Flow-injection preconcentration and electrothermal atomic absorption spectrometry determination of manganese in seawater

Analytica Chimica Acta 435 (2001) 343–350

Flow-injection preconcentration and electrothermal atomic absorption spectrometry determination of manganese in seawater Corrado Sarzanini∗ , Ornella Abollino, Edoardo Mentasti Department of Analytical Chemistry, Via P. Giuria 5, 10125 Torino, Italy Received 20 July 2000; received in revised form 15 November 2000; accepted 26 January 2001

Abstract A procedure for the determination of manganese in seawater by flow-injection preconcentration coupled to electrothermal atomic absorption spectrometry was developed. The analyte is complexed by 1-(1-hydroxy-2-naphthylazo)-6-nitro-2-naphthol4-sulphonic acid (Eriochrome Black T), its complexes are retained onto an anion exchange resin, and eluted with hydrochloric acid. A software driven preconcentration apparatus was set up and a new housing for the enrichment column was designed and adopted. After studying the effect of acidity and ligand concentration, the accuracy of the procedure was assessed by analysis of certified seawater. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Manganese; Seawater; Preconcentration; Ion exchange; Electrothermal atomic absorption spectromety; Flow-injection

1. Introduction The determination of manganese in seawater is of interest, from an environmental point of view, in order to elucidate the biogeochemical cycles in which the element is involved: bioaccumulation in living organisms, remineralisation after their death, burial into sediments, adsorption–desorption equilibria with colloidal and suspended particles. One of the most popular analytical procedures for the determination of trace and ultratrace metals in natural waters is flow-injection preconcentration coupled to atomic absorption or emission spectrometry: this technique has the advantages of sensitivity, specificity,

∗ Corresponding author. Tel.: +39-11-6707628; fax: +39-11-6707615. E-mail address: [email protected] (C. Sarzanini).

versatility, and allows to automate the operations and minimise contaminations [1]. One of the most commonly used substrates for flow-injection preconcentration of manganese is Chelex-100, coupled to flame atomic absorption spectrometry [2,3], ICP AES [4] or ICP MS [3]. Other procedures involve the use of 8-hydroxyquinoline (oxine) or its derivatives, which form stable complexes with Mn and transition metals. Oxine was immobilised, prior to preconcentration, on several substrates such as silica gel [5,6], cellulose [7], Toyopearl TSK (formerly known as Fractogel TSK) [8], Amberlite XAD-2 [9,10], whereas 8-hydroxyquinoline-5-sulphonic acid was retained onto active carbon– silica gel [11] and ion-exchangers [9] or supported on cellulose [12]. Among different approaches, two techniques must be mentioned: one involves the immobilisation of oxine directly onto the inner walls of silicone tubing [13], the other is based on the retention

0003-2670/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 3 - 2 6 7 0 ( 0 1 ) 0 0 8 5 6 - X

344

C. Sarzanini et al. / Analytica Chimica Acta 435 (2001) 343–350

of complexes between Mn and 8-hydroxyquinoline on the walls of a knotted reactor [14]. Other chelating substrates employed for FI enrichment of manganese are a poly(aminophosphonic acid) [15] and a thiolic resin [16]; activated carbon was also used for direct adsorption of trace elements [17]. In other techniques, silica C18 was used to retain pre-formed complexes of trace metals with complexing agents such as dialkyldithiophosphates [18] and chloro-oxine [19]. A procedure for speciation of manganese by flow-injection preconcentration developed in our laboratory [20] is based on the retention of metal species onto a cation or anion exchange resin. The method presented here for the determination of manganese in seawater involves on-line complexation with 1-(1-hydroxy-2-naphthylazo)-6-nitro-2-naphthol4-sulphonic acid (Eriochrome Black T, hereafter EBT) and retention of the negatively charged complexes onto a microcolumn packed with an anion exchange resin. A new kind of microcolumn (see below), which can be used for any kind of solid substrates, was designed and used. Chelation takes place through the azo group of EBT and the two adjacent hydroxy-substituents. EBT forms with manganese 1:1 and 1:2 metal to ligand complexes. The presence of a sulphonic substituent, located apart from the coordination site, allows the interaction with an anion exchanger. The retained metal is eluted with a small volume of acid and determined by electrothermal atomic absorption spectrometry (ETAAS). The accuracy of the method was checked by analysing a seawater sample of certified composition. Advantages of the proposed technique are the stability of the ligand solutions, the wide operating range (pH 0–14) of the sorbent and the use of an acidic eluent, more suitable than an organic one in ETAAS determinations. In fact the measured metal concentration in ethanol or acetonitrile media decreased with time, probably, because of adsorption on the walls of the vessel. An advantage of our procedure over the ones based on the use of immobilised chelates is that on-line precomplexation of the sample allows to minimise analyte losses due to sorption of hydroxo-species onto the container walls or the preconcentration apparatus [21] and avoids their precipitation.

