Scand J Clin Lab Invest 2000; 60: 677 ± 686

Validation of inductively coupled plasma atomic

emission spectrometry technique (ICP-AES) for
multi-element analysis of trace elements in
human serum
R . RA HI L -K HA ZE N, H . H EN R I KSE N, B . J . BO L ANN & R . J . UL VI K
Institute of Clinical Biochemistry, Haukeland University Hospital, Bergen, Norway

Rahil-Khazen R, Henriksen H, Bolann BJ, Ulvik RJ. Validation of inductively

coupled plasma atomic emission spectrometry technique (ICP-AES) for multi-
element analysis of trace elements in human serum. Scand J Clin Lab Invest
2000; 60: 677± 686.

The use of inductively coupled plasma atomic emission spectrometry (ICP-AES)

for the simultaneous determination of Al, B, Ba, Be, Cd, Co, Cr, Cu, Fe, Li,
Mn, Ni, Pb, Se, Sr and Zn in human serum in a clinical laboratory was
validated. Samples were digested and then analysed using yttrium as an internal
standard and a serum-matched calibration standard. The criteria used to assess
the analytical performance of the ICP-AES were detection and quanti® cation
limits, linearity, sensitivity, recovery, interference from alkali and acid, trueness
and precision. Detection limits were 0.002 ± 0.003 mmol/L for Mn, Sr, Ba, and
Cd; 0.014 ± 0.07 mmol/L for Co, Zn, Fe, Be, Li, Pb, Cu, Ni, and Cr; and 0.2 ±
0.9 mmol/L for B, Se, and Al. Trueness, as controlled by analysis of bovine
serum certi® ed reference material, was acceptable for Co, Cu, Se and Zn, while
Fe was 5.1% and Mn 6.2% below the lowest limit of the certi® ed material
interval. We conclude that ICP-AES can be used for multi-element analysis of B,
Ba, Cu, Fe, Li, Se, Sr and Zn in serum. Serum levels of Al, Be and Co were
below the detection limits while serum levels of Cd, Cr, Ni and Pb were below
the quanti® cation limits of the ICP-AES. These trace metals cannot be analysed
as routine by the ICP-AES. However, in cases of intoxication with elevated
serum concentrations mean recovery of 100¡10% was obtained at an addition
of 2.22 mmol/L for Al, 0.11 mmol/L for Be, 0.03 mmol/L for Co, 0.39 mmol/L for
Cr, 0.14 mmol/L for Ni, and 0.12 mmol/L for Pb.

Key words: Inductively coupled plasma atomic emission spectrometry; method

validation; trace element analysis; trace element serum levels

Rune J. Ulvik, Haukeland University Hospital, Institute of Clinical Biochemistry,

N-5021, Bergen, Norway. Tel. z47 5597 3149, fax. z47 5597 3115, e-mail.
rulv@haukeland.no

