determination of total mercury in environmental and biological samples by flow injection cold vapour...

SPECTROCHIMICA ACTA

PART B

E L S E V I E R Spectrochimica Acta Part B 51 (1996) 1867-1873

Determination of total mercury in environmental and biological samples by flow injection cold vapour atomic absorption spectrometry 1

J ames M u r p h y , Phil Jones , S teve J. Hil l*

Department of Environmental Science, University of Plymouth, Drake Circus, Plymouth, PL4 8AA, UK

Received 20 December 1995; accepted 27 June 1996

Abstract

A simple and accurate method has been developed for the determination of total mercury in environmental and biological samples. The method utilises an off-line microwave digestion stage followed by analysis using a flow injection system with detection by cold vapour atomic absorption spectrometry.

The method has been validated using two certified reference materials (DORM-1 dogfish and MESS-2 estuarine sediment) and the results agreed well with the certified values. A detection limit of 0.2 ng g-l Hg was obtained and no significant interference was observed. The method was finally applied to the determination of mercury in river sediments and canned tuna fish, and gave results in the range 0.1-3.0 mg kg -1.

Keywords: Cold vapour atomic absorption spectrometry; Flow Injection; Mercury; Microwave digestion

I. Introduction

The determination of mercury in the natural environment has long been recognised as important when assessing environmental quality. However, the accurate determination of mercury in real samples is not without its difficulties and precautions are required at all stages of the analysis. In addition, the direct determination by atomic spectroscopy can be problematic. For example, the determination of mercury by flame atomic absorption spectrometry is limited to high concentrations of mercury, owing to the poor detection limit offered by the technique (200/~g l-t), whilst the use of electrothermal atomic

* Corresponding author. 1 This paper was published in the Special Issue of Spectrochimica

Acta, Part B, devoted to Flow Analysis.

absorption spectrometry (ETAAS), although offering better detection limits (2/~g l-t), suffers from matrix interference since the high volatility of mercury restricts the ashing temperature [1].

One of the most common analytical approaches for the determination of total mercury at lower concentrations is cold vapour atomic absorption spectrometry (CVAAS). This approach offers a detection limit in the order of 0.008/~g 1-1 [2], based upon the reduction of inorganic mercury by stannous or borohydride compounds. However, the technique has two primary prerequisites if it is to be used successfully and volatilisation losses are to be avoided. These are (i) sufficient oxidation of the organic matter present in the sample to liberate the mercury from the matrix, and (ii) the quantitative production of a single labile mercury species (e.g. Hg(II)) which can afterwards be reduced quantitatively to Hg(0) for spectrometric evaluation. If these conditions are not achieved

0584-8547/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved PII S0584-8547(96)01571-6

1868 J. Murphy et al./Spectrochimica Acta Part B 51 (1996) 1867-1873

serious underestimation of total mercury concentration may result [3].

It is generally agreed that oxidative conversion of all forms of mercury in environmental and biological samples to Hg(II) is necessary prior to reduction to elemental Hg [4-6]. Sodium borohydride is often pre- ferred as the reducing agent because it offers rapid liberation of the mercury vapour from the liquid phase, and is effective with organic compounds that are not reduced to the metal by tin(lI) chloride [7].

The direct determination of mercury in environmental and biological materials may be problematic as a result of the matrix effects encountered, particu- larly with solid samples. The accurate determination of mercury in these materials therefore often requires decomposition of the matrix and conversion of mercury to the inorganic divalent form. Two types of decomposition methods have been employed for this purpose, i.e. wet digestion and dry ashing [8,9]. Dry ashing is not suitable for mercury determinations because of the low boiling point of mercury and its associated compounds. Consequently, wet digestion using either relatively low temperatures (<100°C) or higher temperatures (230°C) with an appropriate selection of oxidising acid mixtures has been successfully employed. However, such methods can be quite lengthy owing to the many steps required and may be prone to contamination. Thus, wet digestion methods have been increasingly succeeded by microwave assisted digestion methods. This approach offers rapid sample preparation and reduced risk of contamination, and has previously been employed for a number of environmental and biological matrixes [10-12].