2. Experimental 2.1. Reagents and apparatus High purity water (HPW) was obtained from a MilliQ purification system (Millipore). Acids and ammonia were purified by sub-boiling distillation in a quartz still. The anion exchanger was analytical grade macroporous 100–200 mesh AG MP-1 (Bio-Rad) resin. The reversed phase sorbent was Merck silica C18 . The sodium salt of EBT was purchased from Aldrich and dissolved in water. The solution was filtered through 0.45 ␮m teflon filters (Nalgene). Ammonium acetate buffer was prepared by adding 5 ml of acetic acid and 25 ml of ammonia to HPW and diluting to a final volume of 250 ml. On-line mixing of this solution with the ligand and the samples gives a final pH value of 10.0 ± 0.2. The wash solution was simply the buffer diluted 100-folds. 8-hydroxyquinoline (Fluka) was dissolved in a few ml of methanol and diluted with water to 3 × 10-3 M. Manganese standard solutions were prepared by dilution of 1000 mg/l Merck Titrisol stock solution. The eluent was 0.1 M HCl, prepared by dilution of concentrated acid purified by sub-boiling distillation. Magnesium nitrate 99.995% was purchased from Aldrich. Seawater was sampled in Ross Sea (Antarctica) during the 1997–98 Italian expedition within the Italian Research Programme on Antarctica (“Programma Nazionale di Ricerche in Antartide”, PNRA). It was immediately filtered through 0.45 ␮m filters and kept frozen during transportation and storage. Before use it was de-frozen, acidified with HNO3 (1 ml/l) and left at least 12 h to equilibrate. Standard reference seawater NASS-3 (seawater from open oceanic origin) from National Research Council, Canada, was used for accuracy assessment [22]. The frits for the enrichment column were adapted from Dionex bed supports for 4 mm i.d. columns. Teflon net was purchased from Zuricher Beuteltuchfabrik (Ruschlikon, Switzerland). The atomic absorption spectrophotometer (ETAAS) was a Perkin-Elmer 5100 with Zeeman-effect background correction. Pyrocoated graphite furnaces were used with wall or platform atomisation for

C. Sarzanini et al. / Analytica Chimica Acta 435 (2001) 343–350

345

Table 1 Heating cycle for manganese platform (P) or wall (W) atomisation Step

1 2 3 4 5 6

Temperature (◦ C)

Ramp time (s)

Hold time (s)

P

W

P

W

P

W

90 130 1400 200 2200 2600

120 – 1200 200 2200 2600

10 20 60 5 0 1

10 – 20 5 0 1

10 30 20 5 4 5

20 – 20 5 3 5

high and low (<1 ␮g/l) concentrations, respectively. Table 1 shows the heating cycle adopted in the two cases. Measurements were performed at 279.5 nm with a 0.20 nm slit. A matrix modifier consisting of 1.95 × 10−2 M Mg(NO3 )2 in water was injected into the furnace together with samples and standards (0.05 mg for each analysis). During ETAAS measurements, auto-zero was performed on 0.1 M HCl in the presence of the modifier. The measured peak-area typically was 0.011 A s (integrated absorbance). For comparison, after subtracting this value, the peak area was 0.072 A s for a 0.5 ␮g/l Mn standard solution and 0.048 A s for an eluate obtained from enrichment of NASS-3. Preconcentration was performed with the aid of a Perkin-Elmer FIAS-200 apparatus, equipped with two peristaltic pumps and a four-way switchable valve. Aspiration and valve switching are softwarecontrolled. Tygon pump tubings were used (i.d. 1.40 mm for sample and wash solution and 0.76 mm for ligand, buffer and acid). Three more manual valves were added to the preconcentration manifold, whose structure is shown in Fig. 1. Connections between containers, pump tubings and valves were made with teflon tubing. Column capacity measurements were performed with a Varian Liberty 100 inductively coupled plasma atomic emission (ICP AES) spectrophotomer. pH values were measured with a Mettler Delta 320 pH meter equipped with a combined glass–calomel electrode. Polyethylene labware was washed by immersion in 1 M HCl for 48 h, rinsed and stored in 1×10−2 M HCl until the moment of use. Sample handling was performed under a laminar flow hood in a laboratory provided with filtered air.

