S e&imifo AM Vol. 48B, No. 2, pp. 193498.1993 l&d in Great B&n.

05%8547/93 %.@I + .oo 0 1993 Pergamml Press Ltd

Reference materials for quality assurance in sea-water analysis: performance of total-reflection X-ray fluorescence in the

intercomparison and certification stages

PETER FREIMANN, DIE-I-HER SCHMIDT and ASTRID NEUBAUER-ZIEBAIUH Bundesamt fur Seeschiffahrt und Hydrographie, Postfach 301220, D-2000 Hamburg 36, F.R.G.

(Received 13 July 1992; accepted 13 October 1992)

Abstract-The certification of a sea-water reference material (CRM 403) was completed by the Community Bureau of Reference (BCR) of the Commission of the European Communities during an intercomparison exercise in which we were participants along with other highly experienced laboratories, who, beforehand, had given evidence of outstanding performance. Further, we participated in a feasibility study on estuarine water. In both studies we used total-reflection X-ray fluorescence analysis (TXRF) for the determination of V, Mn, Fe, Co, Ni, Cu, Zn, Pb and U on the nmohkg level. The enrichment of the trace metals and the separation from the salt matrix were performed by complexation with sodium dibenxyldithiocarbamate and reverse-phase chromatography. In this paper, the high performance of our TXRF results is compared to other analytical techniques like voltammetric and atomic absorption methods.

1. INTRODUCTION

THE BUNDESAMT fur Seeschiffahrt und Hydrographie (BSH, Federal Maritime and Hydrographic Agency) is responsible for the determination of a large number of different pollutants in the marine environment especially in the North Sea and the Baltic Sea. The BSH is involved in different national and international marine monitoring programs.

For the determination of vanadium, manganese, iron, cobalt, nickel, copper, zinc, lead and uranium in sea-water, we mainly use the relatively new method of total- reflection X-ray fluorescence analysis (TXRF) for our routine work.

To control the quality of our TXRF determinations, we participated in two different experiments conducted by the Community Bureau of Reference of the Commission of the European Communities. The first experiment was a certification campaign that allowed production of a sea-water reference material (CRM). The second experiment was a feasibility study in the form of an intercomparison of an estuarine water sample as a first step in producing an estuarine water reference material.

2. EXPERIMENTAL

2.1. Certijication of the sea-water reference material 2.1.1. Sample preparation, homogeneity and stability. Approximately 2600 1 of sea-water from

the central part of the Southern Bight of the North Sea were collected in a pre-cleaned linear polyethylene (LPE) storage tank. The sample was filtered using 0.45 pm membrane filter and acidified to pH 1.5 using nitric acid (purified by sub-boiling distillation). LPE sampling bottles (2 1 volume) were directly filled from the LPE tank using a peristaltic pump.

Homogeneity and stability tests had been performed for Cd, Cu, Pb and Zn by four different laboratories prior to the certification campaign. (For more details about sample preparation, homogeneity and stability tests see Ref. [l].)

2.1.2. Analytial m&o& for sea-water analysis. Each of the 18 European laboratories that participated in the certification exercise was requested to make a minimum of five independent replicate determinations. Table 1 gives an overview of the different methods applied and defines the acronyms that will be used below.

We obtained four 2 1 bottles and made six replicate determinations of samples from each bottle.

193

194 P. FREIMANN er crf.

Table 1. Methods of final detection used for the certification of sea-water reference material (CRM 403) and for the intercomparison of trace metals in estuarine water

Method Acronym Sea water Estuarine water

laboratory no. laboratory no.

