ANALYTICAL SCIENCES DECEMBER 2000, VOL.

16 1309
2000 © The Japan Society for Analytical Chemistry

Determination of Mercury by Cold-Vapor Atomic Absorption

Spectrometry with Preconcentration on a Gold-Trap
Edyta KOPYŚĆ, Krystyna PYRZYŃSKA, S awomir GARBOŚ , and Ewa BULSKA†

University of Warsaw, Department of Chemistry, Pasteura 1, 02-093 Warsaw, Poland

The determination of trace levels of mercury by flow injection cold-vapor atomic absorption spectrometry with gold
amalgamation was evaluated for organic and inorganic mercury species. The effect of the sodium tetrahydroborate(III)
and hydrochloric acid concentrations as well as various instrumental parameters on the analytical performance was
studied. The influence of the blank as well as the laboratory procedure for reagent purification was evaluated. The
procedure was applied to the determination of mercury in lake water, while focusing on the importance of the sampling
protocols.

(Received April 24, 2000; Accepted July 21, 2000)

Cold-vapor atomic absorption spectrometry (CVAAS) is one of fraction of the total mercury present in the sample.
the most popular techniques for determining mercury in a wide The aim of this work was to investigate and optimize the
variety of samples.1–6 Generally, this technique is applied to chemical condition for mercury reduction, as well as the
determine the total mercury content.7 Either tin chloride1–3 or parameters of the flow system, which was coupled on-line to the
sodium tetrahydroborate(III)4,5 is used to reduce mercury to its preconcentration and separation unit. Various parameters
elemental form. Mercury vapor is purged from solution by a affecting the mercury preconcentration and its final
carrier gas, such as air, nitrogen or argon. After passing through determination by the CVAAS method were evaluated. Studies
a gas-liquid separator, it is introduced into the optical path of an were carried out for inorganic mercury (Hg(II)) and
atomic absorption spectrometer. phenylmercury (PhHg). The data were compared to those
Several modifications have been introduced to this technique, obtained without a preconcentration step.
aiming to improve the sensitivity, precision, lower interferences
and increased practicality of the methodology. Some of the
proposals for preconcentrating mercury include extraction8 and Experimental
the amalgamation on silver or gold,1,2,9,10 as well as the
application of solid sorbents.3,8,11 Especially an enrichment Instrumentation
procedure based on the amalgamation of generated mercury A FIMS flow-injection atomic absorption spectrometer
vapor on a gold trap is recommended for the quantification of a (Perkin-Elmer, Germany) was used for all experiments. A
very low mercury content in biological and environmental sample loop of 1 ml volume was used. The pump action,
samples. Beside the preconcentration effect, separation from injection and measurement (Table 1) as well as data processing
other compounds in the gas phase could be achieved. The gold were controlled through Perkin-Elmer WinLab software.
trap is a simple on-line preconcentration system that does not Measurements were performed in both the peak-area and peak-
require the use of reagent and could be easily adopted to the height modes.
flow-injection technique. Recently, the use of gold
amalgamation has been reported in combination with the Reagents
pyrolysis of solid samples in a combusion tube at 750˚C under Inorganic mercury standard solutions were prepared by
an oxygen atmosphere, as reported for mercury reduction.12 appropriate dilution of a 1 g l–1 mercury(II) nitrate stock
Mercury and its compounds, even at low concentration, are
known to be a dangerous neurotoxins for living organisms.
Therefore, it is essential to use an accurate and precise Table 1 Flow injection and preconcentration program in FIMS
analytical method for the determination of trace amounts of (Perkin-Elmer)
mercury. Because the average concentration of Hg in natural
water samples was found to be a few ng l–1 and the typical Au/Pt Gas flow
Valve
Step Time/s Pump speed, rpm gauze
detection limit for CVAAS is 0.05 µg l–1 level,6,13,14 the position
stage Carrier Purge
preconcentration step seems to be essential in this case. Low
detection limits are also necessary in studies where speciation is 1 20 100 fill
involved, because some species may account for only a small 2 25 100 inject *
3 10 100 inject *
† To whom correspondence should be addressed. 4 10 0 fill heat *
5 20 0 fill cool
E-mail: ebulska@chem.uw.edu.pl
1310 ANALYTICAL SCIENCES DECEMBER 2000, VOL. 16

