X射线荧光光谱中低分离度重叠峰分解的方法研究

Research on a Decomposing Method of Low-Resolution Overlapping Peaks in X-Ray Fluorescence Spectra

由于轻元素自身特征X射线以及测量元素间特征X射线的相互干扰,受仪器能量分辨率的影响,实测X射线荧光光谱会产生严重重叠。以色谱分离度Rs作为计算谱峰重叠程度的指标,Rs低于0.5的重叠峰作为研究对象,提出一种峰锐化法结合双树复小波变换分解低分离度重叠峰的新方法,并对模拟X射线荧光光谱和实测X射线荧光光谱进行了新方法的验证。首先,在详细介绍峰锐化法和双树复小波变换分解重叠峰原理的基础上,通过仿真结果发现:当Rs=0.38时,两种方法皆不能单独实现重叠峰的分解;然而,峰锐化法处理后的信号不仅保留了原始信号的峰位特征,还出现了分离度明显变大的现象。因此,可以通过调节峰锐化法的权值实现对低分离度重叠峰的初步锐化,再对锐化后的信号做双树复小波变换,结果实现了对模拟重叠峰的分解,验证了新方法分解低分离度重叠峰的优越性。其中,双树复小波变换的分解层数为2~6层,第一层选择near_sym_b滤波器,第一层以上选择qshift_d滤波器,且当细节系数放大倍数为1~10时,重叠峰的分解结果更准确。然后,模拟了K元素K_α能量峰与其K_β能量峰的重叠光谱(Rs=0.44)以及Fe元素K_β能量峰与Co元素K_α能量峰的重叠光谱(Rs=0.34),用新方法对谱线进行处理,结果实现了重叠峰分解,且分解后峰位和峰面积的相对误差分别在1%和6%以内,验证了该方法分解光谱中低分离度重叠峰的可行性。最后,用新方法对实测的Ca元素X射线荧光光谱进行处理,最终也实现了重叠峰分解,且分解后的峰位相对误差分别为0.8%和0.7%。结果证明:峰锐化法结合双树复小波变换能够有效分解低分离度重叠峰,且在解决X射线荧光光谱中谱峰严重重叠的问题上具有实用性。

Since the element’s own characteristic X-rays with low atomic number and the characteristic X-rays between different elements interfere with each other, the measured spectra will be seriously overlapped due to the limitation of the energy resolution of the instrument. In this paper, chromatographic resolution Rs is used to calculate the separation degree of spectral peaks. Taking the overlapping peaks with Rs below 0.5 as the research object, a new method of decomposing low-resolution overlapping peak by using a peak sharpening method combined with the double-tree complex wavelet transform method is proposed, which is verified by decomposing the simulated X-ray fluorescence spectra and the measured X-ray fluorescence spectra. First, based on the principle introduction of the peak sharpening method and the double-tree complex wavelet transform to decompose overlapping peak in detail, the simulation results showed that when Rs=0.38, both methods can not decompose the overlapping peaks with low-resolution separately. However, the spectrum processed by the peak sharpening method not only retains the peaks position characteristics of the original spectrum, but also has the phenomenon that the resolution becomes larger. Therefore, it can realize the initial sharpening of the low-resolution overlapping peak by adjusting the weight of the peak sharpening method, then, using the double-tree complex wavelet transform is performed on the sharpened spectrum. The results showed that the simulated overlapping peak is decomposed. It is proved that the new method is superior to the decomposition of low-resolution overlapping peak. Wherein, the decomposing level of the double-tree complex wavelet is set to 2~6, the first level selects the near_sym_b filter, the higher level selects the qshift_d filter, and when the detail coefficient magnification is set to 1~10, the decomposing results of the overlapping peak is more accurate. Second, the overlapping spectrum of the K_α energy peak and the K_β energy peak of K(Rs=0.44) and the overlapping spectrum of the K_β energy peak of Fe and the K_α energy peak of Co(Rs=0.34) are simulated. The new method is used to decompose the spectra, and the two overlapping spectra are decomposed. The relative error of peak position and peak area of the decomposed spectra is within 1% and 6% respectively, which verified the feasibility of the new method to decompose the low resolution overlapping peaks in the spectrum. Finally, the actual measurement Ca X-ray fluorescence spectrum is decomposed by a new method, and the errors of the peak position are 0.8% and 0.7% respectively. The results showed that the peak sharpening method combined with the double-tree complex wavelet transform can effectively decompose the low-resolution overlapping peaks, and it has practicality in solving the problem of severe overlapping of X-ray fluorescence spectra.