Internal standardization in graphite furnace atomic absorption spectrometry: Comparative use of As and Ge to minimize matrix effects on Se determination in milk B Adriana Paiva de Oliveira a , Jose´ Anchieta Gomes Neto a, T , Joaquim A. No´ brega b , Pedro Vitoriano Oliveira c a Departamento de Quı ´mica Analı ´tica, Universidade Estadual Paulista, 14801-970, PO Box 355, Araraquara-SP, Brazil b Departamento de Quı ´mica, Universidade Federal de Sa˜o Carlos, 13560-970, PO Box 676, Sa˜o Carlos-SP, Brazil c Instituto de Quı ´mica, Universidade de Sa˜o Paulo, 05513-970, PO Box 26077, Sa˜o Paulo-SP, Brazil Received 14 September 2004; accepted 13 January 2005 Available online 2 April 2005 Abstract Arsenic and germanium have been evaluated as internal standards to minimize matrix effects on the direct determination of selenium in milk by graphite furnace atomic absorption spectrometry (GFAAS) using tubes with integrated platform, pre-treated with W together with Pd as chemical modifier. The efficiency of As and Ge as internal standards for 25 AgL À 1 Se plus 500 AgL À 1 As or Ge in diluted (1 + 9 v/v) milk plus 1.0% (v/v) HNO 3 was evaluated by means of correlation graphs plotted from the normalized absorbance signals (n = 20) of internal standard (axis y ) versus analyte (axis x ). The equations that describe the linear regression were: A As = À 0.004 F 0.019 + 1.02 F 0.019 A Se (r = 0.9967 F 0.005); A Ge = À 0.017 F 0.015 + 1.01 F 0.015 A Se (r = 0.9978 F 0.004). Samples and reference solutions were automatically spiked with 500 AgL À 1 Ge or As and 1.0% (v/v) HNO 3 by the autosampler. For 20 AL of aqueous standard solutions, analytical curves in the 5.00–40.0 AgL À 1 Se range were established using the ratio of Se absorbance to internal standard absorbance (A Se / A IS ) versus analyte concentration, and good linear correlations were obtained. The characteristic mass was 40 pg Se. Limits of detection were 0.55 and 0.40 Ag L À 1 with As and Ge as the internal standard, respectively. Relative standard deviations (RSD) for a sample containing 25 AgL À 1 Se were 1.2% and 1.0% (n =12) using As and Ge, respectively. The RSD without internal standardization was about 6%. The accuracy of the proposed method was evaluated by an addition-recovery experiment and all recovered values were in the 99–105% range with IS and in the 70–80% range without IS. Using Ge as the internal standard, results of analysis of standard reference materials were in agreement with certified values at a 95% confidence level. The selenium concentration for 10 analyzed milk samples varied from 5.0 to 20 AgL À 1 . D 2005 Elsevier B.V. All rights reserved. Keywords: Milk; Internal standardization; GFAAS; Selenium 1. Introduction There is a great interest in the determination of selenium in milk due to the biochemistry of this element and its compounds [1–3]. The milk matrix is rather complex and typically contains proteins, carbohydrates, saturated fats, cholesterol, monounsaturated fats, polyunsaturated fats and essential minerals, such as Na, K, Ca, Mg, Fe, P, Mn, Zn, Cu, and Se, which may cause errors in the analytical results [4–6]. Calibration by matrix matching might not be applicable for the direct determination of Se in milk samples, as the slope of analytical curves obtained in whole, low fat, and non-fat milk samples might be considerably different. Heterogeneous matrices might also affect the quality of sampling by changing the amount of analyte inside the pipette of the automatic sampler of the spectrometer. As a 0584-8547/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.sab.2005.02.005 B This paper was presented at the 8th Rio Symposium on Atomic Spectrometry, held in Paraty, RJ, Brazil, 1–6 August 2004, and is published in the special issue of Spectrochimica Acta Part B, dedicated to that conference. T Corresponding author. Tel.: +55 16 33016611; fax: +55 16 33227932. E-mail address: anchieta@iq.unesp.br (J.A. Gomes Neto). Spectrochimica Acta Part B 60 (2005) 681 – 686 www.elsevier.com/locate/sab