Fresenius Z Anal Chem (1982) 313:200- 202

Determination of Mercury in Seawater by Cold Vapour Atomic Absorption Spectrophotometry

Fresenius Zeitschrilt fiir

�9 Springer-Verlag 1982

Peter Freimann and Diether Schmidt 1

Deutsches Hydrographisches Institut, Labor Siilldorf, Wiistland 2, D-2000 Hamburg 55, Federal Republic of Germany

Bestimmung von Quecksilber im Meerwasser durch Kaltdampf-Atomabsorptionsspektraiphotometrie

Zusammenfassung. Zur Vermeidung yon systematischen Feh- lern bei der Ultraspurenanalyse yon Quecksilber im Meer- wasser wurde ein integriertes Probenahme-Analysensystem entwickelt. Eine 500 ml PTFE-Flasche dient in Verbindung mit einem Greifarm als Probenahmegef~B ftir Oberflfichen- wasser und wird anschlieBend als ReaktionsgeffiB in die Analysenapparatur gesetzt. Die Bestimmung erfolgt mit der Kaltdampf-Atomabsorptionsspektralphotometrie nach Re- duktion des Quecksilbers durch SnC12 und Anreicherung sowie Reinigung mittels Amalgamierung an feinverteiltem Gold. Die Nachweisgrenze dieses Verfahrens betrfigt 0,5 ng. 1-1 der Variationskoeffizient 4 % bei 1,5 ng. 1-1

Summary. To avoid systematic errors in the sampling and determination of mercury in seawater an integrated sampling- analyzing system has been developed. A 500 ml PTFE-bottle attached to a sampling gear is used as sampling vessel for the cold vapour atomic absorption spectrophotometry, using nitrogen aeration, SnC12 reduction, and preenrichment/ purification by amalgamation on finely dispersed gold. The detection limit of this procedure was found to be 0.5 ng. 1-1 and the coefficient of variation 4 ~ for 1.5 ng. 1 1.

1. Introduction

The mercury concentrations of open ocean water are, in general, extremely low. However, values between 0.5 and 300ng.1 1 have been reported. Recent studies [2, 6] show that the higher values seem to be attributed to contamination during sampling, storage and pretreatment.

To prevent these difficulties an apparatus fully made from polytetrafluoro-ethylene (PTFE, "Teflon") has been con- structed, where the sampling and storage bottle is identical to the reaction vessel.

The analytical method preferred for the determination of mercury in natural samples is the closed-system reduction- aeration technique using mercury detection by gas phase atomic absorption (cold vapour atomic absorption spectro- photometry) described by Hatch and Ott [4]. To increase the sensitivity of the method, and to separate mercury from interfering substances a precursory concentration step is needed. In the procedure described here amalgamation with

1 Offprint requests to: D. Schmidt

finely dispersed gold is used to both separate and enrich mercury for analysis.

2. Experimental

2.1 Cleaning Procedure

The 500 ml PTFE-bottles are cleaned by means of a simple two-bottle still described by Mattinson [7]. The feed bottle is filled with 100ml of pro analysi grade HNO3 (typical Hg concentration 2,000 ng. 1-1), while the collecting bottle is in the cleaning position. After 2 days of heating with an infrared lamp the cleaning procedure is finished. The same apparatus is used for sub-boiling distillation of pro analysi grade HNO3. The content of resulting HNO3 is approximately 10 to 30 ng. 1-1 (a mercury decontamination factor of about 70 to 200). All other laboratory equipment used for the de- termination of mercury are first soaked in "sub-boiling" HNO3 and then rinsed at least six times with deionized water ( 1 - 2 ng. 1-1 Hg). The whole cleaning procedure is carried out in a clean bench with laminar air flow.

2.2 Sampling and Storage

In the open ocean a research vessel may be a significant source of contamination by heavy metals. Therefore, sampling of surface water from an inflated rubber dinghy at sufficient distance from the research vessel is recommended. The collection of seawater samples can be done by means of Mart's [5] sampling gear and a precleaned 500ml PTFE- bottle provided with 10 ml "sub-boiling" HNO3.

These preacidified seawater samples (pH about 0.9) show neither contamination nor loss of mercury during a storage period of 6 months.

The sampling of estuarine and coastal subsurface waters (maximum depth about 100m) should be as similar as possible to the sampling of surface waters. Therefore a new PTFE close/open sampler has been constructed and will be described elsewhere [3].

2.3 Pretreatment

The chemical speciations of mercury in the marine environ- ment are not well known. It has been suggested that chloro complexes are the prevailing form of mercury in seawater. However, there may be an appreciable fraction of the element in the form of organomercury complexes, which are unre- active to the S n C I 2 reduction used. The organo-mercury compounds can be digested by UV radiation [1].

