双峰分布颗粒体系的多角度动态光散射数据反演（英文）

采用多角度动态光散射和加权正则化反演方法,对4组模拟的双峰分布颗粒体系(100/600nm,200/600nm,300/600nm和350/600nm)分别选取1、3、6和10个散射角进行测量.粒度反演结果表明,采用加权正则化方法反演双峰颗粒体系的多角度动态光散射测量数据,可获得峰值位置比小于2∶1且含有大粒径(350nm)颗粒的双峰颗粒粒度分布.采用标准聚苯乙烯乳胶颗粒进行实测的结果验证了这一结论.得到含大粒径颗粒的双峰粒度分布反演结果的原因在于,多角度动态光散射能提供更多的大粒径颗粒的粒度信息,加权正则化反演方法能减少测量数据中的噪声,因而多角度动态光散射测量数据的加权反演能实现峰值位置比小于2∶1且含有大粒径颗粒的双峰颗粒体系的测量.     

动态光散射测量的复惩罚正则化反演

多峰颗粒体系粒度及其分布的测量是动态光散射技术的难点,本文在Tikhonov正则化方法的目标函数中加入具有平坦约束功能的惩罚项,增强对解的约束提高对多峰颗粒体系的反演性能.190/443nm、282/953nm、457/553nm双峰分布颗粒体系、564nm单峰分布颗粒体系和292/591/889nm三峰颗粒体系的模拟数据,以及306/974nm、300/502nm双峰颗粒体系的实测数据的反演表明,在正则化反演中增加具有平坦约束功能的惩罚项,可有效消除反演的颗粒粒度分布中出现的毛刺与虚假峰,提高算法的峰值分辨能力和抗噪能力.该研究在发挥多角度动态光散射技术测量中、大超细颗粒时具有信息量多的优势,实现宽范围的双峰及多峰分布颗粒体系的准确测量.     

基于颗粒粒度信息分布特征的动态光散射加权反演

宽分布和双峰分布颗粒的准确反演是动态光散射技术至今未能有效解决的难题,尤其峰值位置比小于2:1且含有大粒径颗粒(350 nm)的双峰分布.造成这一难题的主要原因包括:1)单角度测量数据的粒度信息含量不足;2)常规反演方法对测量数据的噪声抑制以及粒度信息利用缺乏针对性.对测量数据(即光强自相关函数)的研究发现,数据噪声主要分布在长延迟时段,而粒度信息集中分布在衰减延迟时段.基于此,本文提出了采用粒度信息分布为底数、调节参数为指数的权重系数对自相关函数进行加权反演的约束正则化方法.由于采用了与粒度信息分布一致的权重系数,该方法既充分利用了衰减延迟时段的粒度信息,又有效地抑制了长延迟时段的数据噪声.不同噪声水平下,宽分布和双峰分布颗粒体系的反演结果表明,与常规反演方法相比,这一方法可以获得更为准确的宽分布和近双峰分布的反演结果.     

含噪动态光散射测量数据反演中正则化算法与Chahine算法的比较

采用两种常用的粒度反演方法——正则化和Chahine算法,对90nm与250nm单峰分布、50nm与200nm双峰分布、100nm与300nm双峰分布的模拟动态光散射数据,以及105nm、300nm标准颗粒的实测动态光散射数据进行了反演分析.结果表明:噪声水平的高低是影响粒度分布反演准确性的关键因素之一,反演结果的准确性随噪声水平的增加而降低,噪声水平超过某一阈值后,将无法得到有意义的反演结果;不同反演方法具有不同的抗噪能力,在低噪声水平下反演结果无显著差别,随着噪声水平的增加,反演结果表现出很大差异;正则化方法通过正则参数的选择可以有效抑制噪声影响,表现出强于Chahine算法的抗噪能力;与Chahine算法相比,正则化方法不需要假定初始分布,因此,在噪声较大的实验或生产过程中进行颗粒分布测量时,宜采用正则化方法.     

初始模型对含噪动态光散射数据正则化反演结果的影响

分别采用最小模型矩阵、最平坦模型矩阵、最光滑模型矩阵作为初始化模型,对加入5种不同水平随机噪声的90nm窄单峰、90nm宽单峰和250nm窄单峰、250nm宽单峰颗粒体系的模拟分布进行了正则化反演,并对反演结果进行比较。结果表明:当噪声水平为0时,正则化初始模型的选择对反演结果没有明显影响。随着噪声水平的增加,采用三种初始化模型反演得到的峰值误差和粒度分布误差都随之变大,但采用最平坦模型和最光滑模型反演得到的峰值和粒度分布误差明显小于采用最小初始模型的反演误差。当噪声水平大于0.01时,选择最平坦初始模型获得的粒度分布结果优于采用最光滑初始模型和最小初始模型获得的结果,而采用最光滑初始模型反演得到的峰值优于最平坦初始模型和最小初始模型的反演峰值。因此,采用正则化算法处理含噪动态光散射数据时,为得到最优的粒度分布信息,宜采用最平坦初始模型,若需要获取最准确的峰值信息,则应选择最光滑初始模型。     

