Determination of metals in wine with atomic spectroscopy
(¯ame-AAS, GF-AAS and ICP-AES); a review
Maurizio Acetoy*, Ornella Abollinoz,
Maria Concetta Bruzzonitiz, Edoardo Mentastiz,
Corrado Sarzaniniz and Mery Malandrinoz
yDepartment of Sciences and Advanced Technologies, University of
East Piedmont, Corso Borsalino 54, I-15100 Alessandria, Italy
zDepartment of Analytical Chemistry, University of Turin, Via P.
Giuria 5, I-10125 Turin, Italy
(Received 7 November 2000; revised 27 April 2001; accepted
12 June 2001)
Metals in wine occur at the mg l
though not directly related to the taste of the ®nal
product, their content should be determined because
excess is undesirable, and in some cases prohibited, due
to potential toxicity. Lead content in wine, for example,
is restricted in several states by legislation to guarantee
consumer health protection. Of several methods for
metal determination, techniques of atomic spectroscopy
are the most sensitive and rapid. Most of the elements
present in wine can be determined with these techniques,
at concentrations ranging from the mg l
level. Here, inductively coupled plasma-atomic emission
spectrometry (ICP-AES), ¯ame atomic absorption
spectrometry (¯ame-AAS) and graphite furnace-atom-
ic absorption spectrometry (GF-AAS) are compared
for their characteristics as employed in metal determi-
nation in wine.
1level or less and,
1to the ·g l
Keywords : wine, metals, ICP-AES, ¯ame-AAS, GF-
AAS, atomic spectroscopy
The determination of metals in wine is routinely
carried out in most oenological laboratories because
some elements must be kept under control according
to the law. The lead content, for example, should
Parliament 1987). Several countries have rules re-
stricting metal content in wines, which must be ful-
®lled by producers to gain right to export to these
markets. Some examples are reported in table 1.
1, according to Italian legislation (Italian
Atomic spectroscopy is the most commonly used
technique for metal determination in wines. Several
methods are proposed in literature that use ICP-AES
(Eschnauer et al. 1989, Thiel and Danzer 1997,
Fournier et al. 1998), ¯ame-AAS (Caputi and Ueda
1967, Gonzalez et al. 1988) or GF-AAS (Soares et al.
1995, Bruhn et al. 1996, Chmilenko and Baklanova
1997). From an analytical point of view, wine is a
fairly complex matrix owing to the content of its
strongly in¯uences transport properties of the sample
toward atomization devices due to changes in density
and surface tension with respect to aqueous standard
solutions. Various types of interferences are to be
expected in metal determination, but the knowledge
of the concentration of few major components allows
the adoption of simple quantitation procedures.
Particularly important and easy to apply is the matrix
matching method, which o?ers the possibility of
obtaining a simple external calibration by preparing
standard solutions as similar as possible to the
samples. This is possible in wine because although
several hundreds of di?erent compounds are present,
just few are capable of a?ecting the analyte response
by causing interference.
In fact, atomic spectrometry techniques are generally
little prone to interference caused by organic com-
pounds due to the high temperatures involved in the
atomization steps. Moreover, being fast and sensitive,
they are without doubt the most suitable techniques
in metal determination in wines. The present work
deals with ICP-AES, ¯ame-AAS and GF-AAS,
comparing their performances as employed in wine
Food Additives and Contaminants, 2002, Vol. 19, No. 2, 126±133
*To whom correspondence should be addressed. e-mail: aceto@ch.
Food Additives and Contaminants ISSN 0265±203X print/ISSN 1464±5122 online # 2002 Taylor & Francis Ltd
DOI: 10.1080/0265203011007133 6
Materials and methods
Reagents and materials
All reagents were of analytical grade or better. High-
purity water from a Milli-Q apparatus was used to
prepare standard solutions.
ICP-AES determinations were performed with a
Liberty-100 model Varian spectrometer. The in-
strument was equipped with a CETAC U-5000A‡
ultrasonic nebulizer for determination of elements
requiring higher sensitivity. A Perkin-Elmer 5100
spectrometer with Zeeman e?ect background correc-
tion was used for GF-AAS determinations. A Varian
SpectrAA 10 model was used for ¯ame-AAS deter-
minations. An air-C2H2 ¯ame was utilized for all
elements. Absorption and emission lines utilized are
reported in table 2.
