Talanta

ELSEVIER

Talanta 44 (1997) 1219-1223

Determination of mercury in urine by ET-AAS using

complexation with dithizone and extraction with cyclohexane
C. Burrini "'*, A. Cagnini b
lstituto Ricerche Cliniche 'M. FanJimi', Piazza lndipendenza 18 b, 50123 Firenze, Italy
b Universitgt di Firenze Dipartimento di Sanit~ Pubblica, Epidemiologia e Chimica Analitica Ambientale, Via Gino Capponi 9,
50121 Firenze, Italy

Received 11 July 1996; received in revised form 28 October 1996; accepted 29 October 1996

Abstract

A precise and accurate graphite furnace atomic absorption spectrometric method for the determination of mercury
in urine was developed. Samples were subjected to hydrolysis with nitric acid. Then, mercury in the sample was
complexed by dithizone and extracted by cyclohexane. Mercury concentrations were determined against a urinematched calibration curve. Coated graphite notched partition tubes (Varian) and forked pyrolytic platforms (Varian)
were used. The detection limit of the method (Xblan k -~-3 SDbtank) was 1 gg 1- ~. The between run precision CV's were
4.7 and 3.4% for urine with a mercury concentration of 48.2 and 156.2 lamol 1-1, respectively; the within run precision
CV's were 8.9 and 2.9% for urine with a mercury concentration of 17.0 and 172 lag 1- ~, respectively. 1997 Elsevier
Science B.V.
Keywords: Graphite furnace; Mercury; Pyrolytic platform

I. Introduction

In the past, mercury was an important constituent of drugs, but nowadays its use is gradually decreasing. Concentration of mercury in air,
soil and water have been increasing because of
greater utilisation of fossil fuels and for the expanded use in industry and agriculture. Apart
from industrial exposure, the major source of
elemental mercury for humans is dental fillings
(Hg/Ag amalgam) [1]. A recent study [2] showed
that amalgam fillings contribute considerably to
plasma and urine mercury concentration.
* Corresponding author.

The potential hazard of human exposure to

mercury is well known. The novel 'Alice's adventures in wonderland' refers to ' M a d Hatters' who
were hat makers gone mad due to exposure to
mercury used in cleaning felt hats. With regard to
toxicity, three major chemical forms of the metal
must be distinguished: mercury vapours (elemental mercury); salts of mercury; and organic mercurials. Elemental mercury, salts of mercury and
aryl mercurial compounds, containing a labile
H g - C bond, possess the same toxicity with formation of divalent mercuric cations. In contrast
alkilmercurial
(e.g.,
methylmercury)
are
metabolised slowly.

0039-9140/97/$17.00 1997 Elsevier Science B.V. All rights reserved.

Pll S0039-9140(96)02162-5

1220

c. Burrini, A. Cagnini/ Talanta 44 (1997) 1219-1223

There is a linear relationship between plasma

concentration and urinary excretion of mercury
after exposure to vapour [3]. The concentration of
mercury in urine has been used as measure of the
body burden of the metal. In contrast, the excretion of mercury in urine is a poor indicator of the
amount of methylmercury in the blood, since it is
eliminated mainly in faeces.
Mercury in urine can be measured using a
potentiometric method [4], gas-chromatographic
methods [5-7], cold vapour atomic fluorescence
spectrometry method [8], and by atomic absorption spectrometry with the cold vapour technique
[9-12]. The latter has several advantages such as
lower detection limits, less time consuming sample
preparation, higher sample throughput due to a
shorter measurement cycle times and the procedure is full automated. But for this technique,
additional accessories, specifically designed for
measurement of mercury, are required. With our
method we tried to determine Hg without this
apparatus, but using the capacity of mercury to
bind sulphidric groups. Dithizone was used successfully as stabilising agent in the furnace [13],
and it is one of the complexing agents most
commonly used for mercury [14-16]. The complex is generally extracted with chloroform or
carbon tetrachloride. In this work we combined
both the complexing capacity of dithizone with
cyclohexane extraction. The latter was used because of its easier recovery due to its lower density
with respect to water.

