HG Proyecto

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

626 JULSHAMN & BRENNA: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO.

3, 2002

FOOD CHEMICAL CONTAMINANTS

Determination of Mercury in Seafood by Flow Injection–Cold


Vapor Atomic Absorption Spectrometry after Microwave
1
Digestion: NMKL Interlaboratory Study
KAARE JULSHAMN and JAN BRENNA
Institute of Nutrition, Directorate of Fisheries, PO Box 185 Sentrum, N-5804 Bergen, Norway

Collaborators: B. Åsman; A. Ekman; A. El-Ghauoui; K.-O. Gjerstad; D. Grønningen; S.H. Hammer; A.M. Jensen;
H. Liukkonen-Lilja; J. Mikkola; T. Ruth; L.O. Sedal; B. Solli; E.-R. Venalainen; N. Wollertsen

Ten laboratories participated in an interlaboratory Concern about mercury pollution has intensified the search
method-performance (collaborative) study of a for analytical methods that require minimal sample prepara-
method for the determination of mercury in foods tion and give contaminant-free analysis with good analytical
of marine origin by flow injection–cold vapor sensitivity.
atomic absorption spectrometry after wet digestion Since the introduction of microwave digestion in 1975 (3),
using a microwave oven technique. The study was there has been significant interest in using microwave heating
preceded by a training round of samples of known to replace conventional heating during acid digestion. The
identity. The method was tested on a total of 7 sea- major advantage of microwave digestion using a
food products: blue mussel (Mytilus edulis), cod closed-vessel container is that the digestion time is signifi-
muscle (Gadus morhua), crab (Cancer pagurus), cantly reduced, and problems caused by contamination and
scampi (Nephrops norwegicus), black scabbard losses through volatilization are also minimized. Because of
fish (Aphnopus carbo), longnose velvet dogfish these advantages, the microwave digestion technique is be-
(Centroscymus crepidater), and Portuguese dog- coming more widely used in modern analytical laboratories
fish (Cenbroscymus coelolepis) with mercury con- (e.g., in a number of Nordic Committee on Food Analysis
centrations of 0.14, 0.24, 0.35, 0.59, 1.42, 4.2, and [NMKL] methods).
13.2 mg/g, respectively. The materials were pre- The most commonly used method for mercury determina-
sented to the participants in the study as blind du- tion is cold vapor atomic absorption spectrometry (CVAAS)
plicates, and the participants were asked to perform because of its ease of operation, high sensitivity, and relatively
single determinations on each sample. Repeatability inexpensive operational cost. Since it was introduced by
relative standard deviations (RSDr) for mercury Hatch and Ott (4), the method has been significantly im-
ranged from 2.4 to 14.0%. Reproducibility relative proved. More recently, several flow injection (FI) approaches
standard deviations (RSDR) ranged from 7.7 to have been reported (5–7). FI has numerous advantages such as
16.6%. HORRAT values for all samples were <1.0. a high sample throughput, low carryover effects, small sample
size and, thus, better tolerance of interferences, and also low
consumption of reagents (6, 7).
ncreased industrial activity has resulted in higher levels of The NMKL decided to undertake an interlaboratory study

I toxic metals such as mercury in the environment. These


metals accumulate in animal tissues and are eventually
taken up by humans through the food chain. The adverse toxic
of the determination of total mercury in seafood products by
FI–CVAAS, using a wet digestion procedure in a microwave
oven, in order to evaluate performance characteristics (true-
effects of mercury on humans are well documented (1). Fish ness, repeatability, and reproducibility) in the analysis of judi-
and other seafood products usually contain significantly ciously chosen seafood products. The method described in
higher concentrations of mercury than do other foods (2). The this paper has been in use and accredited by Norwegian ac-
authorities in the European Union and other countries that im- creditation authorities for several years at the Institute of Nu-
port seafood products require documentation of the mercury trition, Directorate of Fisheries, Bergen, Norway (8).
content from the producing country.
Interlaboratory Study

Received July 13, 2001. Accepted by AP September 24, 2001. A description of the method and an invitation to participate
Corresponding author’s e-mail: kare.julshamn@nutr.fiskeridir.no.
1
Nordic Committee on Food Analysis (General Secretariat, c/o National
in the study were sent to 16 laboratories in Denmark, Finland,
Veterinary Institute, Department of Food and Feed Hygiene, PO Box 8156, Iceland, Norway, and Sweden. Ten laboratories accepted and
Dep. N-0033 Oslo, Norway). agreed to follow the method procedure and the time schedule
JULSHAMN & BRENNA: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 3, 2002 627