All determinations were carried out on untreated
wine samples; only nitric acid was added to lower
pH (1ml concentrated acid to 100ml sample, the
resulting pH being1.5).
Results and discussion
Acidi®cation yields the following advantages:
. fermentation processes are prevented;
. precipitation of metals is minimized;
. metal ions are released from ligands present in wine
allowing atomization of analytes from free ions.
This is important mainly in atomic absorption tech-
. sorption of metals onto the walls of the containers
For development of methods, a common scheme was
followed. Three main features of wine studied were
those assumed responsible for interference in the
. Potassium content: generally high in wines (300±
1500 mg l
processes by acting as ionization bu?er, a feature
especially important for elements with low ioniza-
tion potentials (alkaline and alkali-earth metals).
This e?ect is particularly important for ¯ame ato-
. Phosphate content: present at 100±1000mg l
wines, and can cause chemical interference by form-
ing refractory compounds (typically calcium phos-
phate). Such interference is expected to be more
relevant in ¯ame-AAS.
. Ethanol: usually 5±15% in wines, and strongly
in¯uences the viscosity of the samples and their
1), and can strongly in¯uence ionization
127Determination of metals in wine with atomic spectroscopy
(concentrations expressed in mg l
Lawlimits for metalcontentin wines
CountryAl AsCdCuNaPb TiZn
yO?ce International de la Vigne et du Vin.
atomic spectrometric techniques.
Lines used for determination of elements with
transport properties towards atomization devices of
The responses of analytes were recorded in standard
solutions containing di?erent contents of these three
compounds. Potassium content was varied between
0 and 1000mg l
between 0 and 1000mg l
between 0 and 20%. The application of each tech-
nique is considered separately below.
1as K‡ion, phosphate content
1as H3PO4, and ethanol
Determination of elements by ¯ame-AAS
Flame-AAS is largely employed in wine analysis
mainly due to the low cost of instrumentation that
makes the technique easily accessible to most oeno-
logical laboratories. Considering the compromise be-
tween the cost and sensitivity needed, ¯ame-AAS can
be considered the technique of choice for alkaline and
alkali-earth metal determination in wine. It is also
well suited for Cu, Fe, Mn and Zn determination,
with respect to concentration ranges of these metals in
wines (table 3). It is not suitable for toxic or undesir-
able elements like As, Cd, Cr, Hg and Pb, except for
highly contaminated samples or in an application
with preconcentration procedures. Finally, it should
be remembered that ¯ame-AAS is the o?cial method
of analysis for determination of Na, K, Mg, Ca, Fe,
Ag and Zn in wine according to EU regulation
(European Union 1990). Precision of analysis of wine
samples is usually very good, being on average >1%,
for all elements considered at mg l
E?ect of potassium content. The high concentration
of potassium in wines acts as a natural ionization
bu?er. This is critical for alkaline metal determina-
tion because it causes a decrease in their ionization
and a corresponding increase in the signal of the free
atoms. Variations of Na and Rb absorbances with
potassium content in standard solution are shown in
®gure 1. It can be seen that in the range 700±
1000 mg l
response is nearly constant. This suggests that the
addition of1000mg l
solutions should be su?cient to allow a simple
external calibration instead of the standard addition
method. In fact, determinations were carried out
with the two methods di?ering by 45% (table 4).
Potassium content has slight e?ects on alkali-earth
metals and aluminium, for whose determination an
acetylene/nitrous oxide ¯ame should be used. We
reported no e?ect of potassium on transition metal
1potassium (suitable for most wines), the
potassium to standard
E?ect of phosphate content. Chemical interferences
are particularly important for Ba, Ca, Mg, Sr and
Al, determinations. The high content of PO3
wines (table 3) suggests that these elements may be
128 M. Aceto et al.
Concentration ranges of inorganic species in
P (as PO4
Adapted from Eschnauer et al. (1989) and Greenough et al. (1997).
mg/l of K+
0 200400 600800 1000
Figure 1. E?ect of potassium concentration on Na and
Rb absorbance with Flame-AAS. = Na‡;= Rb‡.
converted to their refractory phosphates in the ¯ame
environment; absorbance is expected to decrease in
this case. The e?ect of PO3
determination of calcium can be seen in ®gure 2. The
usual procedure suggests addition of a releasing
agent (normally lanthanum chloride) to samples.