677
678 R. Rahil-Khazen et al.
I N TR OD UC T I ON
Up to now, atomic absorption spectrometry
(AAS) has been the method of choice in clinical
laboratories for quanti® cation of trace elements
in blood and biological tissues. However, there
has been an increased interest in evaluating the
usefulness of atomic emission from high-energy
sources like inductively coupled plasma (ICP).
It remains to be seen what role this method will
attain in the laboratories in competition with
AAS [1].
Inductively coupled plasma (ICP) is an
electrical conducting gaseous mixture composed
of cations and electrons and is obtained when
argon is converted to plasma in a high-energy
radio-frequency ® eld, giving rise to an excita-
tion temperature of 6500 ± 7000 K. ICP can be
used as a source for emission of speci® c spectral
lines from atoms and ions formed by metals
which are introduced into the plasma in an
aerosol suspended in argon gas [2, 3]. The two
to three times higher temperature in the
analytical region of ICP gives a more complete
atomization and fewer chemical interferences
than in AAS.
In contrast to conventional ¯ ame or electro-
thermal AAS, the high-energy ICP-technique
allows simultaneous recording of a range of
different spectra, which enables multi-element
analysis of a single sample. Further, the linear
concentration range is several-fold greater than
for AAS.
The two principal methods in current use for
detection of spectra emitted from ICP, are mass
spectrometry (MS) and atomic emission spec-
trometry (AES). ICP-MS is traditionally
regarded as a more versatile method than
ICP-AES. However, since ICP-MS is a much
more complex and expensive technique than
ICP-AES, it is of interest to ® nd out for which
analytical purposes ICP-AES may be an
adequate alternative [4 ± 6].
Since the introduction of the ® rst commercial
ICP-AES instruments at the beginning of the
1970s, the technique has been signi® cantly
developed. The greatest progress has been
achieved in the last 4 ± 5 years by introducing
the axial plasma torch, solid state detectors and
the cyclone spray chamber to improve the
aerosol production. As a result, modern instru-
ments have better long-term stability, higher
resolution and lower detection limits than
instruments manufactured at the beginning of
the 1990s [4, 7, 8].
In addition to important, technical improve-
ments, the matrix problems and the effect of
using external and internal standards have been
thoroughly addressed in the literature [9 ± 12].
Up to now, ICP-AES has been primarily used
in industrial rather than in medical laboratories.
Considering the improved analytical perfor-
mance of modern instruments, ICP-AES may
be a potential alternative or supplement to AAS
in clinical laboratories with interest in trace
element analysis.
Most of the published reports about analy-
tical performance refer to the ICP-MS techni-
que. We lack a recent and more complete
validation of the ICP-AES method useful for
the clinical laboratory. Therefore, the aim of
this study was to validate the ICP-AES method
to see if it may prove useful for analytical
purposes in a clinical chemical laboratory.
M AT ER I A LS AN D M ET H ODS
Chemicals
Multi-element standard solutions were pre-
pared from Spectrascan single element stock
solutions (1000 mg/L) purchased from Tekno-
lab, Drù bak, Norway. Calibration standards
were made by successive dilutions of the stock
solution to obtain a ® nal concentration of
2500 mg/L for Cu (39.3 mmol/L), Fe (44.8 mmol/
L) and Zn (38.2 mmol/L); 500 mg/L for Se
(6.3 mmol/L); 300 mg/L for Al (11.1 mmol/L),
B (27.7 mmol/L) and Pb (1.4 mmol/L); 100 mg/L
for Li (14.4 mmol/L), Ba (0.7 mmol/L), Be
(11.1 mmol/L), Co (1.7 mmol/L), Mn
(1.8 mmol/L), Ni (1.7 mmol/L) and Sr
(1.1 mmol/L); and 50 mg/L for Cd (0.4 mmol/L)
and Cr (1.0 mmol/L).
As internal standard, yttrium was used at a
concentration of 1000 mg/L (11.2 mmol/L). It
was measured at three different wavelengths
242.220, 324.228 and 417.754 nm to cover the
wavelength range of the elements. Ultrapure
nitric acid and suprapure hydrogen peroxide
were from Merck, Darmstadt, Germany.
Seronorm Trace Elements Serum (Nycomed
AS, Oslo, Norway) was used as a commercial
control material.
Bovine serum (SRM 1598) from the National
 ICP-AES technique validation 679

Institute of Standards and Technology (NIST) Expansion Pty. Ltd., Australia was used. The
(Gaithersburg, USA) was used as a standard operating conditions are given in Table I.
reference material. Microwave digestion was carried out using
To imitate the concentrations found in serum, the Milestone microwave system (MLS 1200
suprapure sodium chloride, potassium chloride MEGA, Italy). The Milestone VAC-60 vacuum
and calcium carbonate from Merck, Darmstadt, module was used to evaporate the liquid in the
Germany and magnesium from Spectrascan samples.
were added to the calibration standards and
the blanks to give a ® nal concentration of
3200 mg/L (139.2 mmol/L) for sodium, 200 mg/
L (5.1 mmol/L) for potassium, 100 mg/L Microwave digestion and evaporation procedure
(2.5 mmol/L) for calcium and 20 mg/L Each microwave digestion procedure was
(0.8 mmol/L) for magnesium. carried out using six samples of 3 ml each. To
The acidity in the samples after the digestion each sample 2 ml nitric acid, 1 ml of hydrogen
and evaporation process was measured by peroxide and yttrium (11.2 mmol/L) was added.
titration against alkali with mixed indicators Although we used a sample volume of 3 ml
using the acidity test reagent kit Aquamerck 11 throughout this study, using 1.5 to 2 ml is also
108, Merck, Darmstadt, Germany. possible.
Laboratory glassware, plastic bottles, diges- After completion of the digestion procedure
tion vessels and covers were all soaked in 1.4 N using the program presented in Table II, the
solution of nitric acid overnight (or at least half vessels were cooled in cold water for 20 min.
an hour for the vessels and covers) and then They were then placed in the vacuum module
washed in de-ionized water supplied from a and evaporated (Table II) until a ® nal sample
Milli-Q system (Millipore Corporation, Bed- volume of 0.6 ± 1 ml was obtained. Thereafter,
ford, USA) with a resistance close to the content of each vessel was transferred to a
18 MVcm± 1. 12 ml polypropylene tube (Sarstedt, NuÈm-
brecht, Germany) and diluted with de-ionised
Instrumentation water to 3 ml before it was nebulized into the
ICP.
Trace element concentrations were measured Serum samples, control materials and the
using a programmable simultaneous plasma reference material were all treated in the same
emission spectrometer (IRIS/AP ICP-AES) way mentioned above.
from Thermo Jarell-Ash, Franklin, MA, USA The conditions described here and in Table II
with a ``charged injection device detector’’ and were the result of a systematic study where the
an axial viewing mode. A microconcentric ratio of sample, acid and peroxide, and the
nebulizer (Micromist nebulizer) from Glass microwave programs used in the digestion and
evaporation procedure, were modi® ed until the
sample appeared as a faint yellow, clear
TABLE I. Operating conditions.
solution.
Plasma frequency 27.12 Mhz
Rf Power 1150 watt
Torch gas ¯ ow 0.56 L/min
Auxiliary gas ¯ ow 0.5 L/min TABLE II. Microwave oven programs used.
Plasma viewing mode Axial
Nebulizer Micromist nebulizer Digestion program Evaporation program
Nebulizer gas ¯ ow 0.56 L/min
Spray chamber Cyclone Time Power Time Power
Nebulizer uptake 0.5 ml/min (min.) (W) (min.) (W)
Sample volume used 1.5 ml/2 repeats
Flush time 40 s 2 250 2 200
Purge time 90 s 1 0 3 300
Signal integration time 60 s High wavelength 5 250 7 400
10 s Low wavelength 5 400 6 450
Detector Charge-injection device 7 650
680 R. Rahil-Khazen et al.