In this work a sodium tetrahydroborate(III) solution was

chosen as the reductant because of its ability to reduce both
inorganic and organic forms of mercury.9 Also, when NaBH4
was used, the signal did not depend on the type and
concentration of acid.14 In order to optimize the concentration of
NaBH4 necessary for the quantitative generation of mercury
vapor, a 1 ml sample (volume of injection loop) containing 1 ng
of Hg(II) or 3 ng of PhHg was used. In both cases, the
concentration of the HCl solution was 3% and the NaBH4
concentration was varied from 0.025% to 0.6%. In preliminary
experiments, it was found that without an amalgamation
effective reduction occurred for inorganic and organic mercury
for 0.025% and 0.2% of NaBH4, respectively. When the
amalgamation step was included, the integrated absorbance
Fig. 1 Effect of the NaBH4 concentration on the integrated
absorbance for Hg(II) (1 µg l–1) and PhHg (3 µg l–1) solutions. remained constant within the concentrations range of 0.05 –
0.4% of NaBH4, for inorganic mercury (Fig. 1). Taking into
account the precision as well as the manufacturer’s
recommendation, a 0.2% NaBH4 solution was applied in further
solution (GUM, Poland) in 3% (v/v) hydrochloric acid. experiments. For PhHg the best sensitivity was achieved when
An organic mercury solution (0.1 g l–1) was prepared by the concentration of borohydride was above 0.45%. Therefore,
dissolving an appropriate mass of phenylmercury acetate NaBH4 of 0.5% was selected for further optimization.
(Merck, Germany) in 100 ml of water. The final concentration In the next step, the influence of the flow rate of the sample,
was achieved by dilution with 3% (v/v) HCl. acid and NaBH4 solutions, over the range of 3 – 9 ml min–1 on
Sodium tetrahydroborate(III) solutions were prepared by the sensitivity of the mercury was investigated. The optimum
dissolving NaBH4 (Aldrich, England) in 0.008 mol l–1 of KOH. flow rates were found to be 7.5 ml min–1 for sample and acid
These solutions were prepared daily before use. solutions and 5 ml min–1 for a reductant solution.
A carrier solution, 3% (v/v) HCl, was prepared by diluting After the generation step, mercury vapor was carried to the
concentrated HCl with water. gold trap. It was therefore essential to maintain the carrying
Ultrapure HCl 30% (Cheman, Poland); Suprapur HCl 37% time long enough to permit the total mercury vapor to be
(Merck, Germany); Pure for analysis HCl 36% (POCh, Poland) transported to the trap. When the tube length was constant (≅
and reagent of special purity (max. 0.0000005% Hg) — fuming 95 cm) the efficiency of this step depended mainly on the gas
pure for analysis HCl 37% (Merck, Germany) were used. flow rate. The effect of the argon flow rate on the signals for
A Polyorgs 31 sorbent (SAC) based on cellulose (fiber) both mercury species was examined. It was found that both the
(Vernadsky Institute of Geochemistry and Analytical peak-height or integrated absorbance increased with increasing
Chemistry, Russia), a Dowex 1×4 (Serva, Germany) and an argon flow rate up to 250 ml min–1. Although, in the range of
Amberlyst A-27 (BDH, England) were used. 250 – 300 ml min–1 a slight decrease of sensitivity was observed,