External calibration can still be performed by
provided the phosphate
previously determined (an average of 500mg l
suitable for most wines). No e?ect has been recorded
on the determination of transition metals, while
increasing concentrations of phosphate cause a slight
decrease of absorbance for alkali metals.
concentration on the
E?ect of ethanol percentage. Ethanol content greatly
in¯uences the transport properties of samples. This
can be easily overcome by addition of a ®xed
di?erent than in alcohol-free solutions, is fairly
constant in the range 5±15%. Adding 12% ethanol
to standard solutions allows an accurate determina-
tion of the alkaline and alkali-earth elements.
Determination of transition metals is not a?ected by
the presence of ethanol in samples in the range
Other interferences. No spectral interferences are
reported in literature concerning metal determina-
tion in wine with ¯ame-AAS.
From these indications, it appears clear that external
calibration can be performed for quantitative deter-
minations. Standard solutions of aluminium, alkali
and alkali-earth analytes should contain 1000mg l
potassium, 500mg l
nol, while standard solutions of transition metal
analytes do not need addition of interferents.
1phosphate and have 12% etha-
Determination of elements by GF-AAS
Graphite furnace AAS is often employed for trace
metal determination in wines, though it is a technique
unavailable to most oenological laboratories. It is
well suited for the determination of toxic elements
(table 3). Determination of untreated samples can be
di?cult to perform, due mainly to ethanol which
a?ects the precision on sample delivery. Usually,
dilution with an acidic solution is needed (Soares
et al. 1995). Injection into a hot furnace gives satis-
factory results. Careful conditions must be selected
through appropriate furnace heating cycles. GF-AAS
is the o?cial method of analysis for determination of
Cd and Pb in wine according to EU regulations
(European Union 1990). Precision of analysis in wine
samples is good and >2% for all elements consid-
ered, at the mg l
lyses of undiluted samples may result in a worse
1concentration level, although ana-
E?ect of potassium content. No e?ect is reported.
E?ect of phosphate content. Phosphate is commonly
used as matrix modi®er in the determination of
several elements, between Cd and Pb; the amount of
phosphate added as modi®er largely exceeds the
phosphate present in wines.
E?ect of ethanol percentage. Analysis of untreated
samples is di?cult to perform, especially on red wine
samples, and reproducibility is strongly in¯uenced
by the presence of ethanol, which a?ects sample
delivery in the furnace. Dilution with an acidic
solution is preferable when the concentration of the
129Determination of metals in wine with atomic spectroscopy
comparison between standard additions and external
DeterminationofNaand Rbin wine:
DeterminationNa (mg l
1)Rb (mg l
Standard addition curve
mg/l of PO4
0 200 400600 8001000 1200
Figure 2. E?ect of phosphate concentration on Ca absor-
bance with Flame-AAS.
analytes is high enough, otherwise a minimum
dilution of 1:1 and the addition of a surfactant such
as Triton X-100 to adjust for sample viscosity is
suggested (Soares et al. 1995). A furnace-heating
cycle must be set up with extra steps to evaporate
solvent and destroy organic matter e?ciently.
Other interferences. No spectral interferences are
reported in the literature about metal determination
in wine with GF-AAS.
background correction greatly improves the accu-
racy of determinations.
Determination of metals in wine with GF-AAS is not
very convenient to perform because the technique is
intrinsically time consuming. However, external cali-
bration can be used, provided a su?cient dilution
ratio is taken. Stabilized temperature platform fur-
nace (STPF) conditions should be ful®lled, chie¯y
with the use of matrix modi®ers. A graphite furnace
heating cycle for Cd and Pb determination is reported
in table 5.
Determination of elements by ICP-AES
Plasma spectrometers are not very popular within
oenological laboratories. Both the cost of purchase
and maintenance are well beyond their ®nances.
However, routine use of ICP-AES could prove, in
the long-term, to be more advantageous than ex-
pected. The wide linear dynamic range of this tech-
nique (up to six orders of magnitude) allows the
concomitant determination of several analytes from
major elements down to trace elements. A high-
temperature plasma ( 10000 K) yields optimal con-
ditions for the e?cient breakage of organic matter
present in wine. Interferences are usually less import-
ant than with other atomic techniques. The precision
of the analysis in wine samples is usually good for
most elements and >2% at themg l
level. Determination of some trace metals (Cr, Li, Ni,
Pb, Ti) can be troublesome and generally less precise.
