Determination of Trace Element Contaminants in
Food Matrices Using a Robust, Routine
Analytical Method for ICP-MS
P. Zbinden and D. Andrey
Quality and Safety Assurance Department, Micronutriments & Additives Team
Nestlé Research Center
1000 Lausanne 26, Switzerland
INTRODUCTION
Inductively coupled plasma mass
spectrometry (ICP-MS) is a very
powerful technique for obtaining
very low trace element levels and
high sample throughput. This technique is applicable for the routine
analysis of samples in quality control and safety laboratories of the
food industry, as well as of food
regulatory laboratories. However,
ICP-MS is very sensitive to different
interferences which can lead to
inaccurate results. As food samples
are very complex matrices, interferences occurring in the analysis of
such samples can be very significant.
It is obvious that ICP-MS can
become an appropriate technique
for use in food industry laboratories
when a robust analytical method is
developed. The method developed
should not be too sensitive to the
type of sample matrix to be
analyzed.
A robust method would include
a good sample preparation method
together with a detailed study of
the potential interferences.
Interferences in ICP-MS consist
of general physical interferences,
spectral interferences (1,2), and
carbon-induced interferences (3).
The general and isobaric interferences are usually well known to
ICP-MS users.
In a routine laboratory environment it is necessary to work as fast
as possible. In trace metal determination by ICP-MS, the speed of the
analysis is dependent on the speed
at which the samples are prepared.
AS
Atomic Spectroscopy
Vol. 19(6), November/December 1998
ABSTRACT
ICP-MS is a rapid analytical
technique that shows potential
for use in routine multielemental
analysis in the food industry.
However, in order to take advantage of its high speed of analysis,
the analytical throughput should
not be slowed down by a lengthy
sample preparation step. On the
other hand, a rapid wet ashing
method may cause interferences
due to the presence of residual
carbon, particularly in the determination of As, Se and Pb.
Arsenic and selenium measured
by ICP-MS in samples where residual carbon is present may be
determined with a higher value
up to 30%. At the same time, Pb
may be determined with a value
of 10% lower.
These carbon-related interferences were quantitatively studied.
The study shows that addition of a
set concentration of isopropanol to
wet ashed samples overcomes
interferences from residual carbon.
The accuracy and reproductibility
of the determination of As, Se
and Pb by ICP-MS was improved.
A rapid and robust analytical
method for the trace determination of As, Cd, Hg, Pb, Al and Se,
well-suited to the routine environment of the food analytical
laboratory, has been developed.
Generally, ICP-MS preparation steps
require long digestion times (e.g., 3
hours) at high temperatures to
remove carbon from the sample to
minimize matrix interferences.
214
Even under these extreme
conditions, the quantity of the
residual carbon present in solution
is difficult to evaluate.
The effect of the residual carbon
on quantitative analysis is not wellknown. Potential interferences
occurring in the determination of
27Al, 75As, 114Cd, 202Hg, 208Pb, and
82Se were studied in detail.
A robust analytical method for
the trace element determination in
food useable in a routine laboratory
environment is proposed.
EXPERIMENTAL
Instrumentation
Sample preparation
HPA-S High Pressure Asher™
system (Perkin-Elmer/Paar), maximum pressure 150 bar, maximum
temperature 320°C, used with
quartz vessels.
Decontamination of the HPA-S
quartz vessels
Decontamination of the HPA-S
quartz vessels was performed with
a decontamination system
(TRABOLD, Bern, Switzerland)
using hot HNO3 vapors.
Spectrometer
An ELAN® 6000 ICP-MS (PE
SCIEX, Concord, Ontario, Canada)
was used. A thermostated cyclonic
spray chamber fitted with a concentric nebulizer (Glass Expansion,
Australia) was used instead of the
standard cross-flow nebullizer.
MS-126
Vol. 19(6), Nov./Dec. 1998
Reagents
High-purity ultrafiltered water
(18.2 MΩ, MilliQ® Plus system) was
used for dilution of the standards
and samples. Nitric acid was freshly
sub-distilled.