the best precision (0.8% RSD) was achieved. Therefore, a flow
Procedure rate of 300 ml min–1 was chosen in order to achieve the best
A continuously flowing carrier stream (3% HCl solution) compromise, while considering such factors as the precision and
transported the sample from the injection loop to the manifold, the total time of analysis. Under these conditions, the
where it was mixed with the reductant. The reaction mixture corresponding amalgamation time and the total analysis time
was then merged with an argon stream and transported to the were 10 s and 90 s, respectively. When a longer amalgamation
gas/liquid separator, where the liquid was separated from the time (25 s) was applied, an increase in the blank signal was
gaseous components. The gas phase passed through the wash observed (from 0.05 to 0.08 absorbance). After
flask where soluble contaminants were removed, and then preconcentration, the collected mercury was released directly to
through a drying tube to remove any liquid droplets. The the spectrometer cell by heating the trap for 10 s at 600˚C.
mercury vapor, separated from the contaminants, flowed over a After each measurement the trap was cooled by using an
gold/platinum gauze. After collection, the Au/Pt gauze was external flow of argon for 20 s.
heated rapidly up to 600˚C to remove Hg, and the carrier gas By using optimized conditions for mercury generation and
transported mercury vapor to a quartz cell in the spectrometer. preconcentration, the effect of the sample volume up to 3.0 ml
on the measured absorbance values was investigated. The
results are shown in Fig. 2. Satisfactory linearity of the response
Results and Discussion was observed for the enrichment of 1 ng ml–1 Hg(II) and 3 ng
ml–1 PhHg solutions with correlation coefficients of 0.9998 and
Optimization of the experimental conditions 0.9977, respectively. However, it does not offer unlimited
Several factors could affect the preconcentration of mercury possibilities to lower the detection limit, because with a larger
species on a gold trap, followed by its determination by volume of the sample, increases of the blank signal were noted
CVAAS. In the present work, this method was studied in (Fig. 2). The hydrochloric acid solution was found to be a main
relation to the chemical and flow variables in order to obtain the source of this effect. Several ways to allow a reduction of the
best analytical parameters. blank are described in the next section.
The concentration of hydrochloric acid in the range of 1 – 6% The lowest detection limit (DL), 5.1 ng l–1 (3σ), was obtained
had a very small effect on the absorption signal for PhHg and for a 1 ml sample when integrated absorbance measurements
Hg(II). Due to the possibility of the partial decomposition of were performed. In the case of peak-height measurements, the
phenylmercury into inorganic mercury in a strong acid DL of 4.3 ng l–1 was obtained when 4 ml of a sample was used.
solution,4 further work was performed using 3% HCl. As expected, a measurement based on the peak-area exhibits
ANALYTICAL SCIENCES DECEMBER 2000, VOL. 16 1311