Fig. 1. Preconcentration manifold. Mode: (A) sample loading; (B) analyte elution. Components: a, b, c, d and e: sample, washing solution, eluent, ligand and buffer vessels; P1 and P2: peristaltic pumps; V1, V2, V3 and V4: valves; g: enrichment column; h: outlet to eluent collection vessel; W: outlet to waste.

2.2. Procedure 2.2.1. Enrichment column A new kind of column was designed for solid phase extraction (Fig. 2). The column consists of a teflon tube core, with which sorbent and samples come in contact, placed inside a plexiglas housing. The desired amount of sorbent, in this case AG MP-1, is slurryloaded into the 40 mm long, 2 mm i.d. teflon tube and is kept in place with two frits (cut of the exact size

346

C. Sarzanini et al. / Analytica Chimica Acta 435 (2001) 343–350

Fig. 2. Enrichment column: 1, column housing (central body); 2: column housing (external fittings); 3: teflon tube filled with sorbent; 4: connections to preconcentration manifold; 5: cones; 6: frits.

with the aid of a punch) and two small plugs of teflon net (diameter 4 mm), again obtained with the punch. The latter act as a safety device, like a sort of “filters” in order to prevent sorbent leaking. This phenomenon did not take place with AG MP-1. Only the frit could be used, but the teflon net was added to improve the column packing stability. The tube, which can be completely or partially filled with the sorbent, is inserted into the central part of the plexiglas housing. Connection between substrate and samples is achieved with two pieces of teflon tubing, which are kept in place, i.e. in direct contact with the teflon plugs, by two Omnifit inverted cones, which are set between the central and the two external parts of the housing. The three plexiglas parts are connected by screwing. The total length of the column is 75 mm. The

opposite side of the teflon tube is flanged and inserted in a threaded fitting which allows its connection to the preconcentration valve. The same column packing was used for more than 200 experiments giving undecreased performance. 2.2.2. Preconcentration Two FIAS programs for loading the desired sample volume through the column and elute the retained metals were set up (see Table 2). In the first step, the sample is mixed on-line with the ligand and the buffer, which in turn are mixed in a T-joint. The solution flows through the enrichment column, where EBT complexes are retained, and then to the waste. A portion of loaded sample is collected from the waste outlet in order to ensure that on-line mixing gives the

Table 2 FIAS program for sample preconcentration Step

Time (s)

Pump 1 (rpm)

Pump 2 (rpm)

Valve position

Description

Load-wash 1 2 3 4 5 6 7 8

cycle 60na 10 60 60 60 60 10 60

20 0 20 0 20 20 0 20

20 0 20 0 20 20 0 0

Fill Fill Fill Fill Fill Fill Fill Fill

Sample loading Pausea Sample loading and filling of elution tube Pausea Column washing with diluted buffer Column washing with diluted buffer Pausea Final washing with diluted buffer

Elution cycle 1 60na 2 10 3 99 4 10 5 80

0 0 0 0 30

20 0 20 0 0

Inject Inject Inject Fill Fill

Elution Pausea Washing with eluent Pausea Washing with dilute buffer

a Pauses are technical delays necessary to switch manual valves or to collect the eluate vessel. The first steps of loading and elution are repeated n times depending on the desired volume ratio.