The aim of this study was, therefore, to develop an accurate and sensitive method for determining mercury in environmental and biological samples using a microwave digestion and utilising flow injection analysis (FIA) with on-line cold vapour generation and electrothermal atomic absorption detection.

2. Experimental

2.1. Instrumental

A Perkin-Elmer Model 4100ZL ETAA spectro- meter was used with the furnace head removed and

replaced by a quartz cell in an electrically heated mantle. This was used in conjunction with a Perkin- Elmer Flow Injection Analysis System (FIAS 400) and an AS-90 autosampler. The hardware was con- trolled via an Epson EL2 computer with the Perkin Elmer FIAS-Furnace Version 7.21 software operated in the FIAS only mode. An Epson FX-850 printer was used to print results and peak signals. The measure- ments were made in the peak height mode. The electrically heated mantle contained a quartz tube which was maintained at 100°C to prevent condensation of water vapour in the quartz cell. The ETAA spectro- meter was set up using the optimised operating conditions shown in Table 1.

The flow injection system, shown schematically in Fig. 1, was set up using the conditions in Table 2. The flow rates were determined by measuring the uptake of water at the required pump speeds. The pre-fill step was set at 20 s to allow the solutions to flush through the tubing prior to the next analytical sample. These solutions were then allowed to flow to waste.

In Step 1, a sampling time of 10 s was used, during which the mercury sample/standard solution was drawn into the sample loop (500 #1) with the five- way valve in the "Fi l l" position. Meanwhile, the carrier solution was continuously passed through the manifold to the gas-liquid separator.

In Step 2, an injection time of 20 s was employed to allow the reaction to go to completion. The five-way valve was then switched to the "Inject" position and pump No. 1 stopped so that pump No 2 passed the carrier stream through the sample loop. This intro- duced the 500/~1 volume of sample into the system and transported it to the first section of the manifold.

At this point, the carrier stream and sample met the reductant. The reaction mixture continued through the reaction coil together with the gaseous mercury

Table 1 Instrumental operating conditions

Integration time Data processing Lamp Lamp current Slit width Wavelength Quartz cell dimensions Cell temperature

20 s Peak height 19-point smoothing Mercury hollow cathode lamp 6mA 0.7 nm 253.7 nm 160 mm × 7.5 mm (i.d.) 100*C

J. Murphy et aL/Spectrochimica Acta Part B 51 (1996) 1867-1873 1869

Reductant6.7 mtmin -1 i---]

Pump 2

Waste 10 ml min 4

Carrier 12.0 ml min -1 Pum 1

Waste 18 mT'min -1 L_~

Quartz Cell

[

Ar 100 ml min

5 way

vave [ ~ t

Manifold

Gas-liquid separator

Sample Uptake 8.3 ml min -1

O O O O O O O 0 O O O O O O O 0

O O O O O O Q ~ • O • O I F Autosampler

Fig. 1. Diagram of flow injection system.

vapour and hydrogen gas produced as a by-product. The mixture then flowed into the second section of the manifold where it met the argon carrier gas prior to entering the gas-liquid separator. The mercury vapour was transported with the argon carrier gas to the top of the gas-liquid separator, through a filter membrane (Perkin-Elmer Type B050-8306) and into a quartz cell placed in line with the mercury hollow cathode lamp. The filter membrane was placed between the gas-liquid separator and the quartz cell to prevent any aerosol droplets from reach- ing the cell which could decrease the precision of the measurement.

2.2. Microwave digestion

All the microwave digestions were performed with an unmodified 750W Tecnolec T200M microwave oven. The oven power (0-100% power in 25% incre- ments) and time range (0-30 min) were set manually.

Screw-capped 60 ml PTFE digestion vessels were used. These were cleaned prior to use by filling with concentrated nitric acid and placing in the microwave at full power for 60 s. The vessels were rinsed thoroughly with distilled water.