Fig. 2 Effect of the sample volume on the integrated absorbance

for 1 µg l–1 inorganic and 3 µg l–1 organic mercury solutions. Sample Fig. 3 Efficiency of the purification procedure applied for a 3%
flow rate, 7.5 ml min–1. HCl solution. Initial content of mercury, 1.5 µg l–1 (POCh, Poland).

better precision compared to the peak-height because the hydrochloric acid. For inorganic mercury, Hg(II), a typical
fluctuations in the processes of the release and transport of calibration graph in the CVAAS method with a gold-trap
mercury from the trap are minimized. preconcentration obtained by linear regression corresponded to
A = 0.2453c + 1.6 × 10–5, r2 = 0.9998, where A is the integrated
Purification of hydrochloric acid absorbance and c is the concentration of Hg in µg l–1. The linear
The main source of contamination resulting in a high blank working range extended up to 1.5 µg l–1 for a 1 ml sample
signal, when mercury was determined at the ng l–1 level, was volume. The sensitivity of mercury determination expressed in
found to be the hydrochloric acid used as a carrier solution. The terms of the characteristic mass (e.g. that mass of the analyte
following reagents were tested with respect to the mercury which provides an integrated absorbance of 0.0044) was 15.9
content: (i) HCl# — reagent of special purity for Hg pg. This value was better than that obtained with the CVAAS
determination (Merck, Germany); (ii) suprapur (Merck, method without amalgamation (25.6 pg). The repeatability of
Germany); (iii) ultrapure (Cheman, Poland); (iv) pure for measurements, expressed as %RSD, varied between 0.2% (for
analysis (POCh, Poland). In all solutions, except for the acid of 0.5 µg l–1) and 2.2% (for 1.5 µg l–1). The detection limit (3σ)
special purity HCl#, the metal content was above 1 µg l–1, which obtained in 10 blank runs corresponded to 7.6 ± 0.3 ng l–1 (95%
was found to be unsatisfactory for a trace-mercury confidence limit).
determination. Therefore, laboratory procedures for acid A linear correlation between the integrated absorbance (A)
purification, such as isopiestic distillation or solid-phase and the PhHg concentration (c, µg l–1) was obtained up to 1.5 µg
extraction, were investigated. l–1 with a calibration equation of A = 0.0970c + 2.3 × 10–5 (r2 =
Two procedures, batch and column, were used in the case of 0.9999). The characteristic mass was 45 pg/0.0044A (without
solid-phase extraction. The cellulose sorbent with amidooxime amalgamation m0 was 89.8 pg). The detection limit, based on
groups (SAC) due to the slow kinetics of sorption was used in a integrated absorbance, was equal to 161 ng l–1 (n = 10); the
batch method. The 0.35 g amount of this sorbent was shaken precision obtained at the 1.5 µg l–1 level was 1.5% RSD.
for 1 h with 150 ml of a 3% HCl solution. Anion-exchange The experimental enrichment factor (EF), calculated as the
resins, Amberlyst A-27 and Dowex 1×4, were used for a ratio of the slopes of the calibration graphs obtained without and
column procedure. After a 0.5 g amount of each exchanger was with preconcentration on a gold-trap, was evaluated to be 7.1
packed into the column, an acid solution was passed at a flow and 9.9, respectively, for Hg(II) and PhHg. Because the
rate of 3 ml min–1. The efficiency of the applied purification efficiency of preconcentration depends on the sample volume,
procedure for the most pronounced contaminant reagent, the use of a longer preconcentration time (larger sample
analytical reagent grade (POCh), is shown in Fig. 3. The volume) improved the enrichment efficiency, but
relative concentration of mercury both before and after simultaneously lead to a decrease in the sampling rate. Hence,
purification was evaluated. The efficiency of SAC was found it seems to be very appropriate to use a concentration efficiency
not to be satisfactory. The mercury content decreased by about (CE) value for comparing various flow-injection systems with a
85% when anion exchangers (Amberlyst A-27 or Dowex 1×4) preconcentration step.15 The concentration efficiency is a
were applied. The results were comparable to that when function of EF and the sampling frequency, counted as the
isopiestic distillation of acid was used. For further work, the number of samples analyzed per hour or minute. In the
column method with Amberlyst or Dowex was chosen, because developed method CE values of 4.7 and 6.6 min–1 were obtained
it is simple, and could be performed on-line. After applying the for a 1 ml sample volume.
chosen purification procedures, the hydrochloric acid solution In previous published papers concerning the determination of
contained a comparable amount of mercury (∼0.2 µg l–1) to the mercury in a flow-injection system using microcolumn
reagent of special purity (HCl#) recommended by Merck for the preconcentration, the limit of detection for Hg(II) with silica gel
determination of trace mercury. The blank value for both HCl# functionalized with [1,5-bis-(di-2-pyridyl)methylene
and hydrochloric acid solutions after purification was below tiocarbohydrazide]16 was found to be 1 ng ml–1 and for 1,5-
0.05 absorbance. bis[2-pirydyl]-3-sulphophenyl methylene chelating resin17 was
estimated to be 4 ng ml–1. In this work a higher value of EF =
Analytical figures of merit 10.1 was reported; however, this was achieved by using a larger
The analytical figures of merit were evaluated using purified sample volume of 6 ml.
1312 ANALYTICAL SCIENCES DECEMBER 2000, VOL. 16

Table 2 Procedure for sample treatment Table 3 Results for the determination of mercury in lake water
samples (Szczȩśliwice lake, Warsaw City)
Sample Order of activity
Sub-sample/procedure Hg concentrationa/µg l–1
A filtration on-field acidification on-field
B acidification on-field filtration on-field A 0.04 ± 0.010
C filtration on-field acidification in laboratory B 0.07 ± 0.0001
D filtration in laboratory acidification in laboratory C 0.11 ± 0.002
D 0.05 ± 0.006

a. Mean ± s, 95% confidence limit (n = 4).

