Available online at www.sciencedirect.

com

Talanta 74 (2008) 699–702

Direct determination of iron and manganese in wine using the reference

element technique and fast sequential multi-element
flame atomic absorption spectrometry
Sérgio L.C. Ferreira a,∗ , Anderson S. Souza a , Geovani C. Brandao a , Hadla S. Ferreira a ,
Walter N.L. dos Santos b , M. Fernanda Pimentel c , Maria Goreti R. Vale d
a Universidade Federal da Bahia, Instituto de Quı́mica, Núcleo de Excelência em Quı́mica Analı́tica da Bahia,
Campus Universitário de Ondina, Salvador 40170-290, BA, Brazil
b Universidade do Estado da Bahia, Rua Silveira Martins 2555, Cabula, CEP 41.195.001, Salvador, BA, Brazil
c Universidade Federal de Pernambuco, Departamento de Engenharia Quı́mica, Recife 50740-521, PE, Brazil
d Instituto de Quı́mica, Universidade Federal do Rio Grande do Sul, Av. Bento Gonçalves 9500, Porto Alegre 91501-970, RS, Brazil

Received 1 June 2007; received in revised form 25 June 2007; accepted 27 June 2007
Available online 3 July 2007

Abstract
A procedure is proposed for the direct determination of manganese and iron in wine employing fast sequential flame atomic absorption
spectrometry and the reference element technique to correct for matrix effects. Cobalt, silver, nickel and indium have been tested as reference
elements. The results demonstrated that cobalt and indium at a concentration of 2 and 10 mg L−1 were efficient for quantification of manganese and
iron, respectively. Under these conditions, manganese and iron could be determined with quantification limits of 27 and 40 ␮g L−1 , respectively.
The proposed method was applied to the determination of manganese and iron in 16 wine samples. The content of manganese varied from 0.78 to
2.89 mg L−1 and that of iron from 0.88 to 9.22 mg L−1 . The analytical results were compared with those obtained by inductively coupled plasma
optical emission spectrometry after complete mineralization using acid digestion. The statistical comparison by a t-test (95% confidence level)
showed no significant difference between the results.
© 2007 Elsevier B.V. All rights reserved.

Keywords: Wine; Reference element technique; Iron determination; Manganese determination; Matrix effects; Fast sequential flame atomic absorption spectrometry

1. Introduction wine. They also found that manganese and iron play an impor-
tant role in chemical processes with acetaldehyde: manganese
The mineral content of wine depends on several factors favors its formation and iron catalyzes acetaldehyde combina-
including absorption of metals from the soil, contamination by tion with phenolic compounds. Benitez et al. [7] concluded that
atmospheric emissions, the type of grape, herbicidal treatment iron, copper, manganese and zinc modify the stability, color,
and the wine-making process. The metal content has a great and clarity of wines. Zinc and iron might affect the organoleptic
influence on the quality of a wine, and several studies have characteristics of a wine.
been performed involving this subject [1–5]. Cacho et al. [6] The direct determination of metals in wine employing spec-
studied the influence of iron, copper, and manganese on wine troanalytical techniques such as inductively coupled plasma
oxidation, measuring the evolution of different compounds sen- optical emission spectrometry (ICPOES), inductively coupled
sitive to this process, such as anthocyanins, tannins, total phenol plasma mass spectrometry (ICP-MS) and flame atomic absorp-
content and acetaldehyde. They concluded that the oxidation tion spectrometry (FAAS) is difficult, because wine is a complex
depends directly on the concentration of these cations in the matrix that contains ethanol, sugars and other organic com-
pounds [1]. The methods proposed for the determination of
metals in wine by these techniques therefore typically include
∗ Corresponding author. some sample preparation. This might consist of elimination of
E-mail address: [email protected] (S.L.C. Ferreira). the alcohol, acid digestion, evaporation to dryness and disso-

0039-9140/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2007.06.038
700 S.L.C. Ferreira et al. / Talanta 74 (2008) 699–702

Table 1
Wavelength and concentration used and linear regression equations obtained for the determination of manganese and iron in the absence (standard calibration
technique) and in the presence of four different elements used as RE
Reference element Wavelength (nm) RE concentration (mg L−1 ) Linear regression equation (Mn) Linear regression equation (Fe)

Cobalt 240.7 2.0 A = 0.1311CMn + 0.1462 (R = 0.9995) A = 0.0633CFe + 0.0623 (R = 0.9961)