2. Materials and methods

2.2. Reagents and materials

Suprapur nitric acid, water pro analysis and

cyclohexane were obtained from Merck (Milano,
Italy). Diphenylthiocarbazone (dithizone) was obtained from Sigma, mercury (II) chloride 99.5%,
mercury (II) nitrate monohydrate 98% and standard solution of mercury 995 ~tg ml 1 in 1%
nitric acid were obtained from Aldrich (SigmaAldrich, Milano, Italy). Two stock standard solutions of 4 and 20 pg l-1 in 1% nitric acid using
polypropylene tubes from PBI International (Milano, Italy) were prepared daily [17] from the
latter.
Methylmercury (II) chloride (1 g 1-~ standard
solution in water) was purchased from Alfa Products (Karlsruhe, Germany). All glass tubes (PBI
International) were acid-washed with 1 M nitric
acid solution and rinsed three times with water.
2.3, Samples

Urine was collected at random time from a

volunteer group of men and women (ages 20-40)
with no professional exposure in polypropylene
containers, 10 ml were put in polypropylene
tubes, capped and stored at 4C for a maximum
of 48 h. The amount of creatinine ranged from 0.8
to 1.5 g 1 ~. To monitor the accuracy and precision of analytical procedure a Lyphocheck Urine
Metal Controls (Biorad, Milano, Italy) was used.
The control is prepared from human urine with
added trace elements and heavy metals. The
lyophilised urine samples were reconstituted with
distilled water and stored at 2-8C for 5 days
according the reported instructions.

2,1. Instrumentation and apparatus

2.4. Samples and standards preparation

For all measurements a Varian SperctrAA-300

atomic absorption spectrometer with graphite
tube atomizer and programmable sample dispenser was used. The spectrometer was equipped
with a Zeeman background corrector for the correction of non specific signal. Mercury hollow
cathode lamp, graphite notched partition tubes
and graphite forked pyrolytic platform were obtained from Varian (Torino, Italy).

Sample, 4 ml, were pipetted into glass tubes.

Standard additions calibration curves were prepared adding different volumes of the stock standard solutions to 4 ml of urine from two donors
without neither professional exposure nor dental
fillings. Blank solutions were prepared using 4 ml
of water. Nitric acid 65%, 500 ~tl were added to
each tube, and, after mixing, were capped and
incubated at 95C for 90 min.

122t

C. Burrini, A. Cagnini / Talanta 44 (1997) 1219 1223

Table 1
Time/temperature program
Step no.

Temperature C

Time s

Gas flow I min- ;

Read command

90
180
180
180
1400
1400
2700
2700
80

5.0
5.0
5.0
2.5
1.0
3.0
1.1
2.0
15.0

3.0
3.0
3.0
0.0
0.0
0.0
3.0
3.0
3.0

No
No
No
No
Yes
Yes
No
No
No

2
3
4
5
6
7
8
9

After cooling, 2 ml of a dithizone saturated

solution in cyclohexane were added. The mixture
was vigorously shaken for 30 s and then, after
separation, the upper phase was transferred to a
clean glass.
2.5. I n s t r u m e n t conditions

The m o n o c h r o m a t o r was set at 253.7 with a slit

width of 0.2 nm. L a m p current was 5 m A and
measurements were recorded both in peak height
and peak area mode. The samples were injected
into the furnace at the temperature of 90C (hot
injection) with a rate of 0.7 gl s - 1. The volume of
the injection was 25 lal.

3. Results and discussion

Usually the traditional furnace methods are not

recommended for mercury because it is a very
volatile metal. In order to decrease its volatility
m a n y matrix modificators have been proposed as
palladium [18,19], strong oxidising agents [20] and
complexing agent with sulphidric groups [13-16].
In our procedure, dithizone is responsible both of
the extraction of the metal from a complex matrix
(i.e., urine) and of a decreasing of volatility during
the pre-atomisation step. In fact the determination of mercury in aqueous solution with the same
time/temperature program do not give appreciable absorbance values. Every modification of
time/temperature program do not increase the
absorbance values. In Table 1 is presented the

time/temperature program, As indicated on the

char/atomisation curves in Fig. 1, the maximal
char temperature that could be used without loss
of sensitivity was 200C, while the best atomisation temperature was 1300-1400C.
We used coated graphite notched partition
tubes and forked pyrolytic platforms. With the
same time/temperature program a decreasing of
the absorbance value was observed using coated
graphite partition tubes. With such tubes the optim u m of atomisation temperature was observed at
800C. The resulting values were the same with
respect to the values obtained with pyrolytic platform but they were characterised by a greater
peak broadening (Fig. 2).
Dithizone forms hydrophobic complexes with
most of the divalent or trivalent metal ions. Copper (10-70 lag 24 h 1) and zinc (150-1300 lag 24
h ]) are the most c o m m o n metal ions in urine:
25<)
225

A'FOMIZA I ION

20(I]

175 !