Table 1. Expected mercury concentrations of materials Protocol


included in the study
Before the collaborative study, participants were given the
Expected values, opportunity to become familiar with the method in a pretrial
Material µg/g (dry weight)a
test. The materials included 2 certified reference materials
(CRMs), Dorm-2 (dogfish muscle) and Tort-2 (lobster
Blue mussel (Mytilus edulis) 0.14 ± 0.01 hepatopancreas), both purchased from the National Research
Cod muscle (Gadus morhua) 0.24 ± 0.02 Council Canada. The dogfish muscle contained a certified
Crab (Cancer pagurus) 0.35 ± 0.03 mercury concentration of 4.64 µg/g, and the lobster
hepatopancreas had a certified value of 0.27 µg/g. Eight labo-
Scampi (Nephrops norwegicus) 0.59 ± 0.04
ratories reported results for the CRMs. The results showed
Black scabbard (Aphnopus carbo) 1.42 ± 0.15 that some of the laboratories had problems in analyzing the
Longnose dogfish (Centroscymus crepidater) 4.2 ± 0.2 samples with acceptable trueness and repeatability. The prob-
Portuguese dogfish (Cenbroscymus coelolepis) 13.2 ± 0.4 able reason was insufficient experience with the method.
The 7 test materials used in the collaborative study were
a
Each value is the mean ± standard deviation of 10 independent packed dry in small plastic containers. They were presented to
determinations per sample.
participants as blind duplicates, that is, as 14 randomly coded
materials. The fact that blind duplicates were included in the
set of materials was not disclosed to the participants. Partici-
pants were asked to perform single determinations of the mer-
cury concentration of the materials according to the method
of the study. The design, conduct, and interpretation of the described below and to report results in mg/kg on a dry weight
study followed guidelines recommended by AOAC INTER- basis. Participants were also asked to give information about
NATIONAL (9) and NMKL (10). Participating laboratories the microwave oven and the temperature program used, as
represented industry, commercial laboratories, universities, well as to report name and model of the AAS instrument, type
and government laboratories. of FI system, background corrector, and acid purity used. The
residual moisture content of the dried test samples ranged
Study Materials from 2 to 8%. The participants were asked to determine dry
matter for separate portions and report the values in mg/kg,
Each participant received 7 test materials (samples 1–4 dry weight.
were obtained from Rieber & Søn ASA, Bergen, Norway, and
METHOD
samples 5–7 were obtained from The Institute of Marine Re-
search, Bergen, Norway): blue mussel (Mytilus edulis; meat Areas of Application
from fresh, frozen mussel that was cooked, minced, dried, and
milled to fine powder); cod muscle (Gadus morhua; cod mus- The method is applicable to the determination of mercury
in various types of seafood products. The method has been
cle powder made of fresh, frozen muscle of cod that was
tested primarily on dry products but may be used for fresh
cooked, dried, and milled to fine powder); crab meat (Cancer
samples as well.
pagurus; meat from fresh, frozen crab that was cooked, minced,
dried, and milled to fine powder); scampi powder (Nephrops Principle
norwegicus; tails from fresh, frozen Norwegian lobster that
Concentrated nitric acid and hydrogen peroxide are added
were cooked, minced, dried, and milled to fine powder); black
to samples that are then digested in a microwave oven. Any
scabbard fish (Aphnopus carbo); longnose velvet dogfish
commercially available laboratory microwave oven may be
(Centroscymus crepidater); and Portuguese dogfish
used. The concentrations of mercury are determined by
(Cenbroscymus coelolepis; meat without skin from the 3 fish FI–CVAAS using external standardization.
species were minced, freeze-dried, and milled to fine powder).
The expected values of mercury in the materials used in the Chemicals and Reagents
study were obtained from the homogeneity study (Table 1). (a) Concentrated nitric acid (HNO3).—65% Suprapur
The homogeneity of the test materials was investigated by (Merck, Darmstadt, Germany).
estimating within- and between-container variations of the 7 (b) Nitric acid, 0.65% (w/v).—Dilute 10 mL concentrated
study materials. The homogeneity test was performed by tak- nitric acid (a) to 1000 mL with water.
ing 5 subsamples of 5 g from each seafood sample. Two repli- (c) Hydrogen peroxide (H2O2).—30%, analytical quality.
cates of 0.25 g (according to the method) from each (d) Ultrapure water.—Specific resistance, >18 MΩ/cm.
subsample were digested. In total, 10 replicates of each food (e) Hg standard.—1000 mg/L (commercial standard
sample were analyzed for mercury. Results were analyzed by solution).
1-way analysis of variance (ANOVA). Values of P > 0.05 (f) Hg standard solutions.—Dilute mercury standard (e)
were obtained for all test materials, indicating that the samples to appropriate volumes with 0.65% nitric acid (b). The solu-
were homogeneous. tions are prepared fresh daily.
628 JULSHAMN & BRENNA: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 3, 2002

