Study On The Denoising Method In Multiangle Dynamic Light Scattering Particle Sizing

Multiangle Dynamic Light Scattering(MDLS) technique measures the scatteredlight signal at different scattering angles and calculates the light autocorrelationscorresponding to the angles, then the weighted angular light autocorrelations areanalyzed to get particle size distribution. The technique overcomes the defect ofinadequate information in single-angle autocorrelation function and gives more accuratemeasurement results for broad and bimodal distribution particle systems. However,because of the increased scattering angles, a lot of measurement noise is introduced tothe experiment data, leading to the complexity in data processing and the difficulty inparticle size inversion. So how to effectively eliminate the noise caused by the increasingscattering angles is vital for MDLS particle sizing. This paper mainly studied thedenoising method in MDLS particle sizing,which consists of,1. The basic theory of Multiangle Dynamic Light Scattering technique based onsingle angle dynamic light scattering technique is analyzed. It selected several measuringangles with a rotating apparatus to measure the light intensity autocorrelation functionsof the selected measuring angles, and combined them into a data analysis with properangular weighting coefficients. Further more, the noise introduction in MDLS techniqueis discussed.2. The baseline denoising method in Multiangle Dynamic Light Scattering wasstudied. Compared to single angle measurement, the Multiangle Dynamic LightScattering technique introduces more baseline measurement noise which results in theemerging of false peak at large particle size in PSD. So the baseline error denoisingmethod in MDLS aiming at the problem was put forward.The inversion results ofbimodal simulation distribution data of300nm and600nm measured and a dilutedbimodal experiment data of306nm and974nm polystyrene latex standard spheres showthat the baseline denoising method significantly reduces baseline error and proves bettereffect on the inversion of MDLS from light intensity autocorrelation data includingbaseline noise.3. The influence of angle calibration noise on the PSD was studied. The Johnson’sSBfunction was used to simulate the distribution, and correlate noise with differentsignal to noise ratio (SNR) is added to the autocorrelation functions data correspondingto small, medium and large unimodal distribution and bimodal distribution mixed withthe particles mentioned. The inversion results show that small particles in unimodal distribution was more greatly affected by angle calibration noise than medium and largeparticles,also the bimodal distribution containing small particles was more easily affectedby the angular calibration noise than bimodal distribution without small particle,besides,the broad modal distribution is more easily corrupted by angular calibration noise thanthe narrow modal distribution.4. The angular weighting coefficients denoising method in MDLS was studied.Due to the inevitable introduction of baseline measurement noise, the normalizedintensity autocorrelation data deviates from the true value, leading to the incorrectweighting estimations, which seriously affect the inversion accuracy. Aiming at theproblem, the angular weighting coefficient denoising method is put forward. Theinversion results of the simulated MDLS data of two groups of100~900nm and100~650nm unimodal distribution and two groups of360~900nm and100~900nmbimodal distribution show that, with the denoising method, the error of angularweighting coefficients is significantly reduced, whats more, the performance parametersof the particle size distribution and the relative error of peak value are obviouslydecreased.The experiment data of306/974nm bimodal distribution further verified thedenoising method.MDLS can analyze the scattered light intensity at different angles and avoid thewrong angle selection for an unknown particle system,proving more information aboutparticle size distribution. Thus there will be more broad application prospects than singleangle dynamic light scattering. However, the technique is hindered by the measurementnoise caused by the increasing scattering angle, leading to a slow development in recent20years. This work can effectively decrease the noise influence on the particle sizingand significantly improve the accuracy of PSD estimation, contributing to promote thedevelopment of MDLS particle sizing technique.