Talanta 72 (2007) 10731076

Simultaneous determination of copper and iron in automotive gasoline by X-ray uorescence after pre-concentration on cellulose paper
Leonardo S.G. Teixeira a, , Rodrigo B.S. Rocha a , Eledir V. Sobrinho a , Paulo R.B. Guimar es a , Luiz A.M. Pontes a , Josanaide S.R. Teixeira b a
a

Departamento de Engenharia e Arquitetura, Universidade SalvadorUNIFACS, Av. Cardeal da Silva 132, 40.220-141 Salvador-BA, Brazil b Centro Federal de Educa ao Tecnol gica da Bahia (CEFET-BA), Rua Emdio dos Santos S/N, 40301-015 Salvador-BA, Brazil c o Received 23 September 2006; received in revised form 28 December 2006; accepted 29 December 2006 Available online 9 January 2007

Abstract This paper proposes an alternative analytical method using energy dispersive X-ray uorescence (EDXRF) to determine Fe and Cu in gasoline samples. In the proposed procedure, samples were distilled and the distillation residues were spotted on cellulose paper disk to form a uniform thin lm and to produce a homogeneous and reproducible interface to the XRF instrument. The disks were dried at 60 C for 20 min and copper and iron were determined directly in the solid phase at 6.40 and 8.04 keV, respectively. The calibration curves showed linear response in the 20800 g L1 concentration range of each metal. The precisions (repeatability) calculated from 15 consecutive measurements and dened as the coefcient of variation of solutions containing 100 g L1 of Fe and Cu were 7.8 and 8.1%, respectively. The limits of detection (LOD), dened as the analyte concentration that gives a response equivalent to three times the standard deviation of the blank (n = 10), were found to be 10 and 15 g L1 for Fe and Cu, respectively. The proposed method was applied to copper and iron determination in gasoline samples collected from different gas stations. 2007 Elsevier B.V. All rights reserved.
Keywords: Gasoline; EDXRF; Metals; Iron; Copper

1. Introduction Automotive gasoline is a complex mixture of olenic, parafnic, naphthenic and aromatic hydrocarbons, with carbon content in the C4 C12 range and boiling points between 30 and 220 C, approximately, plus small amounts of sulphur, oxygen, and, to a lesser degree, nitrogen compounds [1]. The combination of these hydrocarbons with the oxygenate compounds present determines the physicochemical properties of the fuel and has great inuence on engine performance. These combination effects are even greater for the gasoline used as automotive fuel in Brazil, that contains between 20 and 25% (v/v) of ethanol, being referred to as gasohol [2]. Some classes of hydrocarbons, particularly olens and diolens, are able to slowly react, at ambient temperatures, with the oxygen in the air. The oxidation of these hydrocarbons is responsible for the formation of an insoluble solid, commonly

Corresponding author. Fax: +55 71 32032712. E-mail address: leonardoteixeira@unifacs.br (L.S.G. Teixeira).

called deposits or gum, which sticks to the metal surfaces of internal combustion engines. Accumulation of these products can cause engine wear and can have adverse effects on engine efciency, performance, and durability [3]. Metals, such as iron and copper, contribute to accelerate the formation of gum by the formation of peroxy radicals. The presence of metallic species in automotive fuels is generally not desirable and can be introduced into the fuel by corrosion of equipment during fuel processing and storage [3]. In addition, fuel combustion may also contribute signicantly to the emission of metals harmful to the environment [4,5]. Thus, the accurate determination of copper and iron in gasoline samples is of great importance. Direct analysis of gasoline is difcult by most analytical techniques because of its volatility, low viscosity, corrosivity and immiscibility with water. Additionally, metallic elements are usually present only in very low concentrations in gasoline, requiring very sensitive techniques or pre-concentration for their determination. Inductively coupled plasma (ICP) is a sensitive multi-element technique, but the introduction of organic solvents into the plasma requires special care, as the organic load may