Analysis of trace elements in human serum on the signal intensity of different elements, we
decided to use a calibration standard prepared
Serum samples were taken from human
in 4N HNO3.
serum pools gathered from patients in the
Fig. 1 shows how the signal intensity of the
hospital. The following 16 trace elements were
elements was suppressed in the presence of 4N
studied at the same time at their speci® c
HNO3 (St0). Taking the response of each
wavelengths given in Table III: Al, B, Ba, Be,
element prepared in 0.1N acid as 100%, a
Cd, Co, Cr, Cu, Fe, Li, Mn, Ni, Pb, Se, Sr, and
decrease in the intensities of the different
Zn. Some elements were studied at an addi-
analyte signals ranging from about 4% to
tional wavelength: Ba at 233.527, Cd at 226.502
more than 20% were seen in the St0 (black
and 228.802, Co at 228.616, Cu at 327.396, Fe
bars). Al, Li and Se were the elements most
at 239.562, Mn at 260.569, Sr at 421.552, and
affected by the acid.
Zn at 213.856 nm.
Seven different levels of trace elements in a
constant serum matrix were obtained by dilut- Use of matrix matched calibration standard
ing serum with added standard with the same
It is known that the alkali elements Na, K,
serum devoid of standard. The procedure was
Ca and Mg which are naturally present in
repeated on three different days. Seronorm was
serum, suppress the signal intensity of trace
used regularly as a control.
metals when analysed by the ICP-method [13 ±
16]. As shown in Fig. 1, when the multi-element
standard with 4N HNO3 (St0) was enriched
R ES UL TS
with Na, K, Ca and Mg (StCaNaMgK) at the
concentrations as described in Chemicals, the
Effect of acid on the signal intensity of the
signal intensity of all elements except Li, was
elements
depressed (dotted bars).
The multi-element calibration standard used
was originally prepared in 0.1N HNO3 acid.
Yttrium as internal standard
The measured acidity in the digested samples,
however, ranged between 3 and 5.5N. Based on By using yttrium as internal standard the
a thorough study of the effect of various relative intensity of the elements was obtained
concentrations of acid between 3 and 5.5 N by dividing their intensity at the respective

TABLE III. Detection limits and quanti® cation limits for the trace elements at their respective wavelengths.

Element Wavelength (nm) Detection Limits (mmol/L)a Quanti® cation Limits (mmol/L)b

Al 396.152 0.9 3.1

B 249.678 0.2 0.7
Ba 493.409 0.003 0.010
Be 313.107 0.024 0.080
Cd 214.438 0.003 0.010
Co 237.862 0.014 0.046
Cr 267.716 0.07 0.23
Cu 324.754 0.05 0.16
Fe 259.940 0.02 0.07
Li 670.784 0.04 0.12
Mn 257.610 0.002 0.005
Ni 231.604 0.05 0.16
Pb 220.353 0.038 0.126
Se 196.090 0.29 0.97
Sr 407.771 0.002 0.008
Zn 206.200 0.015 0.053
a
Calculated as 36SD of the concentration of a blank solution matched to the digested serum matrix with
respect to alkali elements and acid, and measu red 20 times.
b
Calculated as 106SD of the concentration of a blank solution matched to the digested serum matrix with
respect to alkali elements and acid, and measu red 20 times.