0016-1152/82/0313/0200/$01.00

201

17

! i

I

A i I I

\ \ /

[EEXEEEEEJ B

Fig . 1. Schematic diagram of the immersion-irradiation equipment. A 500 ml PTFE-bottle; B magnetic stirrer bar; C magnetic stirrer; D UV- lamp; Ewater cooling vat

i r

D H J

,I L

- - 7

I F

G I<

Fig.2. Schematic diagram of the AI1-PTFE apparatus for the de- termination of mercury in sea water. A Flowmeter; B needle valve; C N2- cleaning tube; D two 3-way stopcocks; E 500ml PTFE-bottle; F magnetic stirrer; G SnCI 2 dispenser; H first drying tube; J second drying tube; K wire-wound Au-column; L variable transformer; Moptical cell; Nwashing bottle

Figure 1 shows the immersion-irradiation equipment used here. A two hour irradiation, with a 150W UV lamp, of a 400 ml preacidified seawater sample is recommended.

Organo-mercury is calculated fi'om the difference between the concentrations of total (photooxidized) Hg and of reactive (without photooxidation) Hg determined in aliquots of the same seawater sample.

2.4 Apparatus

The mercury analysis is conducted using a Coleman Instruments Mercury Analyzer (MAS 50) equipped with a Perkin-Elmer xt-recorder (Model 56).

A schematic diagram of the entire system is shown in Fig. 2. The flow of nitrogen as a purging and carrier gas is regulated by an alI-PTFE needle valve (B), checked by a flow meter (A), and cleaned by a small silica glass ("quartz") tube (C) (i.d. 3ram; length 100ram) which is filled with a gold coated "quartz" wool plug (25 mg SiO2 wool coated with 25rag gold). This tube is connected to the 500ml PTFE sparging vessel (E) (with a PTFE screw-cap containing three threaded boreholes) and/or two Pyrex-glass drying tubes (i. d. 9 mm; length 150 mm) containing MgC104 as a water absor- bent. Two three-way PTFE stopcocks (D) are placed between the N 2 cleaning tube, the sparging vessel, and the first drying tube (H) to permit the sparging vessel to be by-passed during the heating and elution step. A magnetic stirrer (F) is placed below the sparging vessel. The addition of the reduction solution [20 ~ (w/v) S n C I 2 in 3 M H2SO4] is carried out by means of a Pyrex-glass dispenser (G). The mercury collector follows after the water absorbing tube (J). The collector consists of a silica tube (i. d. 3 ram; length 120 ram) which is packed with a gold coated silica wool plug (50 mg silica coated with 50 mg gold) to form a 50 mm column (K). The tube is closely wound with ca. 2 m Nichrome wire (diameter 0.5 ram) connected to a variable transformer (L). The wire-wound tube is placed inside an insulating silica, covering (i. d. 8 mm;

length 80 ram) so that only the sections not wrapped by wire are exposed. The mercury which is rapidly vaporized and eluted from the column by electrical heating (maximum temperature inside the tube 550 ~ is fed directly into the optical cell (M) (length 150 mm) of the Coleman Analyzer. After the carrier gas has passed through the cell it is directed into a 250ml gas washing bottle (N) containing ca. 100ml 10~ KMnO4 in 3M H2SO 4. In this washing step the elemental mercury is oxidized and thus removed from the gas flow.

All connections in the system are made of PTFE tubing (i.d. 0.5ram; o.d. 3.2ram) and PTFE fittings.

2.5 Reagents

All reagents used are of pro analysi grade. The nitric acid is further purified by sub-boiling distillation as described above. The SnCI2 solution is aerated overnight with prepurified nitrogen to remove traces of mercury. The deionized water for reagent preparation and cleaning procedures is produced using a single step mixed bed ion exchanger and then stored in Pyrex bottles. The mercury content of this water was found to be 1 - 2 n g . 1 1. A mercury standard stock solution (0.1 g.1 1) is prepared by means of a Fixanal mercury standard ampoule (Riedel-de Hahn). For the working stan- dard (0.1 rag' 1-1 Hg) 100 lal of the stock solution are diluted to 100ml with 1 M HNO3 (prepared fresh daily).

2.6 Procedure

After shaking the preacidified seawater sample in the 500 ml PTFE bottle, 100ml of the sample are discharged, a pre- cleaned PTFE-coated magnetic stirrer bar is inserted, and the bottle is connected to the system with the three-borehole screw-cap. The three-way stop-cocks are switched to the sparging position and the flow rate of nitrogen is adjusted to 40ml. s -1 (0.5 bar).