Q,\\

15O

'\
( :tl A R

"~\

i25
I O0
tP
75
5O -

'

//

I elnperat ur (

Fig. 1. Char and atomization temperature curves of the determination of mercury in urine.

C, Burrini, A. Cagnini / Talanta 44 (1997) 1219-1223

1222

A)
3888

-8.28|
'tIME ( ~ c )
FIEIGHT

8
6.2
8.226
8.143

B)
1088

-~.28 L
8

'rIME ( ~ c )
I~IG,tlT

6.5
8.167
8.141

Fig. 2. (A) Atomic signal from pyrolytic platform (char temperature: 180C, atomization temperature 1400C); (B) Atomic
signal from coated partition tube (char temperature: 180C,
atomization temperature 800C). Height, A; Area, A.s.

but it was found that even high concentrations of

both ions (300 pg 1 ' for copper and 2000 pg 1for zinc) did not interfere with mercury determination with this method.
Nitric acid did not show an oxidising action to
give dithizone a change in complexing capacity.
But it was important to separate the organic
phase from the urine after the extraction. In fact
after 30 rain the separated dithizone solution did
not show significant absorbance variation, while
the part in contact with urine showed a reduction
of 10%.
The detection limit of the method, calculated as
X b l a n k "t-3SDblan k, was 1 pg 1- j . Solutions for 20
different calibration curves were prepared adding
different volumes of the daily prepared stock standard solutions to urine from donors without neither professional exposure nor dental fillings. The
recorded absorbances for no-spiked samples were
below the threshold value. The results obtained in
peak area and peak height mode were compared.
The response was linear up to 200 pg 1 ~. Both
the systems give a good correlation factor (r) and
reproducible slopes between the different calibration (peak area: average correlation factor r =

0.9993, C V % = 0 . 0 8 and average slope 1.27,

CV% = 0.07; peak height: average correlation factor r = 0.9995, C V % = 0 . 0 7 and average slope
1.74, CV% = 5.84). In order to test the proposed
method in routine work, different furnaces with
different degree of use were utilised. These furnaces were previously used both for mercury in
urine determination and for determination of
other analytes in different matrices (blood, urine,
serum, plasma). The increase in life time of use
led to wider peaks with lower peak heights, resulting in higher coefficient of variation. In Fig. 3
typical calibration curves are shown.
After 48 h of storage at 4C no significant
variation of concentration were measured. Urine
added with Hg z+ in the linearity range and stored
for the same time gave 98 101% recoveries. The
within-run precision (measure of repeatability)
and the between run precision (measure of reproducibility) were measured considering two concentrations of mercury with quality controls. For
the between-run precision Bio Rad Lyphocheck
Urine metal control Level 1 (Lot. No. 60901) and
Level 2 (Lot No. 60902) were used. For the
within-run precision Bio Rad Lyphocheck Level 1
(Lot No. 58501) and Level 2 (Lot No. 58502)
were used. Results are reported in Table 2.
Accuracy was evaluated comparing data obtained with the Bio Rad quality controls (Table 2)
and evaluating the recovery from urine of different subjects added with mercury. In the urine

4o0i
/ J//
300 ,

A (mU)

_ _

/
/

200

/ ~

~J
jJ

At (mUs)

/-q~

YU
0 oo

0 05

0 I0

0 15

0 2(I

Hg COTiccntrationm~,q.

Fig. 3. Linearity of the calibration curve in urine in peak-area

([B) and peak-height ( 0 ) modes using graphite tubes and
platforms. Each point represents the mean _+ S.D. of 20 calibrations. Char temperature, 180C; atomization temperature,
1400C. Instrumental settings as reported in the text.