FIG. 1. Intensity and relative intensity of the differ-
ent elements in St0 (a multi-element calibration stan-
dard prepared in 4N HNO3) and StCaNaMgK (a
multi-element calibration standard prepared in 4N
HNO3 with added alkalis) taking 0.1N HNO3
multi-element standard as 100%. Intensities of St0
and StCaNaMgK are shown as and respectively;
relative intensities to yttrium of St0 and
StCaNaMgK are shown as and respectively.
wavelength by the intensity of yttrium. The
relative intensity measured for some selected
elements in the multi-element standards St0 and
StCaNaMgK is shown in Fig. 1 as grey bars and
oblique striped bars respectively. The relative
intensity of each element prepared in the 0.1N
HNO3 multi-element standard was taken as
100%.
Detection and quanti® cation limits
The detection limits and the quanti® cation
limits were calculated as 36SD and 106SD,
respectively, of the concentration of a blank
solution matched to the digested serum matrix
with respect to alkali elements and acid, and
measured 20 times (Table III).
ICP-AES technique validation 681
Linearity
The multi-element calibration standard in 4N
HNO3 and containing Na, Ca, Mg and K (i.e.
StCaNaMgK) was tested for linearity up to a
concentration of 18.5 mmol/L for Al, 46.2 mmol/
L for B, 7.3 mmol/L for Ba, 11.1 mmol/L
for Be, 0.9 mmol/L for Cd, 1.7 mmol/L for
Co, 1.9 mmol/L for Cr, 78.7 mmol/L for
Cu, 89.5 mmol/L for Fe, 72.0 mmol/L for Li,
1.8 mmol/L for Mn, 1.7 mmol/L for Ni,
2.4 mmol/L for Pb, 6.3 mmol/L for Se,
1.1 mmol/L for Sr, and 76.5 mmol/L for Zn.
Linear regression analysis gave regression
coef® cient R240.998 for all the elements in
the concentration ranges mentioned above.
A series of seven serum samples with a
constant matrix, but different levels of elements,
was prepared by diluting a serum spiked with
elements with the same serum without added
elements. The concentration range of each
element is presented in Table IV.
The concentrations of the added elements
were plotted against the measured values.
Regression coef® cients (R2) between 0.973 and
0.999 were obtained, with the highest coef® -
cients 40.999 for B, Ba, Be, Co, Mn and Sr,
and the lowest values 0.973 for Cu (Table IV).
Trueness
The results of analysis of a human serum
control (Seronorm) and a bovine serum stan-
dard reference material (NIST) with known
concentrations of trace elements are given in
Table V.
Of the seven concentration levels tested, the
minimum level for each element that gave a
mean recovery of 100¡10% or better of three
repeats, is presented in Table VI. Similar results
were obtained at higher concentration levels,
but with even better precision.
The inherent content of trace elements in
serum was calculated by the standard addition
method [17] and compared with measured
values (Table VI). In the standard addition
method, increasing concentrations of an added
element were plotted against their relative
intensities measured by the ICP-AES at the
levels of addition. Linear regression analysis
was performed, and by extrapolating the
regression line to the x-axis the inherent
concentration of the element in the serum was
682 R. Rahil-Khazen et al.

TABLE IV. Linearity was tested by diluting serum samples supplemented with a multi-element standard with
increasing amounts of serum samples devoid of the standard. Linear regression analysis in the speci® ed con-
centration ranges measured on three different days with the slopes, intercepts and the regression coef® cient
(R2) is given.

Elem Concentration Range mmol/L R2 (Lowest value of 3 repeats) Slope (Range) Intercept (Range) mmol/L

Al 0± 5.56 0.997 1.04 ± 1.10 ± 0.02 ± 1.08

B 0± 13.87 0.999 1.05 ± 1.08 1.93 ± 2.05
Ba 0± 1.46 0.999 0.96 ± 1.00 0.51 ± 0.53
Be 0± 2.77 0.999 0.99 ± 1.04 ± 0.02 ± 0.00
Cd 0± 0.22 0.984 0.97 ± 1.00 0.00 ± 0.01
Co 0± 0.85 0.999 1.00 ± 1.04 0.00 ± 0.01
Cr 0± 0.96 0.994 0.95 ± 1.02 0.13 ± 0.23
Cu 0± 7.87 0.973 0.91 ± 1.00 20.60 ± 21.06
Fe 0± 8.95 0.996 1.02 ± 1.09 15.88 ± 16.48
Li 0± 7.20 0.995 0.99 ± 1.06 4.68 ± 4.94
Mn 0± 0.46 0.999 0.96 ± 1.00 0.02 ± 0.04
Ni 0± 0.85 0.980 0.83 ± 0.93 0.03 ± 0.10
Pb 0± 0.48 0.991 0.96 ± 0.96 ± 0.08 ± ± 0.04
Se 0± 3.8 0.996 1.03 ± 1.06 1.01 ± 1.03
Sr 0± 0.29 0.999 0.99 ± 1.01 0.27 ± 0.27
Zn 0± 7.65 0.987 0.97 ± 1.04 11.67 ± 12.42

calculated. A representative plot for the stan-
dard addition procedure is presented in Fig. 2,
taking Mn as an example.
Precision
The data from the within-run and the day-to-
day precision studies showed as expected, that
the coef® cient of variation (CV) was highest for
elements at concentrations close to their detec-
tion limits, such as Cd and Cr (Table VII). Also
Mn at a level three times above the quanti® ca-
tion limit, and to a lesser extent Li had
relatively high CV. B, Ba, Cu, Fe, Sr and Zn
had CV55%, while Se had a CV510%.
The day-to-day precision was done by
measuring 21 samples taken from the same
serum pool, but treated separately and mea-
sured on 21 different days. Thus the variation
re¯ ects both sample treatment and measure-
ment. One outlier 43 SD from the mean was
removed.