Reference Materials
MET 2/95, MET 6/95, Infant
Cereals; DDP 7/95, DDP 8/95, Milk
Powder. These materials were prepared by Nestlé laboratories and are
regularly used as internal reference
samples. BCR 8433, Corn Bran;
NIST 1547, Peach Leaves; NIST
1575, Pine Needles; NIST 1568a,
Rice Flour; NIST 1549, Non-Fat
Milk Powder were obtained from
PROMOCHEM, France.
Sample Preparation
All samples were prepared by
wet ashing using the HPA-S High
Pressure Asher. Except when
specified, 0.4 to 0.5 g of sample
was introduced into 15-mL quartz
HPA-S vessels, and 2 mL of subboiling nitric acid was added. The
HPA-S tubes were closed with two
PTFE strips and a quartz cap. One
strip is used to seal the tube, and
the other to close the quartz cap.
Twenty-one tubes were introduced
into the HPA-S stainless steel heating block. The HPA-S was closed
and a N2 pressure of 90 bar was
applied. The samples were then
heated according to the program
described in Table I.
TABLE I
HPA-S Heating Program
Step Initial
Time
Final
temp.
temp.
(°C)
(min)
(°C)
1
20
2
90
3
150
Total time
30
20
30
80
90
150
180
The samples were diluted with
ultrafiltered water to 10 mL in the
HPA-S quartz tubes.
RESULTS AND DISCUSSION
Automatic Addition of Indium
as Internal Standard
The normal interferences due to
the physico-chemical composition
of the sample viscosity, the difference in acid concentration or in the
quantity of matter injected are
normally corrected by the addition
of an internal standard.
After testing the different
elements as internal standards
(Nb, In, Y, Yb, Be and Ta), indium
(115In ) was found to be the best
choice for correcting the analytes
over the full range of masses.
Usually, the internal standard is
added manually to each sample,
blank, and standard. This operation
is time-consuming and prone to
manipulation errors (addition of
very small volumes). To avoid these
manipulations, a simple device was
used with the ICP-MS sample introduction system. It consists of a mixing
manifold (two ways in, one way
out) and a normal three-channel
peristaltic pump. It automatically
adds indium as an internal standard
to all sample, blanks, and standards.
This very simple system provides
interesting advantages:
•
Fewer manipulation errors and
greater sample throughout. The
internal standard is regularly
pumped at the same rate as the
sample solution. No manipulation of the samples and
standards is required.
content was always by 30% higher
in comparison to the As-certified
value (normally measured by HGAAS). The Pb value was 10–15%
lower than the value obtained by
GFAAS.
Since food samples have a high
carbon content, these problems
were presumably related to the
high carbon concentration remaining in the food samples after wet
ashing. The effect on the quantitative analysis of food matrices has
not been reported previously in
the literature.
The carbon effect on the ICP-MS
intensities of Cd, Hg, As, Pb, Se, and
In is shown in Figure 1, where the
variation of carbon concentration
was simulated by varying the concentration of alcohol and citric acid in a
standard solution. A direct relation
between analyte intensities and carbon was therefore demonstrated.
Figure 1 shows that carbon produces a strong signal enhancement
on 75As and Se (both masses tested
82Se and 77Se). The signals can be
enhanced up to seven times in the
presence of carbon in solution. This
means that measuring As and Se in
organic samples (food samples)
could lead to a value up to seven
times higher, which is unacceptable.
This enhancement effect is often
used to raise the sensitivity for As
(4,5) or Se.
Auto-dilution of the sample is
achieved. The dilution factor is
determined by the ratio of the
internal diameter of the
peristaltic tubes of both sample
and internal standard.
A change in Pb intensities was
also observed, but it is less marked
and the opposite effect occurs. For
Cd, a slight change was observed,
which follows the signal observed
for In. This suggests that for cadmium the intensity change will be
well corrected with indium as the
internal standard.
Carbon Effects on Trace
Element Determination
A systematic error was observed
in the analysis of digested organic
certified food samples. The arsenic
Carbon Effect Measured in
Food Samples
To show the carbon effect in
a more realistic analytical situation,
a Nestlé internal reference material,
•
215
Infant Cereals MET 6/95, for which
the As, Cd, Hg, and Pb concentration is accurately known, was analyzed by varying the dilution factor
of the sample. As the sample volume remains constant, a larger
amount of food sample corresponds
to more carbon in the analytical
solution.