基于Mie散射光强的多角度动态光散射复合角度加权方法

针对多角度动态光散射测量技术中通过Mie散射光强计算的角度权重估计方法存在信息利用率与抗噪性之间的矛盾,提出利用每一角度所有粒度的整体Mie散射空间特征进行角度加权和利用每一粒度对应Mie散射光的细节特征对核矩阵做元素加权的复合角度加权方法,并结合正则化方法进行了模拟和实测的多角度动态光散射数据反演.与采用光强比值法和光强均值法的反演结果比较表明,多角度动态光散射反演结果与角度加权方法密切相关.无噪声影响时,光强比值法和复合角度加权法都能得到准确的颗粒粒度分布,但光强均值法信息利用率不高;随着噪声水平的提高,光强比值法反演结果急剧变差,表现出较低的抗噪性能.复合角度加权方法通过兼顾信息利用率和抗噪性能,使得增加散射角时信息增多的优势得以更好地显现,并且有效地抑制了角度增多带来的噪声影响.该加权方法显著提高了多角度动态光散射进行颗粒测量,特别是对多峰分布颗粒体系测量的准确性.     

多角度动态光散射角度误差影响测量的途径分析

在分析角度误差对测量结果作用机制的基础上,采用不同的角度误差,通过对六组单峰(82nm、104nm、350nm、431nm、816nm和865nm)和三组双峰(137/601nm、242/750nm和470/895nm)颗粒体系的模拟数据及306nm/974nm双峰颗粒体系实测动态光散射数据的反演,研究了权重估计和基线计算两种途径中角度误差对反演结果的作用.结果表明,角度误差对颗粒粒度分布反演结果的影响是由基线计算和权重系数估计的双重作用途径产生的;权重因素对峰值和性能误差的影响明显大于基线因素,仅在小颗粒窄峰测量时,基线因素对峰值误差的影响略超权重因素,但权重因素对性能误差的影响仍然大于基线因素.     

将噪声作为独立变量的动态光散射数据反演

在动态光散射颗粒测量时,为了从含噪的自相关函数数据中准确地反演出颗粒粒度分布,对Tikhonov正则化算法进行改进,将噪声作为一个独立的未知变量应用到正则化方程中进行粒度反演.在计算过程中,相应增加方程中各系数矩阵的行数和列数,对求解的粒度分布数值则仍取其原来方程的行数和列数,从而达到对部分噪声的剔除作用.不同噪声水平下的颗粒粒度反演结果表明,改进后的算法能够显著提高低信噪比动态光散射数据粒度反演结果的准确性,适用于宽分布较大粒径的颗粒粒度反演.     

将噪声作为独立变量的动态光散射数据反演（英文）

在动态光散射颗粒测量时,为了从含噪的自相关函数数据中准确地反演出颗粒粒度分布,对Tikhonov正则化算法进行改进,将噪声作为一个独立的未知变量应用到正则化方程中进行粒度反演.在计算过程中,相应增加方程中各系数矩阵的行数和列数,对求解的粒度分布数值则仍取其原来方程的行数和列数,从而达到对部分噪声的剔除作用.不同噪声水平下的颗粒粒度反演结果表明,改进后的算法能够显著提高低信噪比动态光散射数据粒度反演结果的准确性,适用于宽分布较大粒径的颗粒粒度反演.     

CONTIN算法及其在测量微凝胶颗粒粒度分布中的应用

颗粒粒径大小决定了微凝胶的相变行为,因此采用有效可靠的手段来确定胶体悬浮液中颗粒的平均粒径及粒度分布是至关重要的。CONTIN算法是分析动态光散射实验数据,获取胶体悬浮液中颗粒粒度分布的有效算法,但目前的最优正则化参数选取策略依赖于颗粒粒度分布的先验条件。为此,本文提出利用V-曲线准则获取最优正则化参数,使用CONTIN算法表征微凝胶悬浮液颗粒系的平均粒径和粒径分布信息。实验结果表明,与V-曲线正则化参数选取准则相结合,利用CONTIN算法可以有效的获取微凝胶悬浮液的颗粒粒度信息。     

基于多尺度变换的动态光散射粒径反演范围的自适应调整

 通过多尺度变换实现了反演范围的自适应调整,使其更接近真实范围。分别采用反演范围固定算法与自适应算法对200～600 nm单峰和200～900 nm双峰分布颗粒的模拟相关函数进行了反演,结果表明：自适应算法的结果更接近理论分布,抗干扰能力更强。相对于固定算法,单峰分布颗粒最多可缩小峰值误差4.73%,缩小峰宽误差185 nm。双峰分布颗粒在0～0.001噪声水平时,峰值误差分别小于11.33%,12.45%,峰宽误差分别小于35,160 nm,而固定算法在噪声水平大于0.000 1时,难以得到合理的反演结果。反演范围自适应调整方法能够有效优化粒径反演结果。     

Particle Size Distributions from Static Light Scattering with Regularized Non‐Negative Least Squares Constraints

Simulated data from static light scattering produced by several particle size distributions (PSD) of spherical particles in dilute solution is analyzed with a regularized non‐negative least squares method (r‐NNLS). Strong fluctuations in broad PSD's obtained from direct application of NNLS are supressed through an averaging procedure, as introduced long ago in the inversion problem in dynamic light scattering. A positive correlation between the best PSD obtained from several averaging schemes and the condition number of the respective data transfer matrices was obtained. The performance of the method is found to be similar to that of constrained regularization (CONTIN), which uses also NNLS as a starting solution, but incorporates another regularizing strategy.     