2.3. Reagents

All reagents used were of analytical grade (purchased from BDH Laboratory Supplies, Poole, Dorset, UK) unless otherwise stated. Doubly-distilled water was used throughout.

2.3.1. Reductant solution: 0.2% (w/v) NaBH4 in 0.05% (w/v) NaOH

The solution was prepared by first dissolving 0.5 g of sodium hydroxide pellets and then 2.0 g of sodium borohydride in 1000 ml of distilled water. The solution was stirred for 20 min and then filtered before use. This solution was prepared daily.

Table 2 Sequence programme

Step Time/s Pump 1/ Pump 2/ Valve Read No rpm rpm

Pre-fill 20 100 120 Fill 1 10 100 120 Fill 2 20 0 120 Inject Read

2.3.2. Carrier solution: 3% (v/v) HCl The solution was prepared by diluting 30 ml of con-

centrated hydrochloric acid with 1000 ml of doubly- distilled water.

2.3.3. Stabilising solution: 5% (w/v) K l a n 0 4

A stock solution was prepared by dissolving 5.0 g of potassium permanganate (Spectrosol grade, BDH


Ltd.) in 100 ml of doubly-distilled water. This stock solution was kept for 1 month in a dark glass bottle.

2.3.4. Standard mercury solutions The standard solutions of inorganic Hg(II) were

prepared by stepwise dilution from a stock standard solution containing 1001 ___ 2 mg 1 -I Hg as Hg(NO3)2"H20 (BDH Spectrosol). One to two drops of the stabilising solution was added to every 100 ml of standard solution. All the standards were freshly prepared prior to use.

2. 3.5. Dilu en t solution: 2 % (v/v) HNO 3 + 2 % (v/v) H 2SO 4 This solution was prepared by slowly adding 20 ml

of concentrated nitric acid and 20 ml of concentrated sulphuric acid to about 900 ml of doubly-distilled water, and finally making up to 1000 ml once the solution had cooled down.

All the samples and standard solutions were made up using the diluent solution. To prevent any loss of mercury due to adsorption onto the container walls, 1 ml of concentrated hydrochloric acid per 100 ml of sample solution was added. The hydrochloric acid concentration was kept to a minimum to prevent the premature reduction of the potassium permanganate.

2.3.6. Digestion mixture reagents The reagents used in the various digestion proce-

dures were concentrated sulphuric and nitric acid, and 30% hydrogen peroxide.

2.3.7. Carrier gas The carrier gas used in all experiments was argon

gas with a purity of < 99.95% (Cryospeed, BOC Limited, Plymouth, UK).

2.3.8. Certified reference materials The certified reference materials used in this study

were DORM-1 dogfish muscle reference material and MESS-2 estuarine sediment. Both of these CRMs were purchased from the National Research Council Canada, Ottawa, Canada.

3. Analytical procedure

The environmental/biological samples (~ 0.25 g) were weighed directly into the digestion vessel and

then the digestion mixture was carefully added to the vessel. The digestion mixture for the biological samples consisted of 5.0 ml of concentrated nitric acid and 1.0 ml of hydrogen peroxide, and for the environmental samples 2.5 ml of concentrated nitric acid, 2.5 ml of concentrated sulphuric acid and 1.0 ml of hydrogen peroxide. The hydrogen peroxide was added last and very slowly to prevent the initial reaction from occurring too vigorously. The vessel was left overnight to ensure that excessive pressure did not build up when placed in the microwave oven due to the oxidation of any organic matter in the sediment.

The next morning, the lid of the bomb was tightened and the bomb placed in the microwave oven for 60 s at medium power. After this time, the bomb was inspected to make sure that the seal was still intact. If the seal was damaged, the digestion was repeated, but if the seal was intact the bomb was irradiated in 30 s bursts for 4 min (5 min for the biological samples). The bomb was then left to cool for approximately 1-2 h.