The sample weight was varied
between 100 and 650 mg and
diluted to 10 mL. This corresponds
to a dilution factor varying from
100 to 15.
Fig. 1. Methanol, ethanol, isopropanol and citric acid effect on the ICP-MS
intensities of heavy metals (In, Y, and Se). Plot of weight (%) of alcohol or citric
acid versus analytes intensities.
A strong effect was observed
in the determination of arsenic
(Figure 2). At low dilution factors,
the measured As concentration was
higher. For a dilution near 10, the
value measured for As was 40%
higher. At low dilution factors, the
measured lead concentration was
lower. The effect on Pb is less
important as it corresponds to a
lower value of only 10%. No carbon
effect was observed in the determination of Cd and Hg.
How to Obtain Reliable
Results for 75As and 208Pb
The curve representing 75As
intensities versus isopropanol concentrations (see Figure 3) can be
split into four zones. Parts 1 and 3
are zones where relatively small
changes in carbon concentration
produced an important change in
75As intensities. On the other hand,
parts 2 and 4 are zones where
---- Control limits; .......... Reference value
Fig. 2. Effect of residual carbon on the concentration of As, Cd, Hg, and Pb
measured by ICP-MS in MET 6/95.
Fig. 3. Effect of isopropanol on 75As raw
intensities.
216
Vol. 19(6), Nov./Dec. 1998
changes in carbon concentration
result in a relatively insignificant
change for 75As intensities.
Normally, the analyses are carried out in aqueous solutions. The
normal analytical situation for food
matrices corresponds to zone 1 in
Figure 3, where the change in 75As
versus carbon concentration was
dramatic. This would explain why
As measured by ICP-MS can be
higher. This also correlates well
with the results presented in
Figure 2, which shows that the
measured As content increases
when the dilution factor decreases.
To avoid these changes in 75As
intensities, one could add known
quantities of isopropanol, corresponding to zone 2 or better to zone
4 of the isopropanol curve, where
the effects on 75As are insignificant.
Adding 2% of isopropanol to the
solution should stabilize the As
results. Adding 6% or more
isopropanol should improve the
stability of As results even more.
Nevertheless, for ICP-MS it is more
convenient to work with a low
organic solvent concentration. For
this reason, 2% of isopropanol was
added to all solutions.
Compared to Figure 2, the results
in Figure 4 show that adding 2%
isopropanol to the sample improves
both the As and Pb determination.
The results for Cd determination
remain, as expected, unchanged.
However, the determination of
mercury was unstable in 2% isopropanol. This is probably due to
electrostatic effects due to the
presence of alcohol.
To confirm the observations
obtained for the MET reference
samples, we extended this study to
other sample types such as apple
leaves, pine needles, corn bran,
infant food containing milk, peach
leaves, and rice flour.
For comparison, these samples
were measured successively in a
water solution and in 2% isopropanol.
The simultaneous determination was
extended to Se and Al (see Table II).
The results in Table II show that
the quantitative determination of
As and Pb was ameliorated in the
presence of 2% isopropanol. The Se
results were also better, although
not always well-correlated with the
certified values. This is probably
due to other interferences, which
cannot be corrected by isopropanol.
An amelioration of the repeatability
for As, Pb, and Se was also
observed. This shows that the isopropanol stabilizes the results by
stabilizing the carbon concentration
in solution.
The determination of Al and Cd
was not affected by the presence
of isopropanol.
The results show clearly that Hg
cannot be measured when isopropanol is added to the solutions. The
results obtained for the determination of Hg in normal conditions (in
water acidic solution) were also not
always good. This is especially the
case when Hg concentrations are
low. This is due to the well-known
Hg memory effect and will be the
subject of a future study.
The median of all RSDs was
calculated using the results of all
certified materials (see Table II).
These values, which can be interpreted as the in-house reproductibility, were always ≤ 5%.
Fig. 4. Arsenic, cadmium, mercury, and lead concentration measured in MET –
6/95 versus dilution factor in 2% isopropanol solutions.