Indium 303.9 10.0 A = 0.1610CMn + 0.0992 (R = 0.9480) A = 0.0774CFe + 0.0531 (R = 0.9994)
Nickel 232.0 2.0 A = 0.1456CMn + 0.1175 (R = 0.9952) A = 0.0759CFe + 0.0617 (R = 0.9929)
Silver 328.1 2.0 A = 0.1484CMn + 0.0959 (R = 0.9868) A = 0.0703CFe + 0.0498 (R = 0.9959)
Results using the standard calibration technique A = 0.1363CMn + 0.0128 (R = 0.9991) A = 0.0744CFe + 0.0081 (R = 0.9962)

R: correlation coefficient.

lution of the residue in nitric acid [8,9], mineralization in a Calibration solutions were prepared from 1000 mg L−1 man-
digestion block using concentrated acids [10,11], microwave- ganese and iron stock solutions (Merck) by appropriate dilution
assisted acid digestion [12–14], UV irradiation-assisted diges- with a 1% (v/v) nitric acid solution.
tion [15], dilution of the wine sample and use of the analyte The concentration of the reference element solutions was
addition technique [16] and also the reference element technique 2.0 mg L−1 for silver, nickel, cobalt and 10.0 mg L−1 for indium.
employing ICPOES or ICP-MS [17,18].
The reference element technique is used to overcome matrix 2.3. Procedure for total digestion of wine
effects that might influence the analytical signal in the quan-
tification step. It requires simultaneous or fast sequential A volume of 10.0 mL of wine was introduced into a 125-
measurement of the analyte(s) and the reference element and it mL Erlenmeyer flask, and 3.0 mL of concentrated nitric acid
has therefore been widely used in ICPOES and ICP-MS [19,20]. and 5.0 mL of 30% (v/v) hydrogen peroxide were added. The
The use of the reference element technique in FAAS has become samples were heated and evaporated to dryness on a hot plate; the
feasible with the introduction of a fast sequential (FS) multi- residue was dissolved in 0.5 mol L−1 nitric acid, transferred to a
element system for this technique [21]. It has been used by 25-mL volumetric flask and diluted to volume with 0.5 mol L−1
several authors for FS multi-element FAAS [21–24]. da Silva nitric acid solution. Manganese and iron were determined in
et al. [25] used this system and nickel as reference element for the diluted solutions by ICPOES. All samples were analyzed in
the determination of iron in hydrated ethanol with FSFAAS. triplicate.
The present paper proposes the use of the reference element
technique for correction of matrix effects in the direct determi-
2.4. Direct determination using FSFAAS and the reference
nation of manganese and iron in wine employing FSFAAS.
element technique
2. Experimental
For the direct determination 10.0 mL of wine and 2.5 mL each
2.1. Instrumentation of a 20.0 ␮g mL−1 cobalt and 100 ␮g mL−1 indium standard
solution, respectively, were introduced into a 25-mL volumetric
A Varian model SpectrAA 220 FS flame atomic absorp- flask and diluted to volume using 1.0 mol L−1 nitric acid solu-
tion spectrometer (Mulgrave, Vistoria, Australia), equipped with tion. The final concentration of cobalt and indium in this solution
a deuterium background corrector and automatic switching was 2 and 10 ␮g mL−1 , respectively.
of hollow-cathode lamps was used. The iron and manganese
hollow-cathode lamps were operated with a current of 5.0 mA 3. Results and discussion
for both elements. The analytical wavelengths at 248.3 and
279.5 nm were used with a spectral bandwidth of 0.1 nm for 3.1. Selection of the reference element for the
both metals. An air–acetylene flame was used with an acetylene determination of manganese and iron
flow rate of 2.0 L min−1 , an airflow rate of 13.5 L min−1 and a
burner height of 13.5 mm. The reference elements (RE) tested were cobalt, indium,
nickel and silver. The lines and concentrations used for quan-
2.2. Reagents
Table 2
All reagents were of analytical grade unless otherwise stated. Evaluation of the precision of the method (N = 11)
Ultrapure water was obtained from an EASYpure RF purifi-
Analyte Concentration RSD using RSD without
cation system (Barnstedt, Dubuque, IA, USA). Nitric and added (mg L−1 ) RE (%) RE (%)
hydrochloric acid were of Suprapur® quality (Merck, Darm-
stadt, Germany). Nitric and hydrochloric acid solutions were Iron (RE: In) 0.5 0.52 1.9
2.5 0.88 2.5
prepared from the concentrated acids by dilution with water.
Laboratory glassware was kept overnight in 10% (v/v) nitric
Manganese 0.2 0.91 1.4
acid solution. Before use the glassware was rinsed with deion-
(RE: Co) 1.0 1.14 1.9
ized water and dried in a dust-free environment.
 S.L.C. Ferreira et al. / Talanta 74 (2008) 699–702 701

tification were the same as proposed by Projahn et al. [21].

