Shuaifeng Zhang, Qinghua Zhang,*, Jianzhuang Shang, Zaisha Mao, Chao Yang,*

1CASKey Laboratoryof Green Process and Engineering,Instituteof Process Engineering,Chinese Academy of Sciences, Beijing 100190, China

2School of Chemical Engineering,University of Chinese Academyof Sciences, Beijing 100049,China

3China National Petroleum & Chemical Planning Institute,Beijing100029, China

Keywords:Particle size distribution Emulsion polymerization Off-line On-line In-line

A B S T R A C T The particle size distribution of polymer always develops in emulsion polymerization systems, and certain key phenomena/mechanisms as well as properties of the final product are significantly affected by this distribution. This review mainly focuses on the measurement methods of particle size distribution rather than average particle size during the emulsion polymerization process, including the existing off-line, on-line, and in-line measurement methods. Moreover, the principle, resolution, performance,advantages, and drawbacks of various methods for evaluating particle size distribution are contrasted and illustrated.Besides,several possible development directions or solutions of the in-line measurement technology are explored

1. Introduction

After nearly a century of development, emulsion polymerization (EP) has matured into one of the most versatile and essential techniques for the synthesis of polymer latexes.Industrial applications of these polymers include synthetic rubber (e.g., styrenebutadiene rubber, SBR), synthetic resin (e.g., polyvinyl chloride,PVC), and polymer latexes, which cover adhesives [1,2], paints[3], coatings [4], film materials [5], additives in paper and textiles[6]. Nowadays, emulsion polymerization is increasingly used to produce functional polymer latexes for more specialized fields such as the electronics and biomedical fields [7].

Practically speaking,the particle size distribution(PSD)is a crucial property parameter of latex prepared by emulsion polymerization [8]. First of all, the particle size distribution has a significant influence on several fundamental properties of the final product prepared by emulsion polymerization: viscosity, maximum solid content, adhesion, drying time, polymer rheology, optical properties, mechanical strength, and latex stability [9]. For example, the surface properties of the dried polymer film are primarily determined by the particle size distribution [10]. The solid content of latex most often requires a very well-defined particle size distribution to keep an acceptable viscosity level [11,12].Furthermore, particle size distribution may influence reaction kinetics [13]. Edouardet al.[9] and Herrera-Ordonez [14] also emphasized the important roles of particle size distribution on the reaction kinetics determination and reaction rate. In fact, particle size distribution always occurs in emulsion polymerization systems, and certain key phenomena are significantly affected by this distribution. In particular, particle formation is an excellent example of this type of phenomenon: the kinetics of particle formation dramatically depends on the overall size distribution of the particles[15].Therefore,tools to measure the particle size distribution during the emulsion polymerization are useful and essential.

At present, the particle size distribution of latex prepared by emulsion polymerization can be obtained through experimental methods or numerical calculations. Numerical calculation techniques have been discussed in detail by Sheibat-Othmanet al.[15], but some mathematical models require experimental in-line measurement data for verification. So, this review mainly focuses on the measurement methods of particle size distribution rather than other quality-related properties during emulsion polymerization. According to whether to sample and how to sample, the measurement methods of PSD are divided into off-line, on-line,and in-line measurement. The schematic of method classification is shown in Fig. 1. The off-line measurement requires manual sampling, and the on-line measurement uses automatic sampling,but the in-line measurement does not perform sampling. The off-line measurement is very time-consuming due to manual sample preparation. Therefore, the on-line and in-line measurement methods have been developed and improved in order to shorten the time for measuring particle size distribution and possible instant adjustment of process control parameters.

Fig. 1. Classification of particle size distribution measurement methods.

This work mainly focuses on the measurement methods of particle size distribution during emulsion polymerization, including the existing off-line, on-line, and in-line methods. Furthermore,the principle, resolutions, advantages, and disadvantages of each measurement method are summarized.Meanwhile,some practical guidelines are given, which will help researchers in this field to effectively use these measurement methods to monitor the various processes of particle synthesis and operate in a way that avoids common mistakes. When summarizing in-line and on-line measurement technologies, we will outline the latest technologies and highlight areas where progress needs to be made,either reach a consensus on the conflict of views or just deepen our current understanding of the system. The difficulties and challenges of these in-line methods are discussed in detail, and several possible development directions or solutions of the in-line measurement technology are given to conclude the review.