Adsorptive differential pulse cathodic stripping ADPCSV 6.8 voltammetry

Cathodic stripping voltammet~ csv 5 Differential pulse anodic stripping voitammetry DPASV S,lO,ll 9-30 Differential pulse cathodic stripping voltammetry DPCSV 9 Electrothermal atomic absorption spectrometry ETAAS 4,5,9,12 4,5,11,22,29 Flame atomic absorption spectrometty FAAS 4,12,13 3,11,17,19,23,25 Flow potentiometric stripping analysis FPSA 10 Inductively coupled plasma atomic emission mass ICP-MS 2

spectrometry Isotope dilution mass spectrometry Total-reflection X-ray fluorescence analysis ETAAS with Zeeman background correction

IDMS 16 TXRF 18 12

ZETAAS 1,3,13,14,15 1,3,8,10,23,25,26

2.2. Intercomparison of trace elements in estuarine water 2.2.1. Sample preparation, homogeneity and stability. Two batches of estuarine water samples

with different salinities taken from the Portuguese River Tagus (Tejo) estuary were collected in 300 I LPE tanks. Batch A had a practical salinity of 16.4 while batch B had a practical salinity of 7.4. The samples were filtered using 0.45 &rn filter cartridges and acidified to about pH 2 using hydrochloric acid (Merck, Suprapur). LPE sampling bottles (1 1 volume) were directly filled from the LPE tank using a peristaltic pump.

Homogeneity and stability tests had been performed for As, Cd, Cu, Pb and Zn by six different laboratories. More details about sample preparation, homogeneity and stability tests can be found in the study by KWER et al. [2].

2.2.2. Analytical methods for es&vine water analysk. Each of the 31 European laboratories that participated in the inter~mpa~son exercise was requested to make a ~nimum of five inde~ndent replicate dete~nations for batch A and batch B estuarine water samples. Table 1 gives an overview concerning the different methods applied.

Since we received just 1 1 of each batch, it was only possible to make four replicate determinations by TXRF. The results of batch B alone will be presented here to ensure that we do not unnecessarily lengthen the paper.

2.3. TXRF measurements 2.3.1. Apparatus. The TXW spectrometer consists of an EXTBA II module (Rich. Seifert,

D-2070 Ahrensburg) with double beam excitation (molybdenum and tungsten tubes), an X-ray generator (Iso-Debyeflex lOOl), a 30 mm2 Si(Li) detector with a resolution of 160.5 eV at 5.9 keV, an automatic sample changer and a computer controlled multichannel analyser system (Link Systems AN 10000).

2.3.2. Preconcentration of trace metals. The trace metals are preconcentrated and separated from the sea-water and estuarine water salt matrix by reverse-phase chromatography as described by PRANGE [3] and FREIMANN and SCHMIDT [4]. Two hundred grams of each water sample is spiked with 2.5 &kg Se (Merck, AAS-Standard solution) as internal standard, adjusted to pH 4.8 with sodium hydroxide solution (Merck, Suprapur) and acetate buffer (Merck, Suprapur). The trace metals are complexed by the addition of a methanolic solution of sodium dibe~yldithiocarbamate (NaDBRTC, Fluka). The carbamate complexes are adsorbed on a reverse-phase column (Chromosorb W, Merck).

After drying the column, the complexes are eluted with 3 ml of a mixture of chloroform/methanol (purified by sub-boiling distillation). One hundred microlitres of this solution is dried by means of a special device [5] to achieve a thin film on the silica glass sample carrier.

Using molybdenum excitation and a measuring time of 3000 s vanadium, manganese, iron, cobalt, nickel, copper, zinc, lead and uranium could be determined.

For analytical blanks, 200 g of ultrapure water (MilliQ, ~li~re) are provided with 500 mg sodium chloride (Merck, Suprapur) followed by the procedure for the determination of trace metals in sea-water and estuarine water as described above. All sample treatments are carried

TXRF for quality assurance in sea-water analysis 1%

Volt8minetric Method8

Atomic Abmrptiom Meted

q Totd Reflectkon X-ray FIuoraeenee

Other Methods

S

4

g3 - 2

4% 0

1% 1 3

Lab Namber

6

% 4

12

zi! 0

a 4 5 9 12 38 3 13 4 15

Lab Number

a

6 12 13 17 16 18 1 14 15

Lab Number

L8b Namber

6

6 8 1.114 5 91216183131415

Lab Number

L8b Number

Fig. 1. Comparison of the vanadium, manganese, iron, nickel, copper, zinc and lead content of the sea-water reference material measured by different methods.

out in a laminar flow clean bench (Class 10, U.S. Federal Standard 209b) installed in a clean room laboratory (Class 100).

