56 and 53% (±5% RSD) were obtained for plasma as impurities from the ethylation process are eluted into the source. The same peaks and minor baseline disturb- MeHg+ and HgII, respectively. This confirms the earlier observation that the response was independent of mercury species.ances are also evident in sample and blank solutions (cf. Fig. 4). A further assessment of spike interconversion was undertaken The eciency reported by Bloom25 appears to be 100%, but it is not clear from his data if this is relative or absolute. at the 10 ng level and in no case was any spike observed to result in the recovery of any species other than that which was Significantly, it is clear that the FAPES source serves to dissociate molecular forms of mercury eciently.added. It should be noted that the integrated response from the MeHg+ peak is 93% of that from the HgII peak, suggesting equal generation/trapping/detection eciencies. Accuracy The accuracy of the methodology was assessed through Figures of Merit measurement of the concentrations of MeHg+ and HgII in CRMs. Alkaline hydrolysis via methanolic KOH or quaternary The LOD obtained using this methodology was evaluated from replicate measurements (n=10) of preparative blanks.No peaks ammonium hydroxides has been widely used to solubilize biological tissues.11,15,25–30 This methodology preserves the discernible above the baseline noise located at the elution positions of methylethylmercury or diethylmercury could be species integrity of the mercury and we have found that such solutions are stable for at least 1 year in the usual laboratory detected. As a consequence, a window of observation centred at each elution peak was used to estimate the peak height and environment without species degradation.This is in contrast to the conclusions drawn by Harms,31 which suggested that area of each species from procedural blank determinations. Based on a 3sblank criterion, the peak height and area LODs cysteine was necessary to prevent losses of MeHg+ during such high pH digestions. Fig. 4 presents an example of a for MeHg+ were estimated to be 121 and 7 pg while those for HgII were 56 and 1 pg, respectively.The absolute procedural chromatogram arising from the analysis of TORT-2, wherein 500 ml of a 1% m/v digestate of the tissue were added to the blank was undetectable and hence limited by baseline noise to less than 7 pg for MeHg+ and less than 1 pg for HgII. It should cell. The intensity of the peaks correspond to the presence of 60 pg of HgII and 75 pg of MeHg+. Calibration is accomplished be noted that contributions to the ‘blank’ for HgII can arise from the He purge gas in the form of an elevation of the by direct comparison with standards carried through the ethylation process.Contrary to the conclusions reached by detector baseline noise and also from contaminants in reagents, ambient atmosphere and surfaces in contact with the sample. Fischer et al.,11 there was no need to resort to the method of additions to obtain accurate results. Although no diculties Mercury contamination from the He purge gas is eliminated or minimized byprior passage of theHe throughactivated charcoal were encountered with 150 ml volumes of digest, foaming of sample in the generation cell was problematic when 500 ml traps.Additionally, contamination by HgII from the acetate buer was reduced to below the LOD by prior purification volumes were introduced. This was eliminated by the addition with the Sumichelate Q-10R resin. It should be noted that any Hg0 contamination in the He stream used for sparging the generation cell should not constitute a problem as it is not retained by the Tenax-TA column and, hence, not detected.Fig. 4 Typical chromatogram for the analysis of TORT-2: (A) blank solution,(B) TORT-2 solution. Emission peaks correspondto detection of 75 pg of MeHg+ (I) and 60 pg of HgII (II). Note that curve (A) has Fig. 3 Typical chromatogram for the processing of 3 ng spikes each been deliberately displaced vertically to permit comparison of the two traces. of MeHg+ (A) and HgII (B) added to buered DDW. 600 Journal of Analytical Atomic Spectrometry, May 1997, Vol. 12Table 1 Analytical results Value found/mg g-1 Certified value/mg g-1 Sample MeHg+ HgII Total Hg MeHg+ HgII* Total Hg DOLT-2 (n=4) 0.722±0.005 1.34±0.011 2.06±0.016 0.693±0.053 1.30±0.153 1.99±0.10 TORT-2 (n=2) 0.160±0.014 0.126±0.011 0.286±0.025 0.152±0.013 0.118±0.073 0.27±0.06 DORM-2 (n=2) 4.72±0.22 0.180±0.002 4.90±0.22 4.47±0.32 0.17±0.58 4.46±0.26 * By dierence from measured total Hg and MeHg+.of 30 ml of antifoaming agent to the cell. The latter contributed nation of mercury speciation in other biological samples and at lower concentrations remains to be accomplished. no detectable mercury contamination to the system. Results for the determination of HgII and MeHg+ in several The authors thank Henning Vogt, SGL Carbon, Ringsdor- marine biological reference materials are summarized in Werke GmbH, Bonn, for providing the graphite tubes used in Table 1.Concentrations, corrected to dry mass basis (approxi- this study. mately 5% moisture content), are seen to be in excellent agreement with certified values. The precision of determination REFERENCES is as good as 1% RSD in some cases but is typically better than 10%. Each measurement requires about 30 min: 15 min 1 Organometallic Compounds in the Environment, ed. Craig, P. J., for derivatization and sparging, 5 min for drying of the Wiley, New York, 1986.Tenax-TA column and transfer of species to the analytical 2 Surma-Aho, K., Paasivirta, J., Rekolainen, S., and Verta, M., Chemosphere, 1986, 15, 353. column, and 10 min for chromatography. The last step could 3 Westo�o� , G., Acta Chem. Scand., 1966, 20, 2131. be significantly shortened by running the analytical column at 4 Westo�o� , G., Acta Chem. Scand., 1967, 21, 1790. 