with RE (%)
Deviation Analytical curves were established by spiking a wine sample
with increasing concentrations of manganese and iron. All the
−3.9
−1.1
2.1
−4.2
−2.0
3.6
−2.5
−0.5
−3.8
3.2
−2.0
−6.6
−0.1
−2.8
−4.0
3.6
results found are summarized in Table 1. The analytical signals
obtained in this experiment were corrected employing the RE
technique for all the four RE tested. The corrections were done
without RE (%)

using the expression:

Deviation

(AS◦RE )
−7.8
−3.4
−7.4
−16.0
−13.6
−15.7
−21.3
−11.2
−12.7
−11.2
−13.9
−20.7
−17.3
−22.6
−14.3
−12.9
(ASC ) = (AS) × (1)
(ASRE )
where (AS) is the analytical signal obtained during the quan-
Mn ICPOES

0.03
0.07
0.05
0.04
0.11
0.11
0.03
0.10
0.03
0.03
0.04
0.05
0.11
0.06
0.06
0.05
tification of the analyte, (AS◦RE ) the average of the analytical
(mg L−1 )

±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
signals obtained for the RE in the calibration solutions and blank,
2.30
2.61
2.83
2.38
1.47
1.40
0.80
1.87
2.36
2.51
2.45
1.98
1.62
1.06
2.23
1.39
(ASRE ) the analytical signal obtained for the RE in the sample
solution and (ASC ) is the corrected analytical signal. The rela-
tive standard deviation for the analytical signals obtained for the
0.12
0.22
0.07
0.05
0.11
0.07
0.19
0.08
0.15
0.06
0.04
0.11
0.01
0.07
0.23
0.08
Mn with RE

RE was 2.0% for indium (RE used for iron) and 3.0% for cobalt
(mg L−1 )

±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
(RE used for manganese).
2.21
2.58
2.89
2.28
1.44
1.35
0.78
1.86
2.27
2.59
2.40
1.85
1.61
1.03
2.14
1.44

The slope of the analytical curve for manganese

(0.1363 ± 0.0042)CMn(mg L−1 ) obtained by the standard calibra-
tion technique expressed as confidence interval [26] is quite
Mn without RE

comparable with the slope (0.1311 ± 0.0033)CMn(mg L−1 ) of the

0.15
0.24
0.10
0.08
0.11
0.19
0.23
0.10
0.21
0.10
0.09
0.15
0.04
0.13
0.24
0.14

analytical curve obtained with the RE technique using cobalt

(mg L−1 )

±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
±

as reference element. Similarities are also observed for the

2.12
2.52
2.62
2.00
1.27
1.18
0.63
1.66
2.06
2.23
2.11
1.57
1.34
0.82
1.91
1.21

slopes of the analytical curves established for iron employing

Determination of iron and manganese in wine using FSFAAS without and with RE and using ICPOES after acid digestion

the standard calibration technique (0.0744 ± 0.0063)CFe(mg L−1 )

and the RE technique using indium as reference element
with RE (%)
Deviation

(0.0774 ± 0.0080)CFe(mg L−1 ) . These results demonstrate that

the RE technique (using cobalt as reference element for man-
−1.3
−3.1
−1.8
−2.7
−0.9
−2.9
−4.3
3.2
3.8
−3.2
2.3
4.2
1.0
−3.8
−2.1
3.5

ganese and indium for iron) is perfectly efficient for correcting

the matrix effect observed in the determination of these ele-
without RE (%)

ments in wine using FSFAAS. All the experiments involving

Also shown is the relative deviation of the FSFAAS values compared with the ICPOES results.

silver as reference element were performed in nitric acid media.

Deviation

Hydrochloric acid cannot be used because of precipitation of

−12.7
−21.1
−17.7
−30.5
−8.2
−20.2
−31.3
−31.2
−25.5
−41.0
−8.0
−27.1
−16.0
−23.8
−28.8
−12.9

silver(I) ion as the chloride.