2. Some Fundamental Concepts

This section briefly explains the basic concepts of emulsion polymerization, the definition of particle size distribution, and the classification criteria for measurement methods of particle size distribution.At the same time,the subtle differences and relationships between the results of different measurement methods are described.

2.1. Emulsion polymerization

The basic knowledge and mechanism model of emulsion polymerization is critical to controlling the particle size distribution of products. A simple recipe for emulsion polymerization consists of monomer (usually relatively insoluble in water), continuous aqueous phase, water-soluble initiator, and emulsifier (or stabilizer) [16]. In an aqueous solution, it begins with emulsifying and dispersing the monomer into dropletsviasurfactant aid,which stabilizes the monomer droplets (≥1000 nm). When the surfactant concentration is higher than the critical micelle concentration(CMC), micelles (4-5 nm) will appear in the aqueous solution.The micelles swell by absorbing monomer molecules inside to form monomer-swollen micelles (6-10 nm). The polymerization occurs in the monomer-swollen micelles when the initiator is added, and as the polymerization progresses, polymer particles are formed with diameters (50-500 nm) intermediate to those of micelles and droplets.

According to Harkins [17] and Smith-Ewart’s theory [18], the process of emulsion polymerization can be divided into three distinctive stages, and the schematic diagram of polymer particle growth is shown in Fig. 2.

Interval I: particle nucleation;

Interval II: particle growth in the presence of monomer droplets;

Interval III: particle growth in the absence of monomer droplets.

Nucleation is an essential factor affecting particle size distribution. Particles formed at the beginning of interval I will grow over the entire period of interval I, while particles formed at the end of interval I will have grown very little.In addition,particle secondary nucleation and coagulation maybe occur during Intervals II and III.Fig. 2(a) shows the two main nucleation mechanisms of emulsion polymerization:micellar nucleation and homogeneous nucleation;Fig. 2(b) shows the growth of polymer particles in the presence of monomer droplets, while Fig. 2(c) shows the growth of polymer particles in the absence of monomer droplets. Emulsion polymerization involves four main kinetic steps: initiation, propagation,termination, and chain transfer;detailed emulsion polymerization is described elsewhere [19]. Although the mechanism of emulsion polymerization is gradually improving,some of its aspects are still not completely understood, such as polymer particle nucleation,radical absorption and desorption, and monomer partitioning[15,19,20]. Even for the same reaction recipe, the properties of polymers produced by emulsion polymerization will be greatly affected by the process used. Batch process, continuous or semicontinuous processes are the main ones for the industrial production of polymers.

Fig. 2. Process of emulsion polymerization: (a) Interval I, (b) Interval II, (c) Interval III.

2.2. Particle size distribution

Particle size distribution strictly means particle diameter distribution [19]. In this article, the term ‘PSD’ refers specifically to the particle size distribution of polymer particles during emulsion polymerization, excluding the particle size distribution of monomer droplets. The particle collection of the latex produced by emulsion polymerization always displays PSD, because there is always some random fluctuation in its generation or production processes.

The particles in the emulsion polymerization process may have irregular shapes, and even spherical particles will have irregular shapes due to agglomeration. There is usually no single number or set of numbers to describe the physical size well enough for irregular particles. Generally, the virtual diameter or the equivalent spherical diameter (ESD) of particles that approximates the characteristics of spherical particles will be used [21]. The introduction of ESD into the field of particle technology enables PSD measurement to perform particle statistical analysis in a less ambiguous manner[22].ESD largely depends on the physical principle on which the sizing method is applied,and some examples of equivalent spherical diameters are given in Table 1.

Table 1 Definition of equivalent spherical diameters [22].

In general, there is always a particle size distribution in the emulsion polymerization system. According to NIST regulations of the USA, if at least 90% of the particles are contained within 5%of the median size,these particles are called monodisperse particles [23]. Under some special emulsion polymerization conditions, the particle size distribution of latex is very wide, and the particle size of larger and smaller particles differs largely.This size distribution of particles is called polydispersity.