Precision and accuracy can be improved if ultrasonic
nebulization is employed.
emission intensity of any of the analytes considered.
temperature, formation of refractory phosphates
does not limit the emission response of analytes.
of phosphatecontent. Owingto plasma
E?ect of ethanol percentage. This is particularly
important for techniques in which the sample is
in¯uences emission responses in a rather complex
. by changing excitation conditions, due to a change
in thermal and electrical properties of the plasma.
Good results can be obtained raising plasma power
from 1.0 to 1.3 kW (®gures 3 and 4). In fact, an
increase of ethanol content at 1.0 kW causes a
decrease in the signal for all analytes, the e?ect
being more relevant for zinc. When plasma power
is raised to 1.3 kW, the signals of all elements except
zinc initially increase with ethanol concentration,
130M. Aceto et al.
determination of Cd and Pb.
Graphite furnace heating cycle for GF-AAS
Temperature (°C)Ramp (s)Hold (s)
L’vov platform; sample volume: 20 ml.
05 10 1520 25
Relative intensity (%)
Figure 3. E?ect of ethanol percentage on atomic emission
intensity (plasma power: 1.0kW).
Fe; 5 = Mn; & = Ni; & = V; ^= Zn.
= Cr;= Cu; ! =
and then remain approximately stable in the range
5±20%; the emission intensity of zinc initially
decreases, but its value remains stable in the same
range as for the other elements; and
. by changing sample viscosity, density and surface
tension; this can be overcome by addition of etha-
nol to standard solutions. Taking into account the
trend shown in ®gure 4, a suitable concentration of
added ethanol can be 12%.
The adoption of both of devices allows easy external
calibration for quantitative determinations by ICP-
Other interferences. Unlike the two previous tech-
niques, some spectral interferences are reported to
occur in metal determination in wine with ICP-AES.
In most cases, a positive interference is involved.
Two cases are to be cited.
. Cr determinations using the main emission lines
(267.716 and 205.552nm) su?er from direct overlap
of Fe lines. The sensitivity ratio is 2000:1 (Cr:Fe).
Considering Cr and Fe average concentrations in
wines (table 3), in some cases this interference can
be di?cult to compensate for. Usually, manage-
ment software of instruments allows an interfering
element correction procedure, provided Fe content
is constant in the samples.
. V determination at the 309.311nm line (the prin-
cipal emission line) su?ers from structured back-
determination in wine is not reliable due to a
heavy interference by unknown species centred on
this line. Accurate determinations can be obtained
by using the 292.402nm line, with slightly poorer
sensitivity. In this case, owing to wing overlap by
Fe emission lines, a higher grating order is to be
preferred for determination in wines containing a
high Fe content.
In brief, quantitative determination with ICP-AES
can be performed by external standard calibration.
Standard solutions of all analytes (regardless being
alkaline, alkali-earth or transition metals elements)
should contain at least ethanol up to 12%; neither
potassium nor phosphate addition is needed. These
conditions hold for metalloids and non-metals deter-
mination too (Al, B, P and Si).
Comparison between the methods
Many of the elements considered can be determined
in wine by all three techniques. This holds for alkaline
and alkali-earth metals, Al and some transition me-
tals (Cu, Fe, Mn, Zn). Considering sensitivity, accu-
racy and precision, the method of choice for these
elements is ¯ame-AAS, which yields better perform-
ance for concentrations in themg l
yields similar performance with a slightly lower pre-
cision and use of GF-AAS would require dilution of
the samples to be used. Other metals (Cr, Li, Ni, Pb,
Ti, V) can be determined by both ICP-AES and GF-
AAS; in this case, the higher sensitivity of GF-AAS is
preferred to obtain accurate results. Apart from
method of choice whenever many elements must be
determined in sequence, because of its speed and
Considering the performance o?ered by the three
techniques, it appears that ICP-AES is preferable
for metal determination in wine, for its fastness and
simplicity of analysis. Quantitative determinations
can be performed after easily overcoming interfer-
ences. Flame-AAS guarantees similar conditions of
analysis, but it lacks sensitivity and is more prone to
matrix e?ects. GF-AAS has sensitivity high enough
for determination of all analytes of interest in the
oenological ®eld, but it su?ers from moderate inter-
131Determination of metals in wine with atomic spectroscopy
Relative intensity (%)
Figure 4. E?ect of ethanol percentage on atomic emission
intensity (plasma power: 1.3 Kw).
Fe; 5 = Mn; & = Ni; & = V; ^= Zn.