C. Sarzanini et al. / Analytica Chimica Acta 435 (2001) 343–350

desired pH value. The duration of the loading step depends on the desired preconcentration ratio. In the meantime, the wash solution is aspirated by the eluent tube and driven to the waste. In the third step, valve 2 is switched and the eluent starts filling the corresponding tube, while sample loading continues. Then valve 1 is switched and the washing solution flows through the sample tube. In this step ligand and buffer keep on flowing and mixing with the tube output, which is still filled with the last aliquot of aspirated sample. In the last steps of the loading sequence pump 2 is stopped, the inlet of buffered ligand into valve 3 is closed and only washing solution flows through the system into the column. Before elution, the preconcentration valve is switched, so that the eluent flows through the column, the retained metal is stripped and the eluate is collected in a small vial. The column is then washed with a further aliquot of eluent. Meanwhile buffer and ligand keep flowing, because they are connected to pump 2, directly to the waste; if desired, such flow can be avoided by disconnecting the corresponding tubes. Of course, the sample inlet in valve 3 is closed. In the last step of the procedure, valve 2 is switched and the washing solution flows into the eluent tube and through the column, in order to condition the resin at the pH of the subsequent experience. The determination of manganese in the acidic eluate enabled the evaluation of metal recovery for each experiment. The development and optimisation of the technique was performed on seawater spiked with 2 ␮g/l of Mn. The volume ratio was 5, i.e. from 7.5 ml of loaded sample to 1.5 ml of eluate. The exact volumes of samples and eluate were obtained by weighing the corresponding vessels and taking into account the density of seawater (the density of the eluates was not significantly different from 1) before and after preconcentration, in order to take account of the slight variation in aspiration rates, normally occurring with peristaltic pumps, due to the wearing of the tubing and heating of the pumps. The flow rate was 1.3 ml/min. The resin bed volume was 40 ␮l. 2.2.3. Seawater analysis Seawater analysis was performed at pH 10 with a EBT concentration (after mixing with the sample) of 0.23 mM. The sample volume required for one replicate was 16 ml. The volume ratio was about 10. Three

347

injections of three 30 ␮l aliquots of eluate into the graphite tube of the ETAAS were programmed, each followed by solvent evaporation, in order to obtain a higher signal intensity. This procedure is possible owing to the removal of the sample matrix which occurs during the loading step of preconcentration. The modifier was added only once. The concentrations were evaluated with the standard addition method by performing the enrichment procedure on seawater as such and spiked with a known amount of manganese. 2.2.4. Column capacity Three 150 ml water samples, respectively containing 5, 10, 20 mg/l of Mn in the presence of a 10-fold excess of ligand and buffered at pH 10, were prepared. Aliquots of 100 ml of such solutions were put in contact with the same amount of resin used to pack the preconcentration column and shaken overnight. A blank was run in parallel. Afterwards the solutions were filtered through polypropylene columns (Bio-Rad Poly-Prep Columns), in order to remove the resin, and added with 14 M nitric acid (1 ml of acid to 9 ml of the solution). Before acidification the samples, originally containing 10 and 20 mg/l of manganese, were diluted 2- and 4-folds, respectively, in order to avoid ligand precipitation at low pH. The concentrations of manganese left after contact with the resin were determined by ICP AES by comparison with the ones contained in the original samples (i.e. the 50 ml unused aliquots) acidified and diluted in the same way.

3. Results and discussion All experiments were performed in triplicate and the results were corrected for the blank. The repeatability was typically 3% for a concentration level of 2 ␮g/l. 3.1. Column design The column structure was designed in order to have a versatile, inert and low cost support. The possibility of filling the internal tube to the desired length allows to use it for different kind of sorbents, e.g. for low capacity ones, where a larger amount is required, and for the ones with small particle size such as silica C18 , where a small quantity of material must be used in

348

C. Sarzanini et al. / Analytica Chimica Acta 435 (2001) 343–350

order to avoid backpressures. The sorbent volume can also be changed depending on the concentrations in the sample, i.e. on the amount of active sites required for retention. Samples and eluents come in contact with teflon parts only, since the plexiglas housing simply provides a physical external support. Therefore, it is possible to work with concentrated acids, even HF if required, or strong organic solvents such as acetonitrile. The cost of one plexiglas house is around US$ 40 and lasts indefinitely; each housing can be used for different packings simply by changing the internal tube. 3.2. Preliminary experiments Some preliminary experiments were performed in order to set up the conditions for the optimisation study. These experiments were made at pH 9 with concentrations of 6.3 mM EBT and 10 ␮g/l of Mn. Elution of the retained metal with 2 M HNO3 and 0.1 M HCl gave similar recoveries, therefore the use of the latter was preferred, mainly because it causes less wearing of peristaltic pump tubing and graphite furnace than the more concentrated nitric acid. No alternative eluents were tested since our previous experiences showed that the two considered reagents are suitable for stripping the metal complexes from anion exchange resins [23,24]. Two subsequent elutions with 1.5 ml of 0.1 M HCl on the same loaded column confirmed that no manganese was present in the second aliquot. Since literature indications for titrimetric determination of manganese with EBT as indicator [25] involved the addition of ascorbic acid, in order to avoid ligand oxidation, the effect of the latter on preconcentration efficiency was investigated. No differences in metal recoveries were found with and without ascorbic acid. In the light of these results, the next experiments were performed without ascorbic acid; however, EBT and the buffer were mixed on-line in order to reduce the effect of pH on ligand oxidation. Metal recovery was found to be constant starting from samples containing the same amount of Mn in different volumes (namely, 7.5, 3 and 1.5 ml of 2, 5, and 10 ␮g/l solutions, respectively) and eluting with the same volume of acid (1.5 ml). The procedure