3.2. Analytical characteristics of the method

0.28
0.08
0.13
0.18
0.12
0.22
0.22
0.16
0.03
0.14
0.16
0.09
0.10
0.08
0.19
0.17
Fe ICPOES
(mg L−1 )

±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
±

The precision expressed as relative standard deviation (RSD)

3.08
1.61
7.59
2.62
9.30
3.11
2.56
1.86
5.01
2.22
6.11
0.96
2.00
1.60
3.82
0.85

was evaluated for the quantification of manganese and iron in

solutions of 10.0 mL of wine, additions of manganese and iron,
RE, and dilution with 1.0 mol L−1 nitric acid solution to a final
0.10
0.07
0.07
0.23
0.33
0.18
0.20
0.11
0.34
0.12
0.07
0.07
0.06
0.08
0.39
0.17
Fe with RE
(mg L−1 )

volume of 25.0 mL. The RE used was cobalt for manganese

±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
±

and indium for iron. The results in Table 2 demonstrate that the
3.04
1.56
7.45
2.55
9.22
3.02
2.45
1.92
5.20
2.15
6.25
1.00
2.02
1.54
3.74
0.88

precision of the method improved considerably when the RE

technique was applied for both metals.
RE (mg L−1 )

0.10
0.11
0.29
0.39
0.63
0.39
0.25
0.12
0.53
0.15
0.23
0.12
0.08
0.23
0.48
0.26

The limits of detection (3σ/S) and quantification (10σ/S) were

Fe without

12 and 40 ␮g L−1 for iron, and 8 and 27 ␮g L−1 for manganese,

±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
±
2.69
1.27
6.25
1.82
8.54
2.48
1.76
1.28
3.73
1.31
5.62
0.70
1.68
1.22
2.72
0.74

respectively, calculated according to IUPAC recommendations,

where σ is the standard deviation of the blank (n = 10) and S is
Dry table red
Dry table red
Dry table red
Dry table red

the slope of the analytical curve [27].

Sweet white
Sweet white
White table
White table
White table
Dry white
Dry white
Sweet red
Sweet red
Sweet red
Dry red
Dry red

3.3. Application of the method

Wine

The proposed method has been applied for the determina-

Sample
Table 3

tion of manganese and iron in 16 wine samples (including white

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

wine, red wine, dry wine, table wine and sweet wine) bought
702 S.L.C. Ferreira et al. / Talanta 74 (2008) 699–702