The digestion mixture was then carefully transferred to a volumetric flask, together with washings from the bomb (2% nitric acid/2% sulphuric acid solution). Once the solutions had been made up to volume, the digestion solutions were de-gassed in an ultra- sonic bath for 30 min to remove any NOx which may be present (as a decomposition product of the nitric acid). This is necessary since NOx will scavenge the reductant before the complete reduction of the mercury in the reaction cell [13]. The samples were then transferred into the autosampler vials ready for analysis.

The system was calibrated with a series of eight Hg(II) standards up to 30 ng m1-1 Hg, although the linear calibration range may be extended to 60 ng m1-1. Using the optimised instrumental conditions (Table 1), the calculated detection limit defined as the mercury concentration corresponding to three times the standard deviation of the blank was 0.1 ng as Hg using a 500/.d loop.

The accuracy of the method was tested on the basis of recovery experiments made by adding 1 ml of 100 ng m1-1 Hg standard to each sample before digestion. It was calculated that the recoveries for the determination of mercury in sediments was 91.3-107.8% and in tuna fish was 91-93%.

J. Murphy et al./Spectrochimica Acta Part B 51 (1996) 1867-1873 1871

3.1. Sample collection and pre-treatment

3.1.1. Environmental samples Two sets of samples were collected to reflect both

high levels of mercury (resulting from anthroprogenic input) and low levels of mercury (reflecting the natural level). The River Mersey in Liverpool was chosen as the high level source and the River Avon in Devon was chosen as the low level source.

River sediments were collected at three selected sites along the River Mersey. The collected sediments were then placed in double-skinned polyethylene bags, sealed and transferred back to the laboratory. The same procedure was followed for the three selected sites along the River Avon estuary.

To prevent any exogenous contamination, all of the screw topped plastic bottles (Nalgene, Rochester NY, USA) used for sample collection and storage were cleaned with nitric acid and rinsed three times with doubly-distilled water prior to use.

The sample pre-treatment stage was based upon the Ministry of Agriculture, Food and Fisheries (MAFF) UK Method 43. It is important to dry the sample at room temperature, because at higher temperatures there is a danger of the volatiles being driven off - - including any mercury present in the sample. Thus, the samples were left to dry in air at room temperature until free flowing. The sample was then placed in a two-tower sieve system consisting of a top sieve of 2.00 mm and a bottom sieve of 710/~m. The top sieve removed any large stones, foreign matter and vegeta- tion. The sample that passed through this first sieve was then ground up and placed into the bottom sieve to leave a fine homogenous sample of river sediment, which could be used for the analysis stages.

Some of the original sample was retained to allow a weight loss experiment to be performed in order that the final results for mercury could be quoted on a dry mass basis.

3.1.2. Biological samples A fish sample was chosen, i.e. tuna fish, and a

sample purchased from a local supermarket. The tuna fish sample was pre-treated by freezing in a domestic freezer at below - 10°C for 6 h. After this time, the sample was transferred to a Freeze Dryer (Edward's Supermodulyo Model 12K, West Sussex, UK). The sample was placed in the vacuum chamber

which was evacuated to 0.8 × 10 -3 mbar and left there for 48 h. The condensation trap was set at - 50°C. The sample was then finely ground and stored in a pre-cleaned screw-capped plastic bottle (Nalgene, Rochester NY, USA) until analysis.

Some of the original sample was retained in its original form to allow weight loss experiments to be performed in order to quote the results on a dry mass basis.

4. Results and discussion

Fundamental to this study was the choice of the best digestion technique. The use of microwave digestion had several advantages over conventional digestion techniques. The digestion stage was completed in under 5 min following a pre-digestion stage, in which the digest was left to stand overnight to facil- itate the breakdown of any organic matter present prior to irradiating the sample. The pre-digestion stage clearly lengthens the sample preparation step; however, the microwave technique is comparable to conventional wet digestion techniques in terms of time, and is quicker than conventional cold digestion techniques which can take up to a week. The microwave technique is also less demanding on human resources since it requires less supervision of the digestion process.