TABLE V. Analysis of human serum control material and NIST bovine serum reference material.

Control material Reference material

Recommended Measured mean Certi® eda Measured

Element valuea (range) mmol/L (range)b mmol/L mmol/L mmol/L

Al 2.5 (2.1 ± 2.9) 2.5 (2.0 ± 2.8) 0.14¡0.03c ND

Cd ± ± 0.0008¡0.0001c ND
Co 0.001 ± 0.006c ND 0.022¡0.003 0.022
Cr 0.02 ± 0.03c ND 0.003¡0.002c ND
Cu 20.5 (18.7 ± 21.7) 20.2 (19.4 ± 21.5) 11.7¡0.6 12.1
Fe 19.7 (19.2 ± 19.9) 21.3 (20.2 ± 21.9) 46.98¡1.84 42.85
Mn 0.11 (0.09 ± 0.13) 0.09 (0.07 ± 0.13) 0.070¡0.006 0.060
Ni 0.04 ± 0.06c ND ± ±
Se 1.01 (0.89 ± 1.17) 1.01 (0.96 ± 1.09) 0.55¡0.05 0.52
Zn 22.6 (20.6 ± 25.5) 21.0 (20.0 ± 22.3) 14.0¡0.9 14.2
a
Determined mostly by Electrothermal Atomic Absorption Spectrometry and F lame Atomic Absorption
Spectrometry (see discussion).
b
M ean and range of 8 repeats, for Se mean and range for 6 repeats are given.
c
Values are below ICP-AES detection limits.
 ICP-AES technique validation 683

TABLE VI. Pooled serum levels in (mmol/L) calculated by the standard addition method vs. those measured
and percentage recovery in spiked serum at the level of element addition selected. The mean and range of
three repeats done on three different days are given.

Serum level (mmol/L)

Standard addition Measured Mean Amount Added % Mean Recovery

Element method Mean (Range) (Range) mmol/L (Range)
a
Al (0.02 ± 1.26) 2.22 (101 ± 111)
B (1.93 ± 2.08) (2.13 ± 2.25) 1.39 (91 ± 99)
Ba (0.52 ± 0.55) (0.51 ± 0.52) 0.06 (99 ± 116)
a
Be (0.003 ± 0.019) 0.11 (100 ± 104)
Cd (0.004 ± 0.006) (0.003 ± 0.004)bc 0.01 (78 ± 97)
a
Co (0.002 ± 0.014) 0.03 (86 ± 97)
Cr (0.12 ± 0.26) (0.12 ± 0.20)c 0.39 (95 ± 103)
Cu (18.9 ± 22.6) (20.4 ± 21.1) 1.97 (88 ± 106)
Fe (15.4 ± 16.3) (15.8 ± 16.6) 0.90 (73 ± 116)
Li (4.4 ± 5.0) (4.7 ± 4.8) 2.88 (106 ± 113)
Mn (0.029 ± 0.046) (0.024 ± 0.044) 0.05 (95 ± 97)
Ni (0.002 ± 0.104) (0.019 ± 0.085)bc 0.14 (78 ± 105)
c
Pb (0.043 ± 0.082) 0.12 (92 ± 110)
Se (0.96 ± 1.06) (1.02 ± 1.15) 0.63 (78 ± 104)
Sr (0.269 ± 0.272) (0.269 ± 0.270) 0.03 (93 ± 110)
Zn (11.7 ± 12.8) (11.3 ± 12.4) 1.27 (106 ± 126)
a
Values are below ICP-AES detection limits.
b
Values are near ICP-AES detection limits.
c
Values are below ICP-AES quantiŽ cation limits.