optimisation was subsequently performed on 2 ␮g/l samples with a volume ratio of 5. The effect of the saline matrix was also investigated and showed that manganese recoveries are independent on ionic strength. Therefore, the optimisation for the procedure was directly performed in seawater, taking into account that a relatively high ionic strength improves the interaction between the complex and the resin through a salting-out effect. The flow rate was chosen according to our experience on similar preconcentration procedures involving AG MP-1 and sulphonated ligands [23,24]. A relatively slow rate of 1.3 ml/min ensures a proper contact time between the resin and the complex and thus allows the retention of the latter. 3.3. Optimisation of the procedure 3.3.1. Effect of pH Fig. 3 shows the recovery of manganese as a function of pH. As it can be seen, the yields increase with pH values, owing to the increase in complex stability. However pH values higher than 10 were not chosen to avoid the precipitation of magnesium hydroxide. The subsequent experiments were performed at pH 10, where the highest recovery (83.4%) was observed. The lack of a quantitative recovery is not uncommon in flow-injection procedures [1] and such procedures can be applied for analysis provided the recovery is reproducible. The recovery is directly taken into account when analysis of real samples is performed with the standard addition method.

Fig. 3. Recovery of manganese as a function of pH.

C. Sarzanini et al. / Analytica Chimica Acta 435 (2001) 343–350

3.3.2. Effect of ligand concentration The recovery with different ligand concentrations, corresponding to ligand to metal ratios 250, 200, 100, 50 and 10, was found to be constant for Mn concentration of 2 ␮g/l. Therefore, any of these values is suitable for analysis. In order to avoid a competition between the free ligand and metal complex for the resin sites, during the retention, and a background enhancement, due to a possible release of ligand during the elution, a concentration of 0.23 mM (corresponding to the 10-fold excess after mixing) was adopted. It must be mentioned that for real samples with metal concentration at trace and ultratrace levels, this value ensures the complete complexation of the analyte. 3.3.3. Column capacity The capacity of the column toward the retention of manganese was determined and it was found to be 0.15 ± 0.01 meq Mn/ml resin bed. This value corresponds to the amount of metal bound to EBT, which is the species actually retained by the resin. A microcolumn such as the one used in this study is able to retain 0.127 ± 0.008 mg of Mn: this value is well above the content of Mn in 100 ml of seawater (typically 5 ng) or in river water. In case concentrations, exceeding the capacity of the microcolumn, are to be determined, a larger amount of resin can be used, owing to the versatility of the column packing; on the other hand in this case the procedure with EBT and AG MP-1 can be used only in order to remove the matrix and not for enrichment purposes. 3.3.4. Analysis of seawater The accuracy of the procedure was tested by analysing a certified seawater sample, namely NASS-3. The concentration found for Mn with our technique (n = 4) was 27 ± 3 ng/l, which compares well with the certified value of 22 ± 7 ng/l: there are no statistical differences at the 95% confidence level between the two data. The time required for one cycle of preconcentration and determination of manganese at this concentration level is about 22 min. We worked with relatively high volumes in order to minimise the error in their measurement, because time was not a concern in our study. Anyway our method can be used with lower volume, like similar procedures developed by other authors [6,14,19], thus reducing the analysis time.