202

Table l. Determination of precision: 20 replicate analyses of reactive Hg in a North Sea surface water sample

Sample Hg concentration Sample Hg concentration No. (ng. l- 1) No. (ng. 1-1)

1 1.8 11 1.4 2 1.6 12 1.5 3 1.9 13 1.3 4 1.8 14 1.1 5 1.7 15 1.4 6 1.4 16 1.5 7 1.5 17 1.3 8 1.7 18 1.0 9 1.2 19 1.4

10 1.6 20 0.8

2 = 1.4 +, 0.6ng'1-1 (95 ~ confidence level) (coefficient of variation 47o)

Table 2. Results obtained with the present method from participation in an international intercalibration of mercury in ocean water [8]

Seawater Seawater + 15,4 ng.1-1 spike (original) (ng. 1-1) (ng. 1-1) Recovery (7O)

2.9 +, 0.7 17.9 +_ 3.3 97

Seawater Seawater + 143 ng. 1-1 spike (original) (ng. 1-1) (ng. 1-1) Recovery (7O)

2.9 + 0.7 146 +_ 7 100

The magnetic stirrer is switched on and 3 ml of the SnC12 solution are slowly added through the dispenser to reduce the reactive mercury to its elemental state. The mercury is deposited by amalgamation on the gold coated silica wool plug. After purging has been completed (50 rain) the stop- cocks are switched to the bypass position and the flow rate of nitrogen is increased to 90ml. s -1 (0.5 bar). The elemental mercury is then vaporized and liberated from the Au column by electrical heating of the wire coil using the variable transformer. The absorption at 253.7 nm occurs in the gas cell of the MAS 50 about 1 - 2 s after reaching the mercury elution temperature. The absorption peak is recorded in arbitrary units. When the response has returned to the baseline the heating is discontinued. After a cooling period of about 5 rain the system is ready for the next cycle.

For blank determination 3 ml of the SnC12 solution are added to the "mercury-free" seawater sample (in which the analysis for mercury has just been carried out) and the whole operation is repeated as described above.

3. R e s u l t s and D i s c u s s i o n

Calibration curves are established in "mercury-free" seawater by spiking the sample with 0.5, 1, 2, 4, 10, and 20ng Hg 2§ The calibration curve is linear in this range.

Table 3. Check of the efficiency of photooxidation by using dimethyl mercury spikes

UV-irradiation Recovery (h) of CH 3 - Hg- CH 3

(%)

none 70 _+ 5 2 97 +4 4 101 +- 5

The reproducibility was tested by a 30-fold measurement of 2 ng Hg 2 § spikes. The mean peak height was 29 mm and the coefficient of variation 2 To.

The precision of the method was investigated by 20 replicate determinations of reactive mercury in a North Sea (54~ ' N, 7~ ' E) surface water sample (salinity about 32~ The results are shown in Table 1.

In a recent international intercalibration of mercury in seawater [8] two open ocean samples and two differently spiked seawater samples had to be analyzed. The values obtained with the method described here are summarized in Table 2.

The detection limit (D.L.) is calculated from

D.L, = xbl + k. Sbl

where 2~ mean of the blanks, su~ standard deviation of the blanks, k factor (k = 3). The detection limit of the method has been found to be 0.5 ng. 1-1

The organic-bound mercury can be digested by photo- oxidation (UV) or by chemical oxidation. To prevent con- tamination risks by reagents, the photooxidation method was chosen. The suitability of the UV-irradiation was investigated by spiking previously analyzed preacidified samples with known amounts of dimethyl mercury. The irradiation and the determination of total mercury was carried out in the equipment described above. The results are given in Table 3. Investigations on the applicability of this treatment to other organic mercury compounds and to mercury sulphides are intended for the near future.

R e f e r e n c e s

1. Armstrong FAJ, Williams PM, Strickland JDH (1966) Nature (London) 2t1:481

2. Bothner MH, Robertson DE (1975) Anal Chem 47:592 3. Freimann P, Schmidt D, Schomaker K (1982) Mercos - a Simple

Teflon Water Sampler for Ultratrace Element Analysis. Submitted for publ.

4. Hatch WR, Ott WL (1968) Anal Chem 40:2085 5. Mart L (1979) Fresenius Z Anal Chem 299:97 6. Matsunaga K, Komishi S, Nishimura M (1979) Environ Sci Technol

13:63 7. Mattinson JM (1972) Anal Chem 44:1715 8. Olafsson J (1981) Report on the ICES Intercalibration of Mercury in

Sea Water for the Joint Monitoring Group of the Oslo and Paris Commissions. In: Reports on ICES Intercalibrations of Mercury and Cadmium in Sea Water, 1979, pp 1-25, Coop. Research Rep. No. 110, Internat. Council for the Explor. of the Sea, Copenhagen 1981

Received March 16, 1982