0039-9140/$ see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2006.12.042

1074

L.S.G. Teixeira et al. / Talanta 72 (2007) 10731076

destabilize or extinguish the plasma [6,7]. Atomic absorption spectrometry (AAS) can be used for the determination of metals in petroleum derivatives, but the direct introduction of the sample may cause ames to become excessively fuel-rich and unstable [8]. Moreover, the high organic content of the sample may produce high noise levels, thus reducing detectability by AAS. Chromogenic reagents are used for metal determination in gasoline samples by spectrophotometry, however extraction and previous digestion procedures are necessary [9]. Generally, these methods are time consuming and a preliminary step is required, e.g., dry or wet ashing for decomposition of organic material in order to obtain aqueous solutions [10,11]. Energy dispersive X-ray uorescence (EDXRF) is an instrumental technique with sufcient spectral resolution for simultaneous determination of metals. Furthermore, analytical methods involving EDXRF, do not usually require the sample to be subjected to any drastic or time-consuming treatment, such as concentrated acids, delayed heating or dilution in organic solvents. However, the application of a pre-concentration procedure is an essential step in trace metal determination by this technique because its sensitivity is not enough for trace analysis. In this work, a practical procedure for determination of copper and iron in automotive gasoline samples by EDXRF is proposed. The pre-concentration methodology involves sample distillation followed by dropping and drying of its residue on cellulose paper lter, which demonstrated to be a convenient substrate for EDXRF measurements. 2. Experimental 2.1. Apparatus A Shimadzu EDX 800 (Kyoto, Japan) dispersive X-ray uorescence spectrometer equipped with a rhodium tube and Si(Li) semiconductor detector was used for the analysis. Instrumental conditions were as follows: irradiation time, 100 s; tube voltage 40 kV; tube current, 30 mA; atmosphere, air. Analytical lines were iron 6.40 keV (K ) and copper 8.04 keV (K ). Standard solutions and sample volumes were taken with a 1000 L adjustable air-displacement precision pipette, Pipetman Gilson model (Villiers-le-Bel, France). Distillations have been carried out in an automatic distillation unit model ISL AD 865G (Verson, France), in accordance with international standard methods (ASTM D-86) [12]. 2.2. Reagents and samples All reagents were of analytical grade. Methanol (min. 99.8%) was used to prepare the metal solutions. Solutions of each metal (1000 g mL1 ) were prepared by accurately weighing copper and iron nitrates (Merck, Darmstadt, Germany) followed by dilution with methanol. These solutions were used to prepare the calibration curve. A 10 g mL1 metallo-organic multi-element standard was supplied by AccuStandard Inc. (New Haven, USA). Propan-1ol (Merck) was used for the dilutions of the metallo-organic

standard solution. These solutions were used to prepare the calibration curve and to spike gasoline samples. Whatman 42 lter papers (Maidstone, England) were cut into small disks (140 mm diameter) and used for the preconcentration procedure. Commercial gasoline samples, containing 2025% (v/v) ethanol, were collected at random from different gas stations in the city of Salvador-BA, Brazil. 2.3. General procedure The proposed procedure consisted of the distillation of 100 ml gasoline samples in accordance with ASTM D 86, followed by measurement of the volume of the residue present in the distillation ask. This has then been expressed as a percentage of the charged sample volume for calculation of the pre-concentration factor. Filter disks were adapted into the EDXRF cell and a small amount (500 L) of the distillation residue was dropped on a cellulose paper disk to form a uniform thin lm and to produce a homogeneous and reproducible interface to the XRF instrument. The disks were dried at 60 C for 20 min and covered with Mylar lm. Copper and iron were determined directly in the solid phase at 6.40 and 8.04 keV, respectively. All solutions for the analytical calibration curves and for analyte addition, using inorganic and organic standards, were prepared following the already describe procedure (Section 2.2). Calibration curve were prepared by dilution of the metalloorganic or inorganic standard with puried gasoline to cover the 20800 g L1 range. The analyte concentration in the blank was determined by the analyte addition technique and the results obtained were taken into consideration. The standard solutions were directly dropped and adsorbed on the lter paper disks. 3. Results and discussion Direct analysis of gasoline samples for determination of iron and copper by EDXRF was not possible because the technique sensitivity was not enough to quantify these elements. Then, the determination of Fe and Cu in gasoline samples by EDXRF was rst attempted through distillation of the samples followed by direct analysis of its residue by EDXRF. However, although heavy hydrocarbons and inorganic compounds were identied in this residue, the sensitivity of the technique was not enough to quantify the amounts of iron and copper present. Therefore, the application of the pre-concentration procedure proposed in this work has shown to be an essential step in the analysis. A convenient substrate for EDXRF measurements was only obtained when the procedure of adsorption, by dropping and drying, of the distillation residue on cellulose lter was applied. This showed to be a suitable support, since the residue was evenly adsorbed and the dissolved metals were homogeneously distributed throughout the paper surface, thus forming a uniform thin lm. After drying of the sample on lter paper, a direct analysis by EDXRF was undertaken. In order to minimize matrix effects, a metallo-organic multielement standard material was used for calibration. However,