In addition, with the microwave digestion being performed in a closed microwave vessel and in a single analytical step, there was less chance for contamination of the sample than in conventional techniques, which are performed in open top digestion vessels and may involve a number of heating steps and the addition of reagents.

By coupling the sensitivity of CVAAS to the auto- mated flow injection system, this method offers con- siderable savings in both time and labour over conventional batch analysis. Although the coupling of CVAAS and FIA is not totally novel, most of the previous reports in the literature are for clinical appli- cations involving the measurement of total mercury in liquid samples, e.g. blood and urine, and not for the analysis of complex matrix environmental samples as reported here.

The optimum digestion conditions for each of the two samples under investigation were determined


Table 3 Optimum conditions for the microwave digestion of sediments and certified reference materials

Digestion Digestion mixture time/min

Canned tuna fish 5.0 ml HNO3, 5 and DORM-1 dogfish 1.0 ml H202 Collected sediments 2.5 ml HNO3, 4 and MESS-2 estuarine 2.5 ml H2SO4, sediment 1.0 ml H202

independently. It was found that concentrated nitric

acid and hydrogen peroxide proved most suitable for the digestion of the biological samples, while for the

digestion of the sediment samples, a stronger oxi-

dising media was required. Thus, concentrated sulphuric acid was also added to the digestion mixture. The optimum conditions for the digestion of the

samples are shown in Table 3.

Table 5 Results obtained for the determination of total mercury in river sediments and tuna fish samples

Sample Resulta/ RSD/% n Recovery/ mg kg-l %

(i) River Mersey sediments 1 2.99 _+ 0.05 1.2 6 98.3 2 2.36 --_ 0.04 1.4 6 91.3 3 2.02 _+ 0.03 1.2 6 94.7

(ii) River Avon sediments 4 0.19 -+ 0.05 6.1 6 93.5 5 0.12 -+ 0.05 9.1 6 96.5 6 0.10 + 0.05 8.4 6 107.8

(iii) Tuna fish 7 1.00 ± 0.16 1.6 3 93.0 8 0.92 - 0.17 3.4 3 91.0

n = the number of sample replicates analysed for each sample. a Mean -~ SD at 95% confidence level for three replicate

determinations.

4.0.1. Analysis o f certified reference materials Two reference materials certified for mercury were

analysed: DORM-1, a biological fish muscle tissue,

and MESS-2, an estuarine sediment. These were

chosen since they have a similar matrix to the samples of interest. In both cases, the results obtained using the above method were in good agreement with certified

values. The results are shown in Table 4.

4.0.2. Analysis o f the environmental samples Once the method had been validated by the use of

certified reference materials, a number of real samples

with different matrixes were examined. These were

sediments collected from the River Mersey at Garston and the River Avon at Saltham estuary, and some

canned tuna fish (Table 5). In all cases, the samples were spiked to assess the recoveries, and gave values better than 90%, as shown in Table 5.

Table 4 Results obtained from the analysis of certified reference materials

Reference Certified valueS/ Calculated valueb/ material mg kg -1 mg kg -1

DORM-1 0.798 - 0.074 0.771 + 0.028 MESS-2 0.092 _+ 0.009 0.085 _+ 0.015

Certified values produced by NRCC. b Mean -+ SD at 95% confidence level for three replicate

determinations.

4.0.3. Interferences The last stage of this study examined the effect of

some potential interferences on the method. The elements chosen for this brief investigation are those

previously reported as potentially problematic for the determination of mercury. These elements interfere

by either (a) solution phase reactions, i.e. the metal ion is reduced preferentially to the Hg, e.g. Ni, Cu and Ag, or (b) gaseous phase reactions, i.e. where the

metal ion is converted to a gaseous hydride product

which is then carried into the quartz cell, e.g. As, Sb and Se [14,15]. The selectivity of the proposed

method was therefore investigated by use of a

Table 6 Results of interference study on mercury (10 ng m1-1) by selected transition and hydride forming elements

Ion Conc. of Hg conc. E r r o r Interference added ion found found effect (mg 1-1) (ng m1-1) (ng m1-1)