D I SC US SI O N procedure is required when the concentration of

A major problem with the ICP technique is the
tendency of organic matrix constituents to
interfere with nebulization and formation of
aerosol from the sample material in the
instrument. In general, these problems are
overcome by extensive dilution or acid digestion
of the sample. The more cumbersome digestion
FIG. 2. Representative plot of the standard addi-
tion procedure demonstrating the relative intensity
measured in the serum matrix. The ® gure shows the
regression line and the extrapolation to the x-axis
giving the serum level as calculated by the standard
addition method taking Mn as an example.
the elements to be analysed is so low that
dilution will give a concentration below the
detection limit of the instrument. Since we
wanted to validate the ICP-AES instrument for
a range of trace metals including those at very
low concentrations in serum, we had to perform
acid digestion of serum before analysis.
First it was necessary to de® ne the optimal
conditions for digestion of serum and modi® ca-
tion of the calibration standard to match the
serum sample with respect to the content of
alkali elements and acid. The program for
digestion and evaporation presented in Table II
is the result of a systematic examination to ® nd
the optimal pre-analytical conditions. With
20 min cooling of the sample after digestion,
the pre-analytical procedure was completed in
about 1 h.
Residual inorganic matrix constituents in the
digested sample prone to give spectral inter-
ference in the ICP, were acid and the alkali
elements Na, K, Ca and Mg inherent in serum.
Fig. 1 shows that the suppression of the signal
intensity caused by acid and alkali elements
could be suf® ciently compensated by the use of
yttrium as an internal standard for some
684 R. Rahil-Khazen et al.
TABLE VII. Within-run and day-to-day precision measured in the serum pool.
Element Within-run CV(%)a Day-to-day CV(%)b
B 2.0
Ba 0.6
Cd 23.7
Cr 7.8c
Cu 0.6
Fe 0.5
Li 12.9
Mn 14.7
Se 9.3
Sr 2.3
Zn 0.3
a
Based on 20 repetitive measurements of a digested serum samp le.
b
Based on 20 measurements over 20-day period (one outlier 43SD was removed).
c
Based on 10 repetitive measu rements
d
Level below quantiŽ cation limit of the ICP-AES
elements, but not for all. The elements that were
best compensated by the internal standard were
in falling order Zn4Se4Cd4Cr4Ba. The
remaining elements were not adequately com-
pensated by the internal standard, therefore the
use of the matrix matched calibration standard
was necessary.
Use of a matrix matched calibration standard
excludes the simultaneous determination of the
alkali elements Na, K, Ca and Mg. This is not a
serious objection, however, since the alkali
elements are usually measured by other analy-
tical techniques in the clinical laboratory.
Our detection limits for Cu, Fe, Zn, Li, Mn
and Sr (Table III) are generally lower than or
comparable to the values in other studies [18 ±
24]. For Mn especially, they are comparable to
those of graphite furnace atomic absorption
spectrometry [25 ± 28] and lower than other
ICP-AES techniques [29].
As demonstrated in Table VI, 12 out of 16
trace elements could be simultaneously quanti-
® ed in serum above their detection limits,
covering a concentration range with a 5200
fold difference between the lowest and highest
level represented by Cd and Cu, respectively.
The serum concentration of Al, Be and Co as
determined by the standard addition method,
was below the detection limit, while Pb was
close to the detection limit. The concentration
of Cd, Cr and Ni were measured 2 ± 3 fold
below the quanti® cation limit, at the level of the
detection limit for Cd and Ni, and about 2-fold
above the detection limit for Cr. The concen-
trations in serum of Ba, Cu, Fe, Li, Sr and Zn
Mean serum pool level (mmol/L)
4.5 2.3
1.5 0.67
18.0 0.01d
39.6 0.09d
2.3 19.68
1.9 17.85
9.0 1.54
26.6 0.01
9.1 1.62
4.6 0.27
1.3 14.5
were measured at a level varying from 30 to 245
times above the quanti® cation limit with Sr in
the lower and Fe in the upper end of the range.
The concentrations of B and Mn were measured
about three and seven times above the quanti-
® cation limit respectively, while the concentra-
tion of Se was determined very close to its
quanti® cation limit.
The linearity was satisfying for all elements
with slopes and detection limits that indicated
acceptable sensitivity. Further, recoveries of
100¡10% were obtained by the addition of
rather small amounts of metal (Table VI).
Neither linearity nor recovery was affected by
the inherent serum concentration of trace
metals, which differed substantially between
the metals. The analytical trueness was also
controlled by analysing the Seronorm human
serum control material with recommended
values for some of the trace elements, and a
bovine serum from NIST with certi® ed values
(Table V). These were the only commercial
control and reference materials available.
Except for Fe, none of the elements in the
control and reference material had been ana-
lysed by ICP-AES. The most frequent method
used was electrothermal atomic absorption
spectrometry (EAAS). Some elements were
analysed by ¯ ame atomic absorption spectro-
metry (FAAS); and particularly in the NIST-
material, elements were analysed by several
other methods like mass spectrometry and
neutron activation analysis in addition to
AAS [30]. Although the elements in the control
and reference materials were analysed by other