3. RESULTS AND DISCUSSION

3.1, Trace elevens in sea-water 3.1.1. Vanadium. Only two laboratories measured this element. The results from

the two laboratories, which used cathodic stripping voltammetry and TXRJ?, respectively, agree very well (see Fig. 1). But the number of data sets is insufficient for the certification of vanadium. We found 22.5 +- 3.9 nmol/kg V (arithmetic mean 2 standard deviation).

3.1.2. Manganese. Four laboratories measured manganese. The results of three laboratories were accepted by the BCR (see Fig. l), but the number of data sets is insufficient for the certification of manganese. We found 3.82 it 0.92 nmolkg Mn (n = 24).

1% P. FREIMANN et al.

3.1.3. Iron. Four laboratories measured iron (see Fig. 1). The standard deviations of the results are very high probably owing to contamination problems. Therefore, the iron content could not be certified. TXRF measurements showed that one of the four bottles we acquired seemed contaminated by iron. This contaminated bottle had an iron content of 8.3 + 1.9 nmoYkg (n = 6) in comparison to the other three bottles with an iron content of 3.5 ? 1.3 nmol/kg (n = 18).

3.1.4. Cobalt. Cobalt has only been measured by TXRF and therefore a comparison with other analytical techniques is not possible. We found 0.18 f 0.03 nmol/kg Co (n = 24).

3.1.5. Nickel. Eleven laboratories sent their Ni results to the BCR. Of these, ten were accepted (see Fig. 1). The Ni content was certified to 4.36 + 0.36 nmolikg. Our TXRF result was 4.72 t 0.48 nmoYkg (n = 24).

3.1.6. Copper. Fourteen laboratories measured copper. The BCR accepted thirteen data sets measured by seven different methods (see Fig. 1).

The Cu content was certified to 3.90 f 0.37 nmoYkg. We found 3.86 rt 0.51 nmoYkg (n = 24) by TXRF.

3.1.7. Zinc. Zinc has been measured by twelve laboratories. The BCR accepted nine data sets measured by six different methods (see Fig. 1). The Zn content was certified to 25.7 + 2.9 nmol/kg. Our TXRF result for Zn was 28.2 f 5.2 nmol/kg (n = 24).

3.1.8. Lead. Ten laboratories sent their Pb results to BCR from which nine data sets have been accepted (see Fig. 1). The Pb content was certified to 0.117 + 0.025 nmol/kg. We found 0.112 + 0.043 nmol/kg Pb (n = 24).

3.1.9. Uranium. Uranium has only been measured by TXRF and therefore a comparison with other analytical techniques is not possible. We found a uranium content of 14.0 + 1.3 nmoYkg (n = 24).

3.2. Trace elements in estuarine water (batch B) 3.2.1. Vanadium. Vanadium has only been measured by TXRF and therefore a

comparison with other analytical techniques is not possible. We found a vanadium content of 25.3 it 0.6 nmoYkg (n = 4).

3.2.2. Manganese. Five laboratories measured manganese using FAAS, ZETAAS and TXRF. The results of four laboratories were accepted by the BCR (see Fig. 2). Our TXRF result for Mn was 10.4 + 1.5 nmoYkg (n = 4).

3.2.3. Iron. Seven laboratories sent their iron results to the BCR. Two data sets were rejected due to extremely high values. The remaining five data sets are presented in Fig. 2. It is apparent that in two cases, blank problems led to high standard deviations of the mean values. We found an iron content of 40.9 ? 2.6 nmol/kg (n = 4).

3.2.4. Cobalt. Five laboratories measured this element. The BCR accepted four data sets measured by CSV, ETAAS, ZETAAS and TXRF (see Fig. 2). The TXRF data (0.77 ? 0.08 nmol/kg (n = 4)), appears to be too high when compared to the other analytical techniques.

