Desenvolvimento de método para análise de leite por espectrometria de fluorescência de raios X por reflexão total

Abstract

This work represents a study in which commercial UHT milk samples with different levels of fat (whole, semi-skimmed and skimmed) and special milk with low lactose and calcium-fortified were collected in order to make a comparison and establish a possible correlation between the milk type and the concentrations of P, S, K, Ca, Cu, Zn and Br obtained from the analytical technique Total Reflection X-Ray Fluorescence (TXRF). For this it was first performed an optimization to select the best element to act as internal standard, as well as its optimal concentration and the analysis time that provides the best recovery values of the analytes. The configuration of best response was that used Y with concentration of 10 mg/L with an analysis time of 250 seconds. Thus, analysis of samples digested in a microwave oven using HNO3 and H2O2 was performed. In evaluation of results, a data matrix was constructed for the application of multivariate method Principal Component Analysis (PCA), which showed that there is no effective separation of the different analyzed milk type, but it can be observed a larger group of whole milk that is related to higher content of Cu and Br. In the second part of this study, a method was developed for lead determination in bovine raw milk, in order to attend the Commission Regulation (EC) No 1881/2006 and CODEX STAN 193-1995 2010 that establish 20 g/L as ceiling allowed of lead in milk. Due to the low concentration allowed, the use of preconcentration technique known as Cloud Point Extraction (CPE) was required, in which was used the ammonium O,O-diethyl dithiophosphate (DDTP) as a complexing agent and Triton X-114 as surfactant. As raw milk is a very complex matrix and has high fat content, it was necessary to perform the acid digestion of the sample before extraction cloud point. An optimization of the pre-concentration step was conducted employing central composite design varying volumes of the reagents used in the procedure. The accuracy and precision of the method were studied by analyzing samples doped in three different lead concentrations: 20, 50 and 100 g/L. The method presented values between 80.8 and 112 % for recovery, and between 7 and 10 % for the standard deviation, being within acceptable ranges established by Regulation 657/2002 of the European Community. The average values for the detection limits obtained for the samples was 2,0; 4,5 and 5,8 g/L for concentrations of 20, 50 and 100 g/L, respectively. The calculated pre-concentration factor was 4 to 5 times.