methods, there was an acceptable agreement
between the given intervals and measured values
(Table V). Co, Cu, Se and Zn were measured
within the given NIST-interval, while Fe was
5.1% and Mn 6.2% below the lowest limit of the
NIST interval.
The within-run coef® cient of variation was
between 10% and 15% for Mn and Li, and
about 24% for Cd. The other elements had
variation coef® cients below 10% (Table VII).
Cd and Mn had high day-to-day coef® cients of
variation, but the highest value was found for
Cr. Particularly low variation coef® cients were
obtained for Cu and Zn which are frequently
analysed trace metals in medicine. Also, Se had
coef® cients of variation below 10% both within-
run and from day-to day. In general, and not
unexpected the elements at the highest concen-
tration levels also had the lowest variation
coef® cients. Precision for Cu, Fe and Zn was
high in comparison to other studies [18, 22], but
lower than expected for Mn as compared to
graphite furnace atomic absorption spectro-
metry [26, 27] at the low levels studied.
By using the standard addition method (Fig.
2) it is possible to estimate the concentration of
trace metals in serum at levels below the
detection limit. This was done for Al, Be, Cd,
Co and Pb that were below or very close to the
detection limit. For the 9 elements which are
present in serum at concentrations above the
quanti® cation limit, there was a fairly good
correlation between the results obtained by the
standard addition method and measured values
for Ba, Cu, Fe, Li, Sr and Zn (Table VI). For B
and Se measured values were on average about
8% and 7% higher than those estimated from
the standard addition method; while for Mn the
measured value was 8% lower than the value
estimated from the standard addition method.
Thus, the standard addition method can be
used to analyse trace metals at levels below their
detection limits. Also, it can be used to control
results that appear between the detection and
quanti® cation limits.
In this study we have validated the following
criteria to assess the analytical performance and
usefulness of the ICP-AES method: detection
and quanti® cation limits, linearity, sensitivity,
recovery, interference from alkali and acid,
trueness and precision. Acceptable speci® city
emerges from the selection of speci® c spectral
lines for the elements.
ICP-AES technique validation 685
Of the 16 elements that were studied, we
found that the ICP-AES can be used for multi-
element analysis of the following elements in
serum: B, Ba, Cu, Fe, Li, Se, Sr, and Zn. The
precision of the Mn analysis was not as good as
desired, but given the uncertainty of the
analysis, also determination of Mn can be
recommended.
The concentrations of Al, Be and Co in
serum were below the detection limits and the
serum levels of Cd, Cr, Ni, and Pb were below
the quanti® cation limits of the ICP-AES.
Therefore, these trace metals can not be
analysed as routine by the ICP-AES. However,
in cases of intoxication with elevated serum
concentrations above the quanti® cation limits,
and by using the standard addition method, it is
even possible to get an idea of the serum levels
of these metals by using the ICP-AES.
A CK NO WL E DG EM E NT S
The authors thank Professor Rolf Isrenn and
Ms Torill Standal for providing technical help,
and the Section of Special Analysis in the
Laboratory of Clinical Biochemistry, Hauke-
land University Hospital, for their assistance.
R EF E RE NC ES
1 Milne DB. Trace elements. In: Burtis CA, Ash-
wood ER, editors. Tietz text book of clinical
chemistry, 3rd ed. Philadelphia: W. B. Saunders;
1999. p. 1029 ± 55
2 Houk RS. Mass spectrometry of inductively
coupled plasmas. Anal Chem 1986; 58: 97A ±
105A.
3 Schramel P. Atomic emission spectrometry. In:
Herber RFM, Stoeppler M, editors. Trace element
analysis in biological specimens. Amsterdam:
Elsevier Science B. V.; 1994. p. 91 ± 127
4 Broekaert JAC. Trends of development in plasma
techniques for atomic spectrometric analysis: a
review. Mikrochimic Acta 1995; 120: 21 ± 38.
5 Dean JR. Comparison of techniques. In: Ando
DJ, editor. Atomic absorption and plasma spec-
troscopy, 2nd ed. Chichester: John Wiley; 1997. p.
155 ± 9
6 Rommers P, Boumans P. ICP-AES versus (LA-)
ICP-MS: competition or a happy marriage? A
view supported by current data. Fresenius J Anal
Chem 1996; 355: 763 ± 70.
7 Marshall J, Chenery S, Evans EH, Fisher A.
Atomic spectrometry update--atomic emission