Table 2. Interlaboratory study results (mg/g dry matter) for the determination of mercury in seafooda
Laboratory Blue mussel Cod muscle Crab Scampi Black scabbard Longnose dogfish Portuguese dogfish

1 0.150 0.253 0.350 0.530 1.24 3.64 10.9


1 0.150 0.245 0.356 0.540 1.19 3.86 11.4
2 0.165 0.258 0.368 0.538 1.26 4.07b 11.3
b
2 0.145 0.227 0.297 0.524 1.17 2.46 11.4
3 0.152 0.255 0.337 0.598 1.42 3.61 10.5
3 0.154 0.240 0.353 0.603 1.27 3.56 10.9
4 0.133 0.227 0.373 0.576 1.45 3.81 14.2b
4 0.157 0.257 0.390 0.616 1.51 4.21 12.3b
5 0.151 0.260 0.380 0.600 1.40 4.20 14.1
5 0.150 0.270 0.380 0.620 1.50 4.46 14.0
6 0.115 0.223 0.351 0.551 1.38 4.14 12.1
6 0.113 0.209 0.328 0.544 1.37 4.01 11.8
7 0.083 0.228 0.331 0.280 1.24 4.20 13.0
7 0.102 0.190 0.295 0.510 1.32 4.10 12.9
8 0.045b 0.309 0.373 0.592 1.43 3.79 10.6
8 0.170b 0.321 0.376 0.714 1.40 3.97 11.5
9 0.151 0.318 0.429 0.647 1.49 4.32 13.9
9 0.144 0.245 0.380 0.598 1.33 4.32 13.8
10 0.118 0.228 0.334 0.541 1.30 3.40 11.2
10 0.132 0.219 0.320 0.750 1.34 3.30 11.1

a
Analyzed as blind duplicates.
b
Outlier by the Cochran test (P < 0.01).

(g) Potassium permanganate (KMnO4).—Analytical (c) Laboratory microwave oven.—Equipped with 100 mL
quality. digestion containers and capable of withstanding a pressure of
(h) Potassium permanganate solution, 5% (w/v).—Weigh 30 bar.
25 g potassium permanganate (g) into 500 mL volumetric (d) Dispenser.
flask, and dilute to volume with water (d). Prepare solution (e) Glassware.—25–1000 mL volumetric flasks; 250 mL
daily. beakers.
(i) Concentrated hydrochloric acid (HCl).—37%, analyti- (f) Polyethylene tubes or glass tubes with screw
cal grade. caps.—10–50 mL.
(j) Hydrochloric acid, 3.7% (w/v).—Dilute 100 mL con- (g) Automatic pipets with tips.
centrated HCl (i) to 1 L with water (d). (h) Atomic absorption spectrometer.—Equipped with
(k) Sodium hydroxide (NaOH).—Analytical grade. background corrector (Zeeman, deuterium lamp or
(l) Sodium hydroxide solution, 10% (w/v).—Weigh 100 g high-current HCl pulsing).
NaOH (k) into 1 L volumetric flask, and dilute to volume with (i) FI and cold vapor equipment.
water (d). (j) Autosampler.
(m) Sodium borohydride (NaBH4).—Analytical grade (tin (k) Autosampler cups.—2 mL.
chloride as alternative). (l) Electrodeless discharge lamp, hollow cathode lamp, or
(n) Sodium borohydride solution, 0.3% (w/v).—Weigh boosted hollow cathode lamp for mercury.
3 g sodium borohydride (m) into 1 L volumetric flask, add Procedure
5 mL NaOH solution (l), and dilute to volume with water (d).
(o) Argon.—With purity of $99.998%. Dry matter content has to be determined for separate test
portion by freeze-drying or by thermal drying at 105°C until
Apparatus and Equipment constant weight is obtained (e.g., 12 h). Drying does not ap-
pear to cause a loss of mercury. Weigh into the digestion con-
(a) Freeze-drier and thermal drier. tainer an amount of homogeneous sample corresponding to
(b) Analytical balance. 0.20–0.25 g dry material. Each digestion series must contain
JULSHAMN & BRENNA: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 3, 2002 629