多角度动态光散射加权贝叶斯反演算法

研究加权贝叶斯算法在多角度动态光散射法测量单峰分布颗粒体系的颗粒粒度分布中的应用.采用颗粒粒度信息分布为底数、调节参数为指数的权重系数给各个角度下的光强自相关函数曲线加入不同的权重系数，再利用传统的贝叶斯算法反演.模拟与实验结果表明，加权后的贝叶斯算法能获得分布误差更小的反演结果，有效地抑制了数据噪声的影响，提高颗粒粒度分布反演的准确性.     

The reconstruction of particle size distributions from dynamic light scattering data using particle swarm optimization techniques with different objective functions

In this paper, the reconstruction of particle size distributions (PSDs) using particle swarm optimization (PSO) techniques from dynamic light scattering (DLS) data was established. Three different objective functions containing non-smooth constrained objective function, smooth functional objective function of Tikhonov regularization and L objective function, were employed. Simulated results of unimodal, bimodal and bi-dispersed particles show that the PSO technique with non-smooth constrained objective function produces narrower PSDs focusing on peak position in the presence of random noise, the PSO technique with smooth functional of Tikhonov regularization creates relative smooth PSDs, which could be successfully applied to the broad particles inversion, and the PSO technique with L objective function yields smooth PSDs, which saves calculation amount. Experimental results as well as comparisons with CONTIN algorithm and Cumulants method demonstrate the performance of our algorithms. Therefore, the PSO techniques employing the three different objective functions, which only require objective function and need a few initial guesses, may be applied to the reconstruction of PSDs from DLS data.     

Intercomparison of Inversion Algorithms for Particle‐Sizing Using Mie Scattering

The applicability of different inversion algorithms to retrieve a size distribution of particles in air from light scattering is examined. The investigation is focused on an optical measurement setup with an elliptical mirror as the main optical element. In order to evaluate the capabilities of the individual inversion methods, light scattering by spherical particles is simulated in the size ranges of 0.1 – 10 μm and 0.05 – 1 μm. The distribution of the particle diameters is modeled with three different parametric functions, i.e., RRSB, logarithmic‐normal and a more specific distribution from an ultrasonic nebulizer. Different kinds of noise, e.g., additive and/or multiplicative, are applied in different levels to the simulated scattering measurement to include real physical measurement conditions. The convergence properties of the scattering simulation are investigated with respect to the number of size classes, and thus, information concerning the size resolution required to simulate a measurement for a given particle size distribution is obtained. Further parameters of interest are the minimum angular resolution of the measurements, the number of size classes of the retrieved particle size distribution and the measured polarization of the scattered light.     

A V-curve criterion for the parameter optimization of the Tikhonov regularization inversion algorithm for particle sizing

The regularization parameter plays an important role in applying the Tikhonov regularization method to recover the particle size distribution from dynamic light scattering experiments. The so-called V-curve, which is a plot of the product of the residual norm and the norm of the recovered distribution versus all valid regularization parameters, can be used to estimate the result of inversion. Numerical simulation demonstrated that the resultant V-curve can be applied to optimize the regularization parameter. The regularization parameter is optimized corresponding to the minimum value of the V-curve. Simulation and experimental results show that stable distributions can be retrieved using the Tikhonov regularization with optimum parameter for unimodal particle size distributions.     

Comparative Test of Methods to Determine Particle Size and Particle Size Distribution in the Submicron Range

Among the most important characteristic properties of disperse systems such as latices, pigments, ceramic materials or drug formulations are the particle size and the particle size distribution. To measure these quantities, several methods and measuring instruments based on different physical principles are available. These include turbidimetry, dynamic and static light scattering, electron microscopy with image analysis, ultra- and disc centrifugation, light diffraction and the electrical sensing zone method. All these measuring techniques are doubtless necessary because of the large product variety and the broad particle size range. However, some problems arise if different techniques are used and the results are compared uncritically without considering to the application range and the resolution of the methods. An extensive comparative test was therefore carried out using seven latices in the submicron range with defined monomodal, bimodal and hexamodal particle size distributions. The most important methods of determining average particle size values and particle size distributions were tested and compared. Of the methods to determine only average particle sizes, turbidimetry is the most efficient, followed by dynamic light scattering with cumulants evaluation. Static light scattering only yields accurate results for small particles with narrow particle size distributions. Of the methods to determine particle size distributions, ultracentrifugation and, somewhat less, disc centrifugation and electron microscopy with image analysis are the most efficient. Dynamic light scattering only yields reliable results in the case of small particles with narrow distribution curves. Light diffraction and the electrical sensing zone method are less suitable for the submicron range.