由光子相关谱反演微粒体系粒径分布方法的分析与比较

在光子相关法测量微粒体系粒度分布技术中,反演算法的选择及其正确使用是很关键的。本文中首先概要介绍了目前流行的累积量法(Cumulants)、双指数拟合法(DE)、非负最小二乘法(NNLS)、CONTIN算法和指数采样分析法(ES)等几种反演算法。然后利用计算机模拟产生的光子相关谱对这几种算法的性能进行了详细的对比研究。结果表明:在无噪声的情况下,对于单分散的微粒体系,除DB外,其它4种算法均能给出较准确的平均粒径,而分布宽度以NNLS为最佳。对间隔明显且峰较尖锐的双峰分布微粒体系,除累积量法外,其它4种算法均有可能分辨出双峰的存在,DE对峰高的估计最准,而NNLS对双峰的分辨力最强,CONTIN算法对双峰的分辨力最弱。对于分布较宽的多分散微粒体系,DE无能为力,累计量法也不适用,较为适用的是ES、NNLS和CONTIN算法。

Analysis and Comparison of Arithmetic for Inverting Particle Size Distribution from Photon Correlation Spectrum

The selection and using of the reversing arithmetic are very important for measuring particle size distribution by photon correlation technology. Several inverting arithmetic, such as Cumulants, double exponential(DE), non- negatively constrained least squares(NNLS), CONTIN and exponential sampling (ES) are introduced first. Then by use of computer simulated photon correlation spectrums, a detailed comparison is done among these inverting arithmetic. The results show that, if there is no noise, for a monodispersed system, the other four arithmetic can give rather correct mean diameters except for DE, as for the estimation of the distribution width, NNLS is the best. For well apart double- peak distribution, the other four arithmetic can recognize the peaks except for Cumulants. DE is the best for estimating the peak height, NNLS is the best for recognizing and CONTIN is the worst. For widely distributed polydispersity system, ES, NNLS and CONTIN are capable, but DE and Cumulants are helpless.