2 reagent blanks, that is, HNO3 (Reagents, a) and H2O2 (Re-


agents, c) without sample material. If possible, include CRMs
HORRAT

0.77

0.51

0.68

0.98
0.73

0.58

0.95
containing mercury in amounts corresponding to those found
in the samples to be analyzed in order to reveal systematic or
random errors.
Note: The following section should be regarded as an ex-
Horwitz value of RSDR, %

ample. Digestion programs and amounts of acids will vary


with different digestion systems.
21.6

15.1

12.9

10.9
19.6

18.1

17.5
This program is suggested for 6 containers. For fewer or
more containers, the applied power for each step should be al-
tered proportionally, for example, for 12 containers the power
for each step should be doubled.
Digestion
R, µg/gh

0.064

0.101

0.109

0.269

0.295

0.982

3.65

Add 2 mL concentrated HNO3 (Reagents, a) and 0.5 mL


Table 3. Statistical analysis of interlaboratory study data for the determination of mercury (mg/g dry matter) in seafooda

H2O2 (Reagents, c) to each container. Seal containers and


place them in microwave oven. Start the program: S1, 250 W,
1.00 min; S2, 0 W, 1.00 min; S3, 250 W, 5.00 min; S4, 400 W,
RSDR, %g

16.6

7.7

8.8

10.7
14.3

10.9

16.6

5.00 min; and S5, 650 W, 5.00 min.


Let containers cool sufficiently before opening, and rinse
down with water any condensed water in cap and on sides of
vessels. Quantitatively transfer sample solution to 25 mL vol-
umetric flask, add KMnO4 (Reagents, h) until coloration is
sR, mg/kgf

0.023

0.104

0.347
0.036

0.038

0.095

obtained for 2–3 s, and dilute to volume with water. Then


Each value is the number of laboratories retained after elimination of outliers; each value in parentheses is the number of laboratories removed as outliers.
1.29

transfer sample solution to polyethylene tube (Apparatus, f).


The calibration of the microwave power is described else-
where (11).
Determination
r, µg/ge

0.028

0.190

0.417

0.825
0.065

0.080

0.227

If necessary, replace burner head with quartz cell. Ensure


that sample solutions do not contain any particles. Connect Hg
lamp and switch on AAS instrument. Adjust to correct wave-
length (253.6 nm) and slit (0.7 nm). Let instrument warm up
RSDr, %d

7.28

4.98

3.74

2.43
9.20

8.03

for ca 30 min. Correct gas flow of reductant, carrier solution,


14.0

and argon to obtain maximum sensitivity and repeatability.


Prepare appropriate standard solutions (i.e., 0.5, 1, 2, and
4 µg/L). Add 1 drop KMnO4 (Reagents, h) to standard solu-
tions, and dilute to 25 mL with nitric acid (Reagents, b).
sr, mg/kgc

0.010

0.067

0.147

0.292
0.023

0.028

0.080

Calculation
Calculate calibration curves for mercury by simple linear
regression on the basis of reagent blank-corrected data ob-
Mean, µg/g

tained from FI–CVAAS measurements. Calculate concentra-


0.154

0.277

0.393

0.637

1.50

4.43

tion of Hg in sample, as follows:


12.1

C = ms/w
RSDR = reproducibility relative standard deviation.
RSDr = repeatability relative standard deviation.

where C = concentration in sample (µg/g), ms = amount of


No. of labsb

sR = reproducibility standard deviation.


9(1)
9(1)

10(0)

10(0)
10(0)

10(0)

10(0)

mercury in sample solution (µg), and w = weighed amount of


sr = repeatability standard deviation.