Similar to the definition of particle size,there are also different possibilities to quantitatively describe the particle size distribution. The particle size distribution used depends on the definition of the particle size used, such as volume, weight, or number size distribution. Therefore,the same latex may and often produce different particle size distributions due to different measurement techniques. Of course, these different types of particle size distribution can be converted to each other by calculation because the relationship between them seems clear [22].

PSD measurement results are usually presented in tables or graphs, expressed in a differential or cumulative manner. The cumulative curve can give the percentile PSD parameters easy to be comprehended, such asD10, representing the 10% point in the cumulative undersize PSD. Generally, the data are normalized,and the fractional density of each size category is given in the form of a histogram or smooth curve in the differential size distribution.In the general link between product performance and particle size distribution [22], some form of weighted average diameter is recommended. Some typical examples are shown in Table 2. There are various descriptions of the average particle size in different references, but the calculation formula is essentially the same. For instance, the dynamic light scattering technique measures thezaverage reciprocal diameter, but it is often reported as ‘‘z-average diameter”.The symbolic representation of average particle diameter is also diverse. For example, the number-average diameter can also be represented as ＜D1,0＞, which is a mathematical symbol of the moment-ratio (M-R) system, and the M-R system is more suitable for processing the histogram type PSD than the International Organization for Standardization (ISO) system. Detailed related expertise can be found elsewhere [22].

Table 2 Definition of particle diameter averages.

There are several commonly used differential size distribution functions: the normal distribution (Gaussian, Eq. (1)), the lognormal distribution (Eq. (2)), and the Rosin-Rammler (Eq. (3)):

whereDis the particle diameter,Q(D)is the density function of particle diameter,σ is the population standard deviation,μ is the population means diameter,Q(lnD) is the density function of lnD,σgis the population geometric standard deviation, μgis the population geometric mean diameter,Q3(D) is the cumulative undersize volume (mass) distribution,Deis the reference size constant, andmis the distribution constant.

The complete particle size distribution of the emulsion polymerization system should be a three-dimensional graph, and the three coordinates are a normalized fraction,time,and particle size.The particle size distribution reported mostly in the literature is the particle size distribution at a certain moment(such as the final product).The width of PSD is usually expressed as the ratio ofD90/D10or (D90-D10)/D50.

2.3. Classification of PSD measurement methods

In the past two decades or so,some studies on the classification of PSD measurement methods have been published.Some methods always have fixed meaning and consensus. For example, off-line measurement requires manual sampling,dilution and other preparations before specific analysis and testing can be done. But, online, in-line, andin-situanalysis/measurements are often used interchangeably,and there is some terminological confusion.Some literature argued that the on-line measurement method needed sampling using a sampling loop [24] or by-pass route [25], while other literature reports pointed out that on-line analysis did not require sampling [26,27]. Besides, Danielset al. [28] argued that in-line and on-line analyses were used interchangeably under the condition that the reaction mass remained undisturbed all the time.They deemed that the on-line analysis meant no off-line samples and no by-pass, and another important reason was that the real on-line analysis should not disturb the reacting system, especially for laboratory scale investigations.

This article uses sampling or not and how to sample as the only criteria to classify particle size distribution measurement methods.The classification details are shown in Fig. 1. PSD measurement methods can be categorized into off-line, on-line and in-line. The requirements of each measurement method are shown in Table 3.The detailed definitions are as follows:

Table 3 Characteristics of PSD measurement options.

(1) The term ‘off-line’ is used here to refer to manual sampling,dilution, and analysis on a specific instrument. Results for off-line measurements normally exceed 60 min [25]. The at-line analysis is quite the same as the off-line analysis,but the analytical instrument is now in the neighborhood.

(2) The term ‘on-line’ is used here to refer to automatic sampling, dilution, and analysis at a specific time under computer control. It will be used either in a by-pass of a process/reactor, in which the process conditions have been adapted to the analysis,or samples are being taken continuously from the reactor and then analyzed in some automated way. The time between sampling and results presentation for on-line measurements is usually less than 10 min, which is much faster than off-line measurement.