3.2.5. Nickel. Thirteen laboratories sent their nickel results to the BCR. Of these, three data sets were rejected due to extremely high values. The remaining ten data sets are presented in Fig. 2. The methods applied were CSV, DPCSV, ETAAS, ZETAAS and TXRF. We found a nickel content of 21.8 + 0.5 nmol/kg (n = 4).

3.2.6. Copper. Seventeen laboratories measured copper. Two extremely high data sets and one extremely low data set were rejected by the BCR. The remaining fourteen data sets are presented in Fig. 2. Seven laboratories measured copper by ZETAAS, three by ETAAS, and one each by ASV, DPASV, FAAS and TXRF. Our TXRF result was 33.8 * 1.5 nmol/kg (n = 4).

3.2.7. Zinc. Fourteen laboratories sent their zinc results to BCR, from which one data set was rejected due to the extremely high values (mean: 2459 nmol/kg). The remaining thirteen data sets show a relatively large scatter between 65 and 178 nmol/kg (see Fig. 2) in comparison to the relatively low scatter of the zinc data in the

TXRF for quality assurance in sea-water analysis 197

Voltammotrk Methods

Atomk Absorpth Methuds

Total Rdketion X-ray Fb~oreaeenee

Other Methods

Lab Number

30 9 29 12 3 8 10 10 23 25 30 9 4 5 29 17 12 3 8 10 10 23 25 26

Lab Number I Lab Nnmber

150

i 100 . t8 50

0 309 3 1117192325212 1826

Lab Number ____~

11 12 3 1

Lab Number

‘,mT T c 0,75-

1 OW

8 0$25.-

““- 30 11 12 8

Lab Number

B 096

4 0,4

g 05 n_a

12 3 8 10 10 25

Lab Number .J

Fig. 2. Comparison of the manganese, iron, cobalt, nickel, copper, zinc and lead content of estuarine water measured by different methods.

certification campaign for sea-water. This effect may be explained by methodical problems owing to the high dissolved organic matter content of the estuarine water samples. The methods applied were ASV, DPASV, FAAS, ICPMS, ZETAAS and TXRF. We found 102 + 5 nmol/kg Zn (n = 4) by TXRF.

3.2.8. Lead. Ten laboratories measured lead, from which three extremely high data sets and one extremely low data set were rejected by the BCR. The results of the remaining six laboratories are presented in Fig. 2. The two highest values in the figure show the highest standard deviations. This effect may be explained by contamination problems. Our TXRF result for lead was 0.28 + 0.06 nmoVkg (n = 4).

3.2.9. Uranium. Uranium has only been measured by TURF and therefore a comparison with other analytical techniques is not possible. We found a uranium content of 4.57 f 0.43 nmoVkg (n = 4).

4. CONCLUSIONS

TXRF has successfully contributed, as a relatively new analytical method, to the certification exercise for copper, nickel, zinc and lead in sea-water reference material

198 P. FREMANN et al.

(CRM 403). A report has been published by QUEVAUVILLER et al. [6]. For other elements, which could not be certified, TXRF results show good agreement with other analytical techniques. In the intercomparison exercise for trace elements in estuarine water, TXRF demonstrates high performance (accuracy and precision) in comparison to other analytical techniques such as the voltammetric methods, the different kind of atomic absorption and mass spectrometry.

REFERENCES

[l] Ph. Quevauviller, K. J. M. Kramer, K. Vercoutere and B. Griepink, EUR Report 14061, Luxembourg (1992).

[2] K. J. M. Kramer, E. M. van der Vlies, W. S. Dorten, K. R. Sperling and T. Vinhas, TNO-Report R90/327a, Delft, The Netherlands (1991).

[3] A. Prange, Ph.D. Thesis, Hamburg University, F.R.G. (1983). [4] P. Freimann and D. Schmidt, Spectrochim. Acta 44B, 511 (1989). [S] A. Prange, A. Kn(iche1 and W. Michaelis, Anal. Chim. Acta 172, 79 (1985). [6] Ph. Quevauviller, K. J. M. Kramer, E. M. van der Vhes, K. Vercoutere and B. Griepink, Mar. Pollut.

Bull. 24, 33 (1992).