L.S.G. Teixeira et al. / Talanta 72 (2007) 10731076 Table 1 Calibration equations (I = aC + b)a and correlation coefcients for determination of Cu and Fe in gasoline samples using organic or inorganic standards Calibration Analytical curve using organic standard Analytical curve using inorganic standard
a

1075

Analyte Cu Fe Cu Fe

Equation I = 6.1 104 C + 0.0141 I = 7.2 104 C + 0.0132 I = 5.8 104 C + 0.0122 I = 7.1 104 C + 0.0115

Correlation coefcient 0.9981 0.9984 0.9986 0.9981

I: intensity (cps); a: slope; C: concentration ( g L1 ); b: intercept. Table 2 Determination of copper and iron ( g L1 ) in gasoline samples (n = 3) through spike recovery tests with the addition of 50 g L1 of each metala Sample 1 Sample without copper addition Sample with copper addition Copper recovery (%) Sample without iron addition Sample with iron addition Iron recovery (%)
a

an analytical curve was also obtained using inorganic salts. The resulting equations and their respective correlation coefcients are shown in Table 1. As can be seen, the slopes of the calibration curves obtained using either inorganic or organic standards are very similar. This means that the copper and iron present in gasoline samples can be determined through the calibration technique using either inorganic or organic standards. Simple calibration was used to collect sample data and the calibration curves for both iron and copper using inorganic standards showed a linear response. An important advantage of the proposed method is its potential for sensitivity improvement by increasing the sample volume from which the analyte is concentrated on the cellulose paper. This is an important aspect since it can provide different alternatives for varying the sensitivity of the procedure as a function of the analyte concentration in the sample. This effect can be assessed by measuring the signal intensity on the paper disk with different volumes of a solution containing the same concentration of copper and iron. This has been evaluated by varying the volume of the standard solutions adsorbed on the lter paper disks between 100 and 500 L, following the general procedure described in Section 2.3. As a result, linear relations were observed between the intensity signals and the sample volume. These can be described by the following equations: I = 1.1 10
3

Sample 2 188 7 235 9 94 302 6 351 7 98

Sample 3 111 4 164 5 106 264 8 317 7 106

131 5 183 8 104 252 6 299 8 94

At 95% condence level.

Table 3 Determination of copper and iron ( g L1 ) in four commercial gasoline samples (n = 3, at 95% condence level) Sample Metal Copper Distributor 1 Distributor 2 Distributor 3 Distributor 4 98 2 159 7 188 6 130 4 Iron 446 7 390 5 104 5 410 4