in supermarkets in Salvador City, Brazil. The manganese con- technique might well be applied for the determination of other
tent varied from 0.78 to 2.89 mg L−1 and that of iron from 0.88 trace elements in wine after careful selection of an appropriate
to 9.22 mg L−1 . The analytical results were compared with the RE.
results obtained by the analysis of these samples after com-
plete mineralization using acid digestion and determination with Acknowledgements
ICPOES. All results are summarized in Table 3. The linear
regression method was used for comparing the results obtained The authors are grateful to Fundação de Amparo a Pesquisa
by the proposed method using the RE technique and the ICPOES do Estado da Bahia (FAPESB), Conselho Nacional de Desen-
method for both metals. The equations found expressed as con- volvimento Cientı́fico e Tecnológico (CNPq) and Coordenação
fidence intervals were: de Aperfeiçoamento de Pessoal de Nı́vel Superior (CAPES) for
providing grants and fellowships and for financial support.
Fe : [RET] = 0.9973 ± 0.0101[ICPOES]
−0.0103 ± 0.0416, r = 0.9993, References
Mn : [RET] = 0.9961 ± 0.0268[ICPOES] [1] K. Pyrzynska, Crit. Rev. Anal. Chem. 34 (2004) 69.
[2] M. Aceto, O. Abolino, M.C. Bruzzoniti, E. Mentasti, C. Sarzanin, M.
−0.0254 ± 0.0546, r = 0.9950
Malandrino, Food Addit. Contam. 19 (2002) 126.
These results demonstrate that the calculated slopes and inter- [3] A.L. Waterhouse, V.F. Laurie, Am. J. Enol. Viticult. 57 (2006) 306.
[4] P.L. Teissedre, M. Krosniak, K. Portet, F. Gasc, A.L. Waterhouse, J.J.
cepts do not differ significantly from the “ideal” values of 1 and Serrano, J.C. Cabanis, G. Cros, Food Addit. Contam. 15 (1998) 585.
0, respectively, and that there is no evidence for a systematic [5] A. Santalad, R. Burakham, S. Srijaranai, K. Grudpan, Microchem. J. 86
difference between the two methods for both metals. (2007) 209.
The t-test (95% confidence level) showed no significant dif- [6] J. Cacho, J.E. Castells, A. Esteban, B. Laguna, N. Sagrista, Am. J. Enol.
ference between these results. The accuracy of the method was Viticult. 46 (1995) 380.
[7] P. Benitez, R. Castro, C.G. Barroso, Anal. Chim. Acta 458 (2002) 197.
also evaluated by determination of the relative deviation (%), [8] V.A. Lemos, M. de La Guardia, S.L.C. Ferreira, Talanta 58 (2002) 475.
considering the results obtained with acid digestion and determi- [9] R. Lara, S. Cerutti, J.A. Salonia, R.A. Olsina, L.D. Martinez, Food Chem.
nation using ICPOES as the more probable result. This deviation Toxicol. 43 (2005) 293.
was calculated using the expression: [10] F.F. Lopez, C. Cabrera, M.L. Lorenzo, M.C. Lopez, Sci. Total Environ. 220
  (1998) 1.
Rf − Rfad [11] C. Mena, C. Cabrera, M.L. Lorenzo, M.C. Lopez, Sci. Total Environ. 181
deviation (%) = × 100 (2) (1996) 201.
Rfad
[12] E. Tatár, V.G. Mihucz, I. Virág, L. Rácz, G. Záray, Microchem. J. 85 (2007)
where Rf is the results found with and without application of the 132.
RE technique and Rfad are the results found with acid digestion [13] V.G. Mihucz, C.J. Done, E. Tatár, I. Virág, G. Záray, E.G. Baiulescu, Talanta
70 (2006) 984.
and determination by ICPOES. The deviations for manganese [14] Kh.Z. Brainina, N.Yu. Stozhko, G.M. Belysheva, O.V. Inzhevatova, L.I.
were between −3.2% and −22.6% using the standard calibration Kolyadina, C. Cremisini, M. Galletti, Anal. Chim. Acta 514 (2004) 227.
technique, and between −0.1% and −3.9% using cobalt as RE, [15] C.M. Almeida, M.T.S.D. Vasconcelos, Anal. Chim. Acta 463 (2002) 165.
respectively. For iron, the deviations varied between −8% and [16] I. Karadjova, B. Izgi, S. Gucer, Spectrochim. Acta B 57 (2002) 581.
−41% using the standard calibration technique, and between [17] M.M. Castiñeira, R. Brandt, A. von Bohlen, N. Jakubowski, Fresen. J. Anal.
Chem. 370 (2001) 553.
−4.3% and +3.8% using indium as RE. It is interesting to note [18] M. Iglesias, E. Besalú, E. Ântico, J. Agric. Food Chem. 55 (2007) 219.
that in the absence of an RE all results exhibit a negative bias, [19] A.C. Davis, C.P. Calloway, B.T. Jones, Microchem. J. 84 (2006) 31.
i.e. a systematic deviation, whereas in the presence of an RE the [20] R.M. de Souza, A.L.S. Meliande, C.L.P. da Silveira, R.Q. Aucélio,
deviations are much more randomly distributed and even exhibit Microchem. J. 82 (2006) 137.
positive values. [21] H.D. Projahn, U. Steeg, J. Sanders, E. Vanclay, Anal. Bioanal. Chem. 378
(2004) 1083.
The manganese and iron contents found for the 16 sam- [22] P. Niedzielski, Anal. Chim. Acta 551 (2005) 199.
ples are in agreement with data reported in the literature [23] F.A.C. Amorim, S.L.C. Ferreira, Talanta 65 (2005) 960.
[2]. [24] E.G.P. da Silva, A.C.N. Santos, A.C.S. Costa, D.M.N. Fortunato, N.M.
José, M.G.A. Korn, W.N.L. dos Santos, S.L.C. Ferreira, Microchem. J. 82
4. Conclusion (2006) 159.
[25] J.E. da Silva, F.A. da Silva, M.F. Pimentel, R.S. Honorato, V.L. da Silva,
M.C.B.S.M. Montenegro, A.N. Araújo, Talanta 70 (2006) 522.
The determination of metals in wine is of great interest [26] D.L. Massart, B.G.M. Vandeginste, L.M.C. Buydens, S. de Jong, P.J. Lewi,
because it allows establishing criteria for the evaluation of the J. Smeyers-Verbeke, Handbook of Chemometrics and Qualimetrics: Part
identity and authenticity of the samples. The results demonstrate A, Elsevier, Amsterdam, 1977.
that the RE technique provides reliable results with the advan- [27] IUPAC, Analytical Chemistry Division, Spectrochim. Acta B 33 (1978)
242.
tage that no further manipulation of the sample is needed. The