C. Burrini, A. Cagnini / Talanta 44 (1997) 1219 1223

1223

Table 2
Precision
Acceptable range (lag I ~)

Mean (lag 1 L)

S.D. (lag 1 ~)

CV (%)

Between run (n = 20)

Level I
Level 2

35.0--52.0
116.0--174.0

48.2
156.8

2.3
5.3

4.7
3.4

Within run (n = 4)
Level 1
Level 2

13.0--20.0
137.0--206.0

14.9
151.9

1.3
4.4

8.9
2.9

added with inorganic salts (mercury (II) nitrate

and mercury (II) chloride) 98-101% of recoveries
were obtained for all the concentrations in the
linearity range. For metal organic compounds
(methylmercury chloride), using the same procedure as for inorganic salts, the obtained recoveries
were between 85-90/, for 25 ~tg 1 ~, 84-88% for
50 gg 1 ~ and 75-85% for 100 tag 1 1 of
methylmercury addition. During our work, 25
urine samples were tested. Results, ranging from 1
to 8 lag 1- ' for 23 samples, collected from people
with dental filling, were obtained while for two
people without dental fillings results were below
the detection limit. Though not statistically relevant, these results are in accordance with expected
values and with other published papers [3,8,21].
All these data allow to consider this technique
precise and accurate so that it makes possible a
simple and fast quantification of mercury in urine
even if the equipment necessary to the cold vapour technique is lacking.

References
[l] M.J. Vimj and F.L. Lorscheider, J. Dent. Res., 64 (1985)
1072-1075.
[2] M. Molin, B. Bergman, S.L. Marklund, A. Schutz and S.
Skerfving, Acta Odontol. Scan., 48 (1990) 189-202.
[3] C.O. Klaassen, in A. Goodman Gilman and L.S Goodman (Eds.), The pharmacological basis of therapeutics,
VII edition, Macmillan, New York, 1985, pp. 1605-1627.
[4] D. Jagner and K. Aren, Anal. Chim. Acta, 141 (1982)

157-162.
[5] P. Mushak, F.E. Tibbets, 1.P. Zarnegar and G.B. Fisher,
J. Chromatogr., 87 (1973) 215-226.
[6] P. Zarnegar and P. Mushak, Anal. Chim. Acta, 69 (1974)
389-407.
[7] L. Liang, N.S. Bloom and M. Horvat, Clin. Chem., 40 (4)
(1994) 602-607.
[8] S.A. Winfield, N.D. Boyd, M.J. Vimy and F.L.
Lorscheider, Clin. Chem., 40 (2) (1994) 206 210.
[9] A. Campe, N. Velghe and A. Claeys, At. Absorpt.
Newslett., 17 (1978) 100.
[10] D.R. Boucier, R.P. Sharma and D.B. Drawn, Am. Ind.
Hyg. Assoc. J., 32 (1982) 329.
[11] D.E. Shrader and W.B. Hobbins, Varian Ins. At Work,
September, 1983, AA-32.
[12] T.E. Guo and J. Baasner, Anal. Chim. Acta, 278 (1993)
18% 196.
[13] G.T.C. Shum, H.C. Freman and J.F. Uthe, Anal. Chem.,
5l (3) (1979) 414 416.
[14] L.S. Clesceri, A.E. Greenberg and R.R. Trussel, Standards methods for the examination of water and wastewater, 17th ed., edited by APHA, AWWA and WPCF,
American Public Health Association, Washington, 1989.
[15] S. Nobel and D. Nobel, Clin. Chem., 4 (1958) 150-158.
[16] S. Nobel, in D. Seligson (Ed.), Standard Methods in
Clinical Chemistry, Vol. 3, Academic Press. New York,
1961, p. 176.
[17] J.M. Lo and C.M. Wai, Anal Chem., 47(11) (1975)
1869-1870.
[18] L. Pingo, K. Fuwa and K. Matsumoto, Anal. Chim.
Acta, 179 (1985) 279-284.
[19] B. Welz, G. Bozsai, M. Sperling and B. Radziuk, J. Anal.
Atom. Spectr., 7 (3) (1992) 505-509.
[20] G.F. Kirkbright, S. Hsiao-Chuan and R.D. Snook, At.
Spectrosc., I(A) (1980) 85-89.
[21] V. Liengar and J. Walnier, Clin. Chem., 34(3) (1988)
474- 481.