V + 0.0101,

(r = 0.996) for iron (r = 0.995) for copper

I = 0.95 103 V + 0.0162

where I is the intensity signal (cps) and V is the sample volume ( L) for a solution containing 800 g L1 of each metal, copper and iron. It has also been observed that increasing the sample volume resulted in an increase in the time required for the sample to dry. A 500 L sample volume, dried at 60 C for 20 min, was enough to obtain an appropriate support and allow a much greater sensitivity from the uorescence measurements. Sample volumes greater than 500 L caused uid leakage. The precisions (repeatability) calculated from 15 consecutive measurements and dened as the coefcient of variation of solutions containing 100 g L1 of each metal, Fe and Cu, were 7.8 and 8.1%, respectively. The limits of detection (LOD), dened as the analyte concentration that gives a response equivalent to three times the standard deviation of the blank (n = 10), were found to be 10 and 15 g L1 for Fe and Cu, respectively. The proposed method has been applied to the determination of copper and iron in gasoline samples from different fuel distributors. Initially, the method was veried through spike recovery tests, by adding 50 g L1 of each metal (metallo-organic stan-

dard) to three different fuel samples. The results were obtained as the average of three replicates of each fuel sample and are shown in Table 2. As can be seen, the method has good accuracy and the recoveries were between 94 and 106%. The proposed method was then applied to copper and iron determination in four more gasoline samples collected from different gas stations in Salvador-BA, Brazil. The results, obtained as the average of three replicates of each sample, are shown in Table 3. 4. Conclusion The developed procedure provides a sensitive and simple approach for the determination of copper and iron in automotive gasoline samples by EDXRF, after application of the pre-concentration procedure involving distillation and the dropping and drying of the distillation residue on cellulose lter. The method is simple, fast and does not require the sample to be subjected to any drastic or time-consuming pretreatment, such as concentrated acids heating. Acknowledgments The authors would like to acknowledge the nancial support received from Conselho Nacional de Desenvolvimento

1076

L.S.G. Teixeira et al. / Talanta 72 (2007) 10731076 [5] D.S.S. dos Santos, A.P. Teixeira, M.G.A. Korn, L.S.G. Teixeira, Spectrochim. Acta Part B 61 (2006) 592. [6] J.K. Vilhunen, A. von Bohlen, M. Schmeling, R. Klockenk mper, D. a Klockow, Spectrochim. Acta Part B 57 (2002) 1991. [7] W.N.L. dos Santos, F.S. Dias, M.V. Reboucas, M.G. Pereira, V.A. Lemos, L.S.G. Teixeira, J. Anal. At. Spectrom. 21 (2006) 1327. [8] P.S. Roldana, I.L. Alc ntara, C.C.F. Padilha, P.M. Padilha, Fuel 84 (2005) a 305. [9] Q. Wei, B. Du, Talanta 45 (1998) 957. [10] Z. Yang, X. Hou, B.T. Jones, Talanta 59 (2003) 673. [11] R.Q. Aucelio, A.J. Curtius, J. Anal. At. Spectrom. 17 (2002) 242. [12] ASTM-D86Standard Test Method for Distillation of Petroleum Products at Atmospheric Pressure, American Society for Testing and Materials, Washington, DC.

Cientco e Tecnol gico (CNPq), Financiadora de Estudos e o Projetos (FINEP/CTPetro), Petrobras and Fundacao de Amparo ` a Pesquisa do Estado da Bahia (FAPESB). References
[1] F.S. de Oliveira, L.S.G. Teixeira, M.C.U. Ara jo, M. Korn, Fuel 83 (2004) u 917. [2] L.S.G. Teixeira, E.S. Le o, A.F. Dantas, H.L.C. Pinheiro, A.C.S. Costa, a J.B. de Andrade, Talanta 64 (2004) 711. [3] L.S.G. Teixeira, J.C. Souza, H.C. dos Santos, L.A.M. Pontes, P.R.B. Guimar es, E.V. Sobrinho, R.F. Vianna, Fuel Process. Technol. 88 (2007) a 73. [4] L.S.G. Teixeira, M.A. Bezerra, V.A. Lemos, H.C. dos Santos, D.S. de Jesus, A.C.S. Costa, Sep. Sci. Tech. 40 (2005) 2555.