Determination of mercury in water samples and sampling

protocols
In natural ground water, especially under aerobic conditions, 2. R. C. Campos, C. L Porto da Silveira, and R. Lima, At.
mercury is present mainly in the inorganic form.18 Therefore, Spectrosc., 1997, 18, 55.
despite the fact that the sensitivity of the developed procedure is 3. J. Chwastowska, W. Skwara, E. Sterlińska, J. Dudek, and
different for organic and inorganic mercury, the method was L. Pszonicki, Chem. Anal. [Warsaw], 1998, 43, 995.
applied to natural samples from surface water from small Lake 4. X. Yin, W. Frech, E. Hoffman, C. Lüdke, and J. Skole,
Szcz ȩśliwice in the north-west area of Warsaw City. The main Fresenius J. Anal. Chem., 1998, 361, 761.
aim of this part of the project was to investigate the influence of 5. E. Wieteska and A. Drzewińska, Chem. Anal. [Warsaw],
sampling protocols for the determination of trace amount of 1999, 44, 547.
mercury. 6. E. Bulska, W. Kandler, P. Pas awski, and A. Hulanicki,
The samples were collected in acid-washed bottles. The Mikrochim. Acta, 1995, 119, 137.
sample preparation always included filtration (using 0.45 µm 7. W. I. Clevenger, B. W. Smith, and J. D. Winefordner, Crit.
filter Sartorius, Germany) and acidification of the sample Rev. Anal. Chem., 1997, 27, 1.
(HCl#). Different sampling protocols were used, as is shown in 8. P. Canada Rudner, J. M. Cano Pavón, F. Sánchez Rojas,
Table 2. The determination results are in Table 3. and A. Garcia de Torres, J. Anal. At. Spectrom., 1998, 13,
Generally, the content of mercury in samples which were first 1167.
filtrated and then acidified, was not dependant on whether both 9. R. Dumarey, R. Heindryck, R. Dams, and J. Hoste, Anal.
steps were performed in the field or in the laboratory. As an Chim. Acta, 1979, 107, 159.
example, the results for sub-samples A and D could be 10. C. G. Bruhn, A. A. Rodriquez, C. Barrios, V. H. Jaramillo,
compared. However, both values were much lower when J. Becerna, U. González, N. T. Gras, O. Reys, and S. Salud,
compared with sub-samples B and C. This could be explained J. Anal. At. Spectrom., 1994, 9, 535.
by the fact that the aim of filtration was to remove any non- 11. M. Okumura, K. Fukushi, S. N. Willie, and R. E. Sturgeon,
dissolved particles from the sample solutions that may Fresenius J. Anal. Chem., 1993, 345, 570.
subsequently be dissolved by the acid. If mercury is partially 12. C. T. Costley, K. F. Mossop, J. R. Dean, L. M. Garden, J.
deposited on the particle, this could be the reason for the Marshall, and J. Caroll, Anal. Chim. Acta, 2000, 405, 179.
increase in the mercury content. This is in a good agreement 13. M. Fairulnizal Noh, A. H. Tumijah, and I. Zakiah, At.
with the results obtained by Reimann et al.,19 and supports our Spectrosc., 1998, 19, 95.
experience that the sampling protocols for the collection and 14. E. Bulska, W. Kandler, and A. Hulanicki, Proccedings
preservation of water samples for trace mercury analysis are CANAS ’95, ed. B. Welz, 1996, 601.
very important, because they influence the final results. 15. Z. L. Fang, Spectrochim. Acta Rev., 1991, 14, 235.
16. P. Canada Rudner, A. Garcia de Torres, J. M. Cano Pavón,
and E. Rodriguez Castellon, J. Anal. At. Spectrom., 1998,
Acknowledgements 13, 243.
17. P. Canada Rudner, A. Garcia de Torres, J. M. Cano Pavón,
This study was supported partially by European Community and F. Sanchez Rojas, Talanta, 1998, 46, 1095.
project No. IC15-CT98-0119 (DG12-JNCN). 18. D. R. Turner, “Lead, mercury, cadmium and arsenic in the
environment”, ed. T. C. Hutchinson and K. M. Meema,
1987, J. Wiley, Chichester, 107 – 121.
References 19. C. Reimann, U. Siewers, H. Skarphagen, and D. Banks, Sci.
Total Environm., 1999, 234, 155.
1. J. M. Ombaba, Microchem. J., 1996, 53, 195.