(3) The term‘in-line’is used here to refer to automatic analysis without sampling.It provides continuous data from the process using the analytical probes, sensors, soft-sensor, endoscope, and so on, which construct a permanent direct connection with the process flow. In general, no sampling required for in-line measurement,and the local process conditions should not be changed by the analytical probe.In this paper, the terms ‘in-situreal-time’ and ‘in-line’ are used interchangeably.If ignoring the delay of sensors and control center, in-line and in-situ real-time measurement is realtime.

3. Off-line Measurement Methods

As mentioned above, off-line measurement methods require sampling and preparation (such as dilution) before using related instruments for measurement and statistics. Among the many reported particle size distribution measurement methods, the off-line method is the most commonly used. They mainly include(but certainly not limited to) light scattering technology (e.g.,dynamic light scattering, DLS), microscopy technology (e.g., scanning electron microscopy, SEM), and separation technology (e.g.,capillary hydrodynamic fractionation,CHDF)[29].This part mainly introduces the principles, methods, advantages and drawbacks of various off-line measurement technologies.

3.1. Light scattering measurement methods

Light scattering has been used to characterize latex particles and polymer solutions for many years[30-34].The light scattering measurement technology of PSD is based on the principle of light scattering, that is, when a light beam is an incident on a particle,the light beam will be absorbed and scattered by the particle,and then the outgoing light will deviate from the propagation direction of the incident light and scatter around[35,36].The types of light scattering can be divided into two categories:the first type is that there is no change in the wavelength of light before and after scattering, such as Mie scattering, which is mainly used in the development of laser particle size analyzers; the second type is that the wavelength changes, such as Raman scattering, which is mainly used in the field of molecular spectroscopy.Light scattering measurement methods include classic static light scattering(SLS)and dynamic light scattering(DLS).Using appropriate numerical methods to process experimental data, either DLS or SLS will provide PSD [37].

In a typical experimental setup,the photomultiplier tube detector that typically covers an angular range from 15°to 150°records the intensity of the scattered light and the photon counts. Moreover, how to handle these photon counts will determine whether to perform a DLS experiment or a SLS experiment [38]. It is worth noting that although both static light scattering and dynamic light scattering use the principle of light scattering technology to measure PSD, the two technologies have different emphases on light intensity.Static light scattering usually refers to a kind of measurement technique of the total intensity (or absolute intensity) of scattered light,but dynamic light scattering is mainly used to analyze the intensity fluctuation of scattered light rather than absolute scattered intensity. This technique is widely used mainly because the test is relatively fast, the equipment is cheap, and the calibration is not required compared to other off-line measurement methods.

3.1.1. Dynamic light scattering

DLS is also known as photon correlation spectroscopy (PCS) or quasi-elastic light scattering (QLS) [39]. The basic principles of DLS are the Mie scattering theory and Brownian motion. For asolid-liquid homogeneous dispersion system, the Brownian motion of small particles is more substantial, and the scattered light waves detected by the photoelectric detector are faster. The opposite is true for large particles.Therefore,the Brownian motion intensity of the particles can be obtained by measuring the fluctuation of the scattered light over time,and the particle size in a uniform dispersion system can be obtained indirectly.In general, DLS technology needs to use time-related correlation detection and particle size inversion algorithm to complete the calculation of particle size distribution. Correlation detection is to extract real features from complex noise-contaminated signals through autocorrelation or cross-correlation operations,and particle size inversion algorithm calculates the particle size from these laws.

In a typical experimental setup, the DLS measurement uses a helium-neon laser as the light source, and the detector position is either at a back angle of 173°or a right angle of 90°with the incident light. The dynamic light scattering experimental platform is shown in Fig. 3. A laser beam is used to hit the sample, and the scattered light intensity of a certain angle of the sample is collected.Then the time autocorrelation function of the scattered light intensity(Eq. (4))can be given through the autocorrelation operation based on the previously obtained scattered light intensity[36].The diffusion coefficient,DΓ, can then be identified and further used to calculate the hydrodynamic radius,D, through Eqs. (5)and (6).

Fig. 3. Schematic diagram of measuring PSD using dynamic light scattering.

In Eq. (4), τ is the delay time;qis the scattering vector, also called the optical constant,given by Eq.(5);DΓ is the translational diffusion coefficient of the particles in the suspension. In Eq. (5),nis the refractive index of the suspension, θ is the scattering angle,and λ0is the wavelength of the incident laser. In Eq. (6),KBis the Boltzmann’s constant,Tis the absolute temperature of the solution,η is the viscosity coefficient of the dispersion medium.