349

Moreover, the developed procedure was compared with another one, commonly used in our laboratory for seawater analysis and based on a previously developed method [10], modified in order to be adapted to FIAS. The method is based on the complexation with 8-hydroxyquinoline at pH 9, retention onto silica C18 and elution with a small volume of acetonitrile or 0.1 M HCl. Also with this technique the preconcentration was performed with FIAS following the steps described above. Antarctic seawater was analysed with both procedures. The manganese concentration found with AG MP-1/EBT technique was 83 ± 5 ng/l, in good agreement with the value of 77 ± 2 ng/l found with silica C18 /8-hydroxyquinoline. Both methods are therefore suitable for seawater analysis. The absolute blank level was 0.16 ± 0.01 ng. The detection limit, evaluated as three times the standard deviation of the blank, was 2 ng/l (for a volume ratio of 10). Lower concentrations could be determined by adopting higher volume ratios, provided the blank is further reduced. 4. Conclusions The enrichment procedure developed allows to determine manganese at ultratrace levels with accuracy in natural waters, as confirmed by the analysis of certified seawater. The enrichment column designed in this study has the advantages of versatility, low cost and compatibility with most eluents. Acknowledgements Financial support from the Ministero dell’Università e della Ricerca Scientifica e Tecnologica (MURST, Rome), from the Italian National Research Council (CNR, Rome) and from ENEA under the Italian Research Programme on Antarctica (Programma Nazionale di Ricerche in Antartide) is gratefully acknowledged. References [1] Z. Fang, Flow-Injection Separation and Preconcentration, VCH, Weinheim, 1993. [2] Y. Liu, J.D. Ingle Jr., Anal. Chem. 61 (1989) 520.

350

C. Sarzanini et al. / Analytica Chimica Acta 435 (2001) 343–350

[3] R.A. Nickson, S.J. Hill, P.J. Worsfold, Anal. Chim. Acta 351 (1997) 311. [4] S.D. Hartenstein, J. Ruzicka, G.D. Christian, Anal. Chem. 57 (1985) 21. [5] S. Nakashima, R.E. Sturgeon, S.N. Willie, S.S. Berman, Fresenius Z. Anal. Chem. 330 (1988) 592. [6] L.C. Azeredo, R.E. Sturgeon, A.J. Curtius, Spectrochim. Acta 48B (1993) 91. [7] P. Schramel, L. Xu, G. Knapp, M. Michaelis, J. Anal. Chem. 345 (1993) 600. [8] K.W. Warnken, G.A. Gill, L.S. Wen, L.L. Griffin, P.H. Santschi, J. Anal. Atom. Spectrom. 14 (1999) 247. [9] O. Abollino, E. Mentasti, V. Porta, C. Sarzanini, Anal. Chem. 62 (1990) 21. [10] V. Porta, C. Sarzanini, E. Mentasti, O. Abollino, Anal. Chim. Acta 258 (1992) 237. [11] X. Peng, Z. Jiang, Y. Zen, Anal. Chim. Acta 283 (1993) 887. [12] K.S. Huang, S.J. Jiang, Fresenius J. Anal. Chem. 347 (1993) 238. [13] S.N. Willie, H. Tekgul, R.E. Sturgeon, Talanta 47 (1998) 439. [14] E. Ivanova, K. Benkhedda, F. Adams, J. Anal. Atom. Spectrom. 13 (1998) 527.

[15] M.C. Yebra-Biurrun, A. Bermejo-Barrera, M.P. BermejoBarrera, M.C. Barciela-Alonso, Anal. Chim. Acta 303 (1995) 341. [16] D. Ye, H. Zhang, Q. Jin, Talanta 43 (1996) 535. [17] H. Zhang, X. Yuan, X. Zhao, Q. Jin, Talanta 44 (1997) 1615. [18] R. Ma, F. Adams, Anal. Chim. Acta 317 (1995) 215. [19] K.A. Tony, S. Kartikeyan, B. Vijayalakshmy, T. Prasada Rao, C.S. Padmanabha Iyer, Analyst 124 (1999) 191. [20] O. Abollino, M. Aceto, M.C. Bruzzoniti, E. Mentasti, C. Sarzanini, Anal. Chim. Acta 375 (1998) 299. [21] O. Abollino, M. Aceto, C. Sarzanini, E. Mentasti, Anal. Chim. Acta 411 (2000) 223. [22] NASS-3, Open Ocean Seawater Reference Material for Trace Metals, National Research Council, Ottawa, Canada. [23] O. Abollino, C. Sarzanini, E. Mentasti, A. Liberatori, Spectrochim. Acta 49A (1993) 1411. [24] M. Colognesi, O. Abollino, M. Aceto, C. Sarzanini, E. Mentasti, Talanta 44 (1997) 867. [25] E. Bishop (Ed.), Indicators, Pergamon Press, Oxford, 1972.