= Cr; = Cu; ! =
ferences and slowness of analysis. Suggested con-
ditions of analysis with methods described are sum-
marized in table 6 and the proper addition of
interferents to standard solutions is indicated for each
technique. Table 7 summarizes suggested technique
for each analyte.
Financial support from the Ministero dell’Universita Á
e della Ricerca Scienti®ca e Tecnologica (MURST,
Rome) and from the Italian National Research
Council (CNR, Rome) is gratefully acknowledged.
Bruhn, C. G., Ambiado, F. E., Cid, H. J., Woerner, R., Tapia, J.,
and Garcia,R.,1996, Determination of heavy metals in waters
and drinks by electrothermal atomic-absorption spectrometry
with a tungsten coil atomizer. Quim. Anal. (Barcelona), 15,
Caputi, A., Jr, and Ueda, M., 1967, The determination of copper
and iron in wine by atomic absorption spectrophotometry.
American Journal of Enology and Viticology, 18, 66±70.
132 M. Aceto et al.
Table 6. Summary of suggested conditions of analysis.
Analytes Flame-AASGF-AAS ICP-AES
(Na, K, Rb)
(Mg, Ca, Sr, Ba)
add to standards
K‡1000 mg l
500 mg l
sample dilution with HNO3,
add to standards ethanol 12%
(Mn, Fe, Cu, Zn)
no addition neededsample dilution with HNO3,
add to standards ethanol 12%
(B, P, Si)
b.d.l.* sample dilution with HNO3,
add to standards ethanol 12%
(Li, Ti, V, Cr, Ni, Pb)
b.d.l. sample dilution with HNO3,
add to standards ethanol 12%
(Ag, As, Cd, Co, Hg, Mo, Se, Sn, )
b.d.l.sample dilution with HNO3,
*Below detection limits of the technique.
Table 7.Suggested technique of analysis for each analyte.
AnalyteFlame-AAS GF-AAS ICP-AES
Alkaline metals (Na, K, Rb)suitablesuitablesuitable
Alkali-earth metals (Mg, Ca, Sr, Ba)suitable suitablesuitable
Alsuitable suitable suitable
Transition metals (Mn, Fe, Cu, Zn)suitablesuitablesuitable
Metalloids and non metals (B, P, Si)not enough sensitivity suitablesuitable
Trace elements (Li, Cr)not enough sensitivity suitablesuitable
Trace or ultratrace metals (Ti, Ni, Pb, V) not enough sensitivitysuitablesuitable with ultrasonic nebulization
Ultratrace elements (Ag, As, Cd, Co, Hg,
Mo, Se, Sn)
not enough sensitivitysuitable not enough sensitivity
Chmilenko, F. A., and Baklanova, L. V., 1997, Atomic absorption Download full-text
determination of standardized metal impurities in wines using
ultrasound. Zh. Anal. Khim., 52, 1093±1098.
Eschnauer, H., Jakob, L., Meierer, H., and Neeb, R., 1989, Use
and limitations of ICP-OES in wine analysis. Mikrochim. Acta,
European Union, 1990, Commission Regulation No. 2676/90 of 17
September 1990. O?cial Journal, L272, 1±192.
Fournier, J. B., Hirsch, O., and Martin, G. J., 1998, Interet de
l’analyse elementaire des produits viticoles: dosage de vingt-
cinq elements par spectrometrie d’emission atomique dans un
plasma a couplage induit. Analusis, 26, 28±32.
Gonzalez, M. J., Martinez-Para, M. C., and Aguilar, M. V.,
1988, Determination of iron, copper, zinc, manganese and lead
in D. O. C. Mentrida wines. Z. Lebensm. Unters. Forsch., 187,
Greenough, J. D., Longerich, H. P., and Jackson, S. E., 1997,
Element ®ngerprinting of Okanagan Valley wines using ICP-
MS: relationships between wine composition, vineyard and
wine colour. Australian Journal of Grape and Wine Research,
Italian Parliament, 1987, Italian O?cial Gazette, o.s., no. 13 (17
Soares, M. E., Bastos, M. L., and Ferreira, M. A., 1995,
Quanti®cation of Ag, Co, Si and Zn in Port wine by AAS.
At. Spectrosc., 16, 256±260.
Thiel, G., and Danzer, K., 1997, Direct analysis of mineral com-
ponents in wine by ICP-OES. Fresenius Journal of Analytical
Chemistry, 357, 553±557.
133Determination of metals in wine with atomic spectroscopy