Eq.(4)represents the autocorrelation function of the monomodal distribution. For multimodal distribution,g(q,τ) is the sum of multiple particle swarms. The main difficulty, in this case, is to determine the contribution of each population to the overall signal.In addition,if the intensity caused by one population is lower than that scattered by other populations,it is not easy to detect the particle population.

Elizaldeet al.[40]used four different commercial techniques to measure the average particle size and particle size distribution of standard polystyrene latexes in the sub-micron range: DLS, CHDF,disk centrifuge photosedimentometry (DCP), and transmission electron microscope (TEM). The comparative analysis of the measurement results shows that DLS for monodisperse latex gives similar results to other measurement techniques, but DLS may not be reliable for measuring broad range or very polydisperse particle size distribution. Schneider and McKenna [29] established a round-robin evaluation of monomodal and bimodal latices using DLS(and/or multi-angle dynamic light scattering),SLS,SEM,CHDF,and field flow fractionation (FFF); and the comparative analysis of measurement results shows that most of the currently available techniques are sufficient to measure the average particle size of a monomodal latex and provide very similar PSD results;for bimodal latex, the results based on multi-angle dynamic light scattering(Zetasizer5000 in mode NNLS) seem to be the closest to the reference value. Other comparative researches [41-44] also show that DLS has good consistency with other measurement technologies.

Some scholars believed that the PSD measured by DLS was slightly different from other measurement techniques. Leeet al.[45] pointed out that the particle size distribution measured by SEM was wider than that measured by DLS. The main reason for this phenomenon is that the intensity of Rayleigh scattering light is proportional to the sixth power of the diameter of nanoparticles.As a result,the signal of larger particles is easy to be recorded,and the smaller population will be missed.It must be said that the particle size measured by DLS is larger than that measured by CHDF[46], and this difference is attributed to instrument calibration and expression uncertainty. The size of micelles estimated by DLS is also larger than that estimated by TEM or SEM [47-50]because the drying samples for SEM cause the particles to shrink,or there are several small aggregates of particles during DLS analysis.

With the continuous improvement of commercial particle size analyzers based on DLS, more and more often DLS is used alone to characterize the particle size distribution of latex. To name a few,Huanget al.[51]used DLS to measure the particle size distribution of PBA(poly(butyl acrylate))latex particles synthesized by batch polymerization and pre-emulsion semi-continuous polymerization.Kimet al.[52]used DLD and multi-photon confocal microscopy to measure the particle size distribution of PEG-b-PDVBP(PEG-b-P (diethyl-(4-vinylbenzyl) phosphate (DEVBP))-based submicron particles)prepared with different emulsion polymerization synthetic formulations. Besides, many reports that only used DLS to measure PSD can be found elsewhere [53,54].

Measurement instrument based on DLS can provide the following aspects: (1) particle size distribution curve; (2) harmoniczaverage particle diameter; (3) polydispersity index (PDI); (4) zeta point position.In order to facilitate comparison among techniques,it is usually necessary to normalize the particle size. For example,Farias-Cepedaet al.[55]used a normalized particle size to plot the distribution in order to compare the width of PSD. Some scholars believed that the PDI value below 0.2 was generally considered monomodal [45], but another believed that the PDI value lower than 0.1 and not more than 0.05 indicated that the final latex particles exhibited a unimodal size distribution [56]. Zeta potential is a very good indicator of the degree of interaction between latex particles and is usually used to evaluate the stability of latex systems [49].

It is also pointed out that DLS particle size analyzer has strict requirements on sample concentration. The theoretical model of dynamic light scattering is based on the single scattering light;that is, the scattered light received by the detector at a certain angle is that scattered from a single particle. Therefore, the tested latex sample needs to be diluted to make the sample concentration low enough.In general, the latex sample is diluted with deionized water under mechanical stirring to a latex concentration of 5 %(mass). In addition, DLS can be used to measure PSD as long as the sample is transparent.