Ag(I) 1.0 9.02 -0.98 Suppression As(V) 2.0 10.02 +0 .02 Enhancement Cd(II) 5.0 10.44 +0 .44 Enhancement Cu(II) 2.0 10.32 +0 .32 Enhancement Ni(II) 10.0 10.35 +0 .35 Enhancement Pb(lI) 5.0 10.44 +0 .44 Enhancement Sb(V) 2.0 10.44 +0 .44 Enhancement Se(VI) 1.0 9.77 -0.23 Suppression

J. Murphy et al./Spectrochimica Acta Part B 51 (1996) 1867-1873 1873

mercury standard (10 ng m1-1) in the presence of Ag, As, Cd, Cu, Ni. Pb, Sb and Se.

The results of the study are given in Table 6 and show that silver and selenium had a suppression effect, whilst all the other interferences had a small enhancement effect. However, the levels at which the interference was noted (with the exception of silver) was at least two orders of magnitude greater than that of mercury, and not considered to be a major problem when determining mercury in the majority of environmental and biological samples.

5. Conclusion

The objective of this investigation was to develop a sensitive and accurate analytical method for the determination of total mercury in environmental and biological samples by flow injection CVAAS.

CVAAS was used owing to its ease of operation, its low detection limits compared with other techniques, and its suitability for use on-line with a FIA system. The use of flow injection in conjunction with the CVAAS technique allows for the rapid analysis of between 20 and 30 samples h -t. The results of the interference study, although not exhaustive, show that the method is relatively robust with regard to the selected elements, and in most cases gave less than 5% error when their concentration was at least two orders of magnitude greater than that of mercury. The method offers a detection limit of 0.2 ng g-1 (3 x SD of baseline (blank) signal), a relative standard deviation for real samples of less than 10%, and has been validated using standard reference materials.

Acknowledgements

The authors acknowledge the support of Bodensee- werk Perkin-Elmer GmbH in providing the FIAS-400 equipment used in this study, and in particular Dr Ian L. Shuttler for his valuable advice.

References

[1] B, Welz, Atomic Absorption Spectrometry, 2nd edn., VCH, Weinheim, 1985, Chapter 3.

[2] C, Vandecasteele and C.B. Block, Modern Methods for Trace Element Determination, Wiley, Chichester, 1993, Chapter 5.

[3] J.E. Tahan, V.A. Granadillo, J.M. Sanchez, H.S. Cubillan and R,A. Romero, J. Anal. At. Spectrom., 8 (1993) 1005.

[4] T. Guo and J. Baasner, Talanta, 40(12) (1993) 1927. [5] B. Weiz, D.L. Tsalev and M. Sperling, Anal. Chim. Acta, 261

(1992) 91. [6] D.C. Baxter and W. Frech, Anal. Chim. Acta, 236 (1990) 377. [7] M. Colina de Vargas and R.A. Romero, At. Spectrosc., 5(10)

(1989) 160. [8] S.B. Adeloju, H.S. Dhindsa and R.T. Tandon, Anal. Chim.

Acta, 285 (1994) 359. [9] S.B. Adeloju, Analyst, 114 (1989) 455.

[10] P. Quevauviller, J.-L. Imbert and M. Olle, Mikrochim. Acta, 112 (1993) 147.

[11] J. Nieuwenhuize and C.H. Poley-Vos, At. Spectrosc., 5(10) (1989) 148.

[12] G. Vermeir, C. Vandercasteele and R. Dams, Anal. Chim. Acta, 220 (1989) 257.

[13] 1. Rokkjaer, B. Hoyer and N. Jensen, Ta|anta, 40(5) (1993) 729.

[14] Y. Madrid, J.M. Gutierrez and C. Camara, Spectrochim. Acta Part B, 49 (1994) 163.

[15] N. Bernth and K. Vendelbo, Analyst, 109 (1984) 309.

determination of total mercury in environmental and biological samples by flow injection cold vapour...

Documents