R = reproducibility; R = 2 2 × S R .

sample (g).
r = repeatability; r = 2 2 × S r.
Analyzed as blind duplicates.

Results and Discussion


Portuguese dogfish
Longnose dogfish

The mean result from 8 laboratories in the pretrial test for


Black scabbard

mercury in dogfish muscle was 4.38 ± 0.39 mg/kg (1 standard


Blue mussel

Cod muscle

deviation [SD]); the certified value was 4.64 ± 0.26 mg/kg


Sample

Scampi
Crab

(95% uncertainty). Agreement was not found between the re-


a

h
c

f
630 JULSHAMN & BRENNA: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 3, 2002

ported results and the certified results by using z-scores, pro- Laboratory 6 reported that the method description needs a
cedure 11. The mean result from 8 laboratories for mercury in more precise formulation for “keeping the color for a mo-
lobster hepatopancreas was 0.26 ± 0.04 mg/kg (1 SD); the cer- ment.” The laboratory requested a description of the quality of
tified value was 0.27 ± 0.06 mg/kg (95% uncertainty), and an the water used for dilution. The laboratory indicated that the
agreement was found (12). determination of dry matter should have been performed with
Ten laboratories reported results in the interlaboratory separate samples to avoid losing mercury.
study (Tables 2 and 3). Results were statistically evaluated ac- Response to the comments from Laboratory 6.—It is nec-
cording to the protocol of the International Union of Pure and essary to keep the color only for 1–2 s. The text of the method
Applied Chemistry (IUPAC) for the design, conduct, and in- was corrected. Only high-quality water is used throughout the
terpretation of collaborative analytical studies (13). Results analysis. There is no indication of mercury loss by
were tested for the presence of outliers by using the Cochran freeze-drying or thermal drying at 105°C.
and the Grubbs tests. Results judged as outliers according to Laboratory 8 reported using 0.05% (w/v) potassium per-
one of these tests were not included in the final estimation of manganate solution instead of 5% (w/v) potassium permanga-
the method-performance parameters. The results for blue nate solution (Reagents, h), 3% (v/v) HCl instead of 10% (v/v)
mussel from Laboratory 8, longnose velvet dogfish from Lab- HCl (Reagents, j), 0.08% (w/v) sodium borohydride instead
oratory 2, and Portuguese dogfish from Laboratory 4 were of 0.3% (w/v) sodium borohydride (Reagents, n), standard so-
found to be outliers by the Cochran test at the P < 0.01 level: lutions with concentrations of 2.5, 5.0, 10, and 20 µg/L for the
the replicate values of 0.045 and 0.170, 4.07 and 2.46, and external standard curve, and the quartz cell warmed to 100°C.
12.3 and 14.2 µg/g, respectively, were too far apart. Laboratory 8 questioned the stability of the potassium per-
None of the data represented extreme values according to manganate solution and the Hg standard solutions.
the Grubbs test. Table 3 presents the estimated performance Response to the comments of Laboratory 8.—The potas-
characteristics of the method. The repeatability relative stan- sium permanganate solution should be prepared weekly, and
dard deviation (RSDr) of the method for the determination of the Hg standard solutions should be prepared daily. This infor-
mercury was estimated to be between 2.4 (Portuguese dog- mation is included in the text of the method.
fish) and 14.0% (scampi). The reproducibility relative stan- Laboratory 9 reported that a minimum of potassium per-
dard deviation (RSDR) was estimated to be between 7.7 (black manganate solution was used. The laboratory recommended
scabbard fish muscle) and 16.6% (scampi and blue mussel). keeping the sodium borohydride solution in an air-tight flask
The last column in Table 3 shows the HORRAT values, to minimize contact with air.
computed as follows:
Acknowledgments
RSDR/2C–0.1505
We are indebted to the following collaborators and co-
where C is the relative concentration of the analyte in ques-
workers for their skillful participation in this study:
tion. The expression in the denominator is suggested by
Horwitz et al. (14) as an empirical formula to estimate RSDR Birgitta Åsman, AgroLab Scandinavia AB, Kristianstad,
in an “acceptable” method-performance study. The formula is Sverige, Sweden
based on several thousand such studies, although no further Asta Ekman, Helsingfors Stads Miljøcentral,
details of the deviation are quoted. According to Horwitz et Miljølaboratoriet, Finland
al. (14), a series of ratios close to or consistently smaller than Anu El-Ghauoui, Lahtis Stads Kontroll-och
1.0 indicates acceptable precision of a method. Correspond- Forskningslaboratoium, Lahtis, Finland
ingly, HORRAT values consistently near or greater than Karl Olav Gjerstad, Næringsmiddeltilsynet for
2 probably indicate an unacceptable method with respect to pre- Midt-Rogaland, Stavanger, Norway
cision. The same approach has been adopted by IUPAC (13). Dag Grønningen, Veterinærinstituttet, Oslo, Norway
Table 3 shows that the present method has an acceptable Svein Harald Hammer, Næringsmiddeltilsynet i Salten,
precision for mercury determination. The HORRAT values Bodø, Norway
are <1.0 for all test materials. Arne M. Jensen, Næringsmiddelkontrollen i Trondheim,
Trondheim, Norway
Collaborators’ Comments
Helena Liukkonen-Lilja, VTT Bio-och, Livsmedelsteknik,
Laboratories 2 and 5 reported that they used tin chloride as Esbo, Finland
a reducing agent instead of sodium borohydride. Jorma Mikkola, Vanda Stads Livsmedels-och
Response to the comment of Laboratories 2 and 5.—The Milølaboratorium, Finland
method gives a choice between tin chloride and sodium Tomas Ruth, SGAB, Luleå Tekniska Univeristet, Sverige,
borohydride as proposed in the method of the European Com- Sweden
mittee for Standardization (CEN) for mercury (15). Laila O. Sedal, Fiskeridirektoratets Ernæringsinstitutt,
Laboratory 4 reported that it diluted the sample solutions to Bergen, Norway
50 mL with water and stabilized the calibration solutions to a fi- Berit Solli, Fiskeridirektoratets Ernæringsinstitutt, Bergen,
nal concentration of 1% (v/v) HNO3 and 0.5% (w/v) K2Cr2O7. Norway
JULSHAMN & BRENNA: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 3, 2002 631