217
TABLE II
Analysis of Heavy Metals in Different Types of Organic Certified Samples -Comparison Between Analysis Performed in Aqueous Soluitons (“Normal Analysis”)
and in 2% Isopropanol
Analysis N of A
Al
As
Cd
Hg
Pb
Se
N of A Al
As
Cd
Hg
Apple Leaves, NIST 1515
Peach Leaves, NIST 1546
Pb
Se
Cert-min
277,000
31
11
40
446
41
241,000
42
23
24
840
111
Cert-max
295,000
45
15
48
494
59
257,000
78
29
38
900
129
12
368,672
91
11
49
393
282
6
340,569
125
24
34
730
231
9
336,434
43
14
35
463
167
3
316,351
75
25
30
843
168
Error in water 12
±17,222
±5
±2
±2
±10
±9
6
±12,785
±10
±1
±2
±23
±39
Error in 2%
isoporpanol
±13,857
±1
±2
±1
±9
±3
3
±9052
±1
±1
±1
±24
±6
Water
Iso-OH 2%
9
Corn Bran, BCR 8433
Pine Needles, NIST 1575
Cert-min
460
0
7
2
106
37
515,000
170
n.c.
100 10,300
n.c.
Cert-max
1560
4
17
4
174
53
575,000
250
n.c.
200 11,300
n.c.
Water
3
510
3
8
46
104
54
6
591,141
207
183
78 10,144
72
Iso-OH 2%
9
636
0
13
32
138
45
6
590,780
206
188
64 10,448
65
Error in water 3
±84
±0
±1
±4
±16
±3
6
±4616
±5
±13
±2
±447
±6
Error in 2%
isoporpanol
±71
±0
±0
±3
±7
±1
6
±8903
±2
±12
±6
±147
±2
9
Infant Cereals Product, MET 2/95
Rice Flour, NIST 1568a
Cert-min
n.c.
408
460
417
160
n.c.
3400
260
20
5
n.c.
340
Cert-max
n.c.
639
531
472
339
n.c.
5400
320
24
6
n.c.
420
Water
6
n.a.
648
482
450
212
n.a.
6
3705
277
21
44
–16
347
Iso-OH 2%
3
n.a.
525
494
141
228
n.a.
6
3978
296
30
30
4
358
Error in water 6
n.a.
±7
±4
±7
±17
n.a.
6
±156
±3
±1
±3
±21
±10
Error in 2%
isoporpanol
n.a.
±3
±5
±3
±17
n.a.
6
±33
±3
±0
±2
±3
±2
3
Infant Cereals Product, MET 6/95
Cert-min
n.c. 1538
147
120
898
n.c.
Cert-max
n.c. 1769
218
175
1077
n.c.
Water
6
n.a.
2001
180
149
904
n.a.
Iso-OH 2%
6
n.a. 1719
186
–14
978
n.a.
Error in water 6
n.a.
±79
±2
±2
±25
n.a.
Error in 2%
isoporpanol
n.a.
±12
±1
±0
±11
n.a.
6
All results are expressed in mcg/kg;; n.c. = not certified; n.a. = not analyzed; N of A = Number of analyses.
218
Vol. 19(6), Nov./Dec. 1998
CONCLUSION
REFERENCES
A new method is proposed for
the analysis of food samples, which
minimizes the effect of carbon on
As, Se, and Pb by adding 2% isopropanol to the analytical solution.
1.
Meng-Fen Huang and H. Jiang,
J. Anal. At. Spectrom. 10, 31
(1995).
2.
L. Ebdon, J. Anal. At. Spectrom.
9, 611 (1994).
Cadmium can be determined
either in water or in isopropanol
with similar results. Mercury is better
determined in aqueous solution,
because of the poor repeatability and
accuracy observed in 2% isopropanol.
3.
J. Campbell, C. Demesmay and
M. Ollé, J. Anal. At. Spectrom.
9, 1379, (1994).
4.
P. Thomas, J. Anal. At. Spectrom.
10, 615 º1995).
5.
E.H. Larsen and S. St¸rup, J. Anal.
At. Spectrom. 9, 1099 (1994).
The simple sample preparation
suggested provides reliable precision and accuracy in the ICP-MS
determination of toxic minerals. It
can be applied to a wide variety of
food matrices and is well-suited for
routine food analysis.
219