Using Iterative Discrete Wavelet Transform to Improve Trace Element Analysis by XRF Spectrometer

Abstract

X-ray fluorescence (XRF) spectrometry has certain difficulties of detecting trace amount material components accurately when measuring material samples composed of variable elements, mainly due to low Signal to Noise Ratio (SNR) issues of the characteristic spectroscopic peaks from the measurement. In this paper, a novel method called background noise reduction using Iterative Discrete Wavelet Transform (IDWT) methodology for trace element material analysis by advanced X-ray fluorescence spectrometer is proposed to improve SNR, thereby decreasing the Limit of Detection (LOD) for elemental qualitative analysis, and then achieve a more accurate quantitative analysis of trace elemental concentration. This paper utilized handheld X-ray fluorescence spectrometer to obtain the content of Sulphur in petroleum and 4 major pollution elements in soil. A total of 81 standard samples were collected and measured. The hardware parameters of the instrument were adjusted to optimize the SNR before background noise reduction. Experimental results illustrate that X-ray tube parameters have great influences on the calibration regression. Different X-ray tube voltages were tested and the optimal results were achieved at 5kV. Furthermore, IDWT algorithm was implemented and the optimal results were achieved by wavelet base “db5” and “sym4” with 7 level decomposition. The calibration regression curves were established for the Sulphur in petroleum. The regression R2 values after IDWT were increased effectively when compared with original data without IDWT. Finally, the experimental results demonstrate a very good linearity between the weight contents of the target material and the XRF spectral characteristic peak intensity, and also it is found the LOD for Sulphur in petroleum can be reduced when combing with the IDWT.