686 R. Rahil-Khazen et al.
spectrometry. J Anal At Spectrom 1998; 13:
107R ± 30R.
8 Mermet JM, Poussel E. ICP emission spectro-
meters: 1995 analytical ® gures of merit. Appl
Spectrosc 1995; 49: 12A ± 8A.
9 Chudinov EG, Ostroukhova II, Varvanina GV.
Acid effects in ICP-AES. Fresenius Z Anal Chem
1989; 335: 25 ± 33.
10 De Boer JLM, Velterop M. Empirical procedure
for the reduction of mixed-matrix effects in
inductively coupled plasma atomic emission spec-
trometry using an internal standard and propor-
tional correction. Fresenius J Anal Chem 1996;
356: 362 ± 70.
11 Jason LI. Improving analytical precision by
utilizing intrinsic internal standards for determin-
ing minor constituents by inductively coupled
plasma atomic emission spectrometry. J Anal At
Spectrom 1996; 11: 683 ± 7.
12 Marshall J, Rodgers G, Campbell WC. Myers --
Tracy signal compensation in inductively coupled
plasma atomic emission spectromtery with high
dissolved solids solutions. J Anal At Spectrom
1988; 3: 241 ± 4.
13 Brenner I, Zischka M, Maichin B, Knapp G. Ca
and Na interference effects in an axially viewed
ICP using low and high aerosol loadings. J Anal
At Spectrom 1998; 13: 1257 ± 64.
14 Dubuisson C, Poussel E, Todoli JL, Mermet JM.
Effect of sodium during the aerosol transport and
® ltering in inductively coupled plasma atomic
emission spectrometry. Spectrochim Acta Part B
1998; 53: 593 ± 600.
15 KovacÏic N, BudicÏ B, Hudnik V. Matrix effects
from magnesium and lithium in inductively
coupled plasma atomic emission spectrometry. J
Anal At Spectrom 1989; 4: 33 ± 7.
16 Tripkovic MR, Holclajtner-Antunovic ID. Study
of the matrix effect of easily and non-easily
ionizable elements in an inductively coupled
argon plasma. J Anal At Spectrom 1993; 8:
349 ± 57.
17 Dean JR. Methodology in trace element analysis.
In: Ando DJ, editor. Atomic absorption and
plasma spectroscopy, 2nd ed. Chichester: John
Wiley; 1997. p. 1 ± 20
18 Chappuis P, Poupon J, Rousselet F. A sequential
and simple determination of zinc, copper and
aluminium in blood samples by inductively
coupled plasma atomic emission spectrometry.
Clin Chim Acta 1992; 206: 155 ± 65.
19 Le¯ on P, Plaquet R, Rose F, Hennon G, Ledeme
N. Rapid determination of lithium in human
serum and urine, at physiological concentrations,
by inductively coupled argon plasma atomic
emission spectrometry. Anal Chim Acta 1996;
327: 301 ± 6.
20 Lim HB, Han MS, Lee KJ. Determination of trace
elements in human serum by inductively coupled
plasma atomic emission spectrometry with ¯ ow
injection. Anal Chim Acta 1996; 320: 185 ± 9.
21 McLeod CW, Worsfold PJ, Cox AG. Simulta-
neous multi-element analysis of blood serum by
¯ ow injection--inductively coupled plasma
atomic--emission spectrometry. Analyst 1984;
109: 327 ± 32.
22 Melton LA, Tracy ML, Moller G. Screening trace
elements and electrolytes in serum by inductively-
coupled plasma emission spectrometry. Clin Chem
1990; 36: 247 ± 50.
23 Mauras Y, Allain P. Automatic determination of
aluminum in biological samples by inductively
coupled plasma emission spectrometry. Anal
Chem 1985; 57: 1706 ± 9.
24 Nixon DE, Moyer TP, Johnson P, McCall JT,
Ness AB, Fjerstad WH, Wehde MB. Routine
measurement of calcium, magnesium, copper, zinc,
and iron in urine and serum by inductively
coupled plasma emission spectroscopy. Clin
Chem 1986; 32: 1660 ± 5.
25 Halls DJ. Manganese. In: Herber RFM, Stoeppler
M, editors. Trace element analysis in biological
specimens. Amsterdam: Elsevier Science B. V.;
1994. p. 385 ± 401
26 Baruthio F, Guillard O, Arnaud J, Pierre F,
Zawislak R. Determination of manganese in
biological materials by electrothermal atomic
absorption spectrometry: a review. Clin Chem
1988; 34: 227 ± 34.
27 Subramanian KS, Meranger JC. Graphite furnace
atomic absorption spectrometry with nitric acid
deproteinization for determination of manganese
in human plasma. Anal Chem 1985; 57: 2478 ± 81.
28 Slavin W. Graphite furnace AAS for biological
materials. Sci Total Environ 1988; 71: 17 ± 35.
29 BussieÁre L, Dumont J, Hubert J. Direct determin-
tation of calcium, copper, iron, magnesium,
manganese and zinc in amniotic ¯ uid samples
using inductively coupled plasma-atomic emission
spectrometry. Anal Chim Acta 1989; 224: 73 ± 81.
30 National Institute of Standards & Technology.
Certi® cate of analysis. Standard Reference Mate-
rial 1598 inorganic constituents in bovine serum.
Gaithersburg: NIST, 1990
Received: 30 May 2000
Accepted: 31 August 2000