100 °C and increasing the flow rate of He. Omission of the 5 Emteborg, H., PhD Thesis, University of Umea, Sweden, 1995. 5 min drying step used to purge water vapour from the 6 Puc, R., and Weber, J. H., Appl. Organomet. Chem., 1994, 8, 293. Tenax-TA column completely suppressed the chromatographic 7 Magos, L., Analyst, 1971, 96, 847. 8 Rapsomanikis, S., Donnard, O. F. X., and Weber, J. H., Anal. elution of species from the analytical column. Chem., 1986, 58, 35. 9 Bulska, E., Baxter, D. C., and Frech, W., Anal. Chim. Acta, 1991, 249, 545. CONCLUSIONS 10 Rapsomanikis, S., and Craig, P. J., Anal.Chim. Acta, 1991, As the US Food and Drug Administration has targeted 1 mg g-1 248, 563. 11 Fischer, R., Rapsomanikis, S., and Andreae, M. O., Anal. Chem., (wet mass) as the action level for the concentration of mercury 1993, 65, 763. in fish, it is clear that the FAPES source oers adequate 12 Liang, L., Horvat, M., and Bloom, N. S., T alanta, 1994, 41, 371. detection power, stability and reproducibility for application as 13 Ceulemans, M., and Adams, F.C., J. Anal. At. Spectrom., 1996, an element-specific detector for the GC speciation of mercury 11, 201. in such tissues. Concentration LODs can be further improved, 14 Hintelmann, H., Douglas Evans, R., and Villeneuve, J. Y., J. Anal. At. Spectrom., 1995, 10, 619. if necessary, by processing larger sub-sample volumes. 15 Cai, Y., and Bayona, J., J. Chromatogr., 1995, 696, 113. Application to determination of mercury in aqueous systems is 16 Rapsomanikis, S., Analyst, 1994, 119, 1429.evident, with corresponding sub-pg ml-1 detection power. The 17 Liang, D. C., and Blades, M. W., Spectrochim. Acta, Part B, 1989, methodology is reasonably rapid and solvent free and does not 44, 1059. require that the derivatized species undergo any pyrolysis prior 18 Sturgeon, R. E., Willie, S. N., Luong, V., Berman, S. S., and Dunn, J. G., J. Anal. At. Spectrom., 1989, 4, 669. to detection (as for AFS techniques). Throughput may conceiv- 19 Sturgeon, R. E., Luong, V.T., Willie, S. N., and Marcus, R. K., ably be enhanced by use of a Nafion dryer32 to prevent water Spectrochim. Acta, Part B, 1993, 48, 893. vapourfrom collecting in the Tenax-TA trap, thereby eliminating 20 Pavski, V., Chakrabarti, C. L., and Sturgeon, R. E., J. Anal. At. the 5 min drying time currently used before desorption of the Spectrom., 1994, 9, 1399. analytes to the GC column. 21 Sturgeon, R. E., Willie, S. N., Luong, V. T., and Berman, S. S., Anal. Chem., 1990, 62, 2370.It is clear from this procedure that any ethylmercury or 22 Sturgeon, R. E., Willie, S. N., and Berman, S. S., Anal. Chem., diethylmercury which may be indigenous to the tissue samples 1985, 57, 2311. could not be distinguished from HgII present. There is every 23 Lansens, P., Meulman, C., and Baeyens, W., Anal. Chim. Acta, reason to believe that other organomercurials, such as 1990, 229, 281. dimethylmercury which may be present in biota,33 can be 24 Cappon, C.J., in Analysis of Contaminants in Edible Aquatic Resources, ed. Kiceniuk, J. W., and Ray, S., VCH, New York, quantified with this system. Additionally, speciation of 1993, ch. 8. organometallic forms of other elements of interest, such as 25 Bloom, N., Can. J. Fish Aquat. Sci., 1989, 46, 1131. Sn34 and Pb,13 should be amenable with this system. 26 Trachman, H. L., Tyberg, A. J., and Branigan, P. D., Anal. Chem., The purge-and-trap injection–GC system can probably be 1977, 49, 1090. substantially improved with respect to sample dispersion. The 27 Dogan, S., and Haerdi, W., Int. J. Environ. Anal. Chem., 1980, 8, 249. current half-widths of the peaks are approximately 15 s. 28 Zhou, C. Y., Wong, M. K., Koh, L. L., and Lee, Y. V., T alanta, Reducing the size of the Tenax-TA trap and the application of 1996, 43, 1061. alternative chromatographic columns may serve to resolve this 29 Szpunar, J., Schmitt, V. O., £obin�ski, R., and Monad, J.-L., problem. Concurrent with improvements in this area will be J. Anal. At. Spectrom., 1996, 11, 193. corresponding enhancements in the detection power of the 30 Ceulemans, M., Witte, C., £obin�ski, R., and Adams, F. C., Appl. Organomet. Chem., 1994, 8, 451. system. The statistically calculated LOD suggests that 1 ng g-1 31 Harms, U., Appl. Organomet. Chem., 1994, 8, 645. of mercury can be detected in solid tissue samples using this 32 Kingston, K. J., and McIntosh, S. A., At. Spectrosc., 1995, 16, 118. protocol. It is clear from the traces presented in Fig. 4 that a 33 Puc, R., and Weber, J. H., Anal. Chim. Acta, 1994, 292, 175. larger sample aliquot would be necessary for this level of 34 Carlier-Pinasseau, C., Lespes, G., and Astruc, M., Appl. quantification, or an improvement in the baseline noise of the Organomet. Chem., 1996, 10, 505. system. Alternatively, changes to the chromatographic hard- Paper 6/07729C ware to reduce the dispersion may serve to address this level Received November 13, 1996 Accepted January 15, 1997 of analyte routinely. Application of the system to the determi- Journal of Analytical Atomic Spectromet, Vol. 12 601

ISSN:0267-9477    

DOI:10.1039/a607729c

出版商:RSC    

年代:1997

数据来源: RSC