Eija-Riitta Venalainen, Anstalten for Veterinæmedisin och (8) Norwegian Accreditation Authority (2000) Accreditation
Livsmedel, Helsingfors, Finland Document P072, July 31
Nina Wollertsen, Fiskeridirektoratets Sentrallaboratorium, (9) Guidelines for Collaborative Study Procedures to Validate
Bergen, Norway Characteristics of a Method of Analysis (1995) J. AOAC Int.
78, 143A–160A
(10) Nordic Committee on Food Analysis (1992) Validation of
References Methods for Chemical Analysis, Report No. 11, NMKL,
Espoo, Finland
(1) World Health Organization (1989) Mercury—Environmental (11) Julshamn, K., Thorlacius, A., & Lea, P. (2000) J. AOAC Int.
Aspects, World Health Organization, Geneva, Switzerland 83, 1423–1428
(2) Reilly, C. (1991) Metal Contamination of Food, 2nd Ed., (12) Nordic Committee on Food Analysis (2001) Evaluation of
Elsevier Applied Science, London, UK Results Derived from the Analysis of Certified Reference Ma-
(3) Abu-Samra, A., Morris, J.S., & Koirtyohann, S.R. (1975) terials, Report No. 9, NMKL, National Veterinary Institute,
Anal. Chem. 47, 1475–1477 Oslo, Norway
(13) Horwitz, W. (1988) Pure Appl. Chem. 60, 855–864
(4) Hatch, W.R., & Ott, W.L. (1968) Anal. Chem. 40, 2085–2087
(14) Horwitz, W., Kamps, L.R., & Boyer, K.W. (1980) J. Assoc.
(5) Scheuhammer, A.M., & Bond, D. (1991) Biol. Trace Elem.
Off. Anal. Chem. 63, 1344–1354
Res. 31, 119–129
(15) European Committee for Standardization (CEN) (2001)
(6) Fauske, H. (1994) Master Thesis, University of Bergen, CEN/TC 275 prEN 13806, “Foodstuffs—Determination of
Bergen, Norway, 95 pp Trace Elements—Determination of Mercury by Cold Vapour
(7) Lippo, H., Jauhiainen, T., & Peremeki, P. (1997) At. Atomic Absorption Spectrometry (CVAAS) after Pressure
Spectrosc. 18, 102–108 Digestion,” CEN, Brussels, Belgium

You might also like