At present, there are many commercial high-precision measuring instruments based on DLS. For example, the measurement range of Zetasizer Ultra is 0.3 nm to 15 μm.DLS provides accurate and repeatable data of 2 nm and above for most systems under test.As to the PSD measured for monodisperse latex,DLS has good consistency with other measurement techniques; multi-angle DLS can be used to measure PSD for bimodal or multimodal latex but need to be verified.In contrast,DLS may not be reliable for measuring broad range or very polydisperse particle size distribution.The autocorrelation function (Eq. (4)) is actually the inverse Laplace transform of the scattered light intensity signal in the time domain,so the PSD measured by DLS is not absolute.In order to ensure the accuracy of PSD measurement, PSD is generally measured by DLS and confirmed by electronic microscope images [57].

Today, DLS may be the cheapest, easiest, and most convenient commercial particle size analysis instrument in off-line measurement. However, some inherent shortcomings exist. Firstly, due to the high dependence of scattering light intensity on particle size,light scattering often misrepresents the true particle size dispersity.The signal of larger particles can dominate the signal of smaller size, and the smaller population will be missed. Secondly, the particle size distribution by DLS is not a direct measurement but extracted indirectly from the light intensity signal by complex mathematical calculation, and there must be statistical error and systematic deviation.In addition,the measurement of particle size distribution by DLS is usually time-consuming because of the preparation work such as sampling and dilution.

3.1.2. Static light scattering measurement methods

The basic process of static light scattering (SLS) technology to obtain PSD is to first measure the relationship between the scattered light intensity and the scattering angle, and then extract the particle size distribution information from the complex information through a numerical inversion algorithm. The SLS theory is the Mie scattering theory, and the detail can refer to Drake and Gordon [58]. The mathematical model of Mie scattering is a summation expression of infinite series, which cannot be solved accurately at present.Although there are many calculation methods for Mie scattering [59], every algorithm has a corresponding scope of application. When calculating the scattered light intensity, an approximation of the Mie theory can be used, namely Fraunhofer diffraction theory[60]and Rayleigh scattering theory[61].In general,the simplified calculation formula for scattered light intensity is simple in form and fast in the calculation.

In a typical SLS experiment, the sample is irradiated by a monochromatic beam with wavelength λ and then the detector records the scattered light intensityI(θ), which is a function of monitoring angle θ. In the relationship between scattering intensity and scattering angle,particles of each size have a unique scattering pattern,and the scattering of a group of particles is the sum of the respective effects of each particle:

whereAis a scale factor that depends on the geometry of the experimental device,Nirepresents the number of particles of a specific typei(radius,refractive index),Ii(θ)represents the scattering intensity of a single particle of typei, andNrepresents the number of particle types in the solution.

For the uniform spherical particles in a single scattering regime state, the SLS measurement,I(θ), is related to the PSD,f(D) [62]. If we know the refractive index of a particular diluted solution, the average particle size can be quickly calculated [29]. However, if we want to get the complete PSD, the calculation process will be very complicated mainly because of the inversion process, which needs to extract the required information from noisy experimental data [38]. The methods for this inversion are usually divided into two categories, namely analytical techniques and empirical methods [38]. Analysis techniques require solving integral equations,which are ill-conditioned and do not result in unique solutions.Therefore, analysis techniques need prior information, and its small errors may lead to large errors in the final results.The empirical method is based on computer numerical simulation, which handles different types of problems in different ways and the calculation accuracy depends on the algorithm. Inversion techniques such as Tikhonov regularization [63], CONTIN [64], maximum entropy method (MEM) [65,66] and the nonnegative leastsquares method (NNLS) [67], have been applied to obtain PSD.The details of these inversion techniques can be found in the above references.

Although SLS technology is widely used in the field of molecular spectroscopy, there are few reports on its specific application to the particle size distribution of latex prepared by emulsion polymerization.Some authors reported the combination of SLS technology and other measurement techniques to study particle size distribution. Finsyet al. [68] used the inversion technique to extract the particle size distribution from the information of the polystyrene latex dispersion measured by SLS, and compared it with the particle size distribution measured by DLS and electronic microscope. The comparative analysis of the measurement result shows that SLS can obtain a better resolution in particle size compared with PCS. Schneider and Mckenna [29] used SLS, DLS and SEM devices to measure the PSD of latex and conducted cyclic evaluations. An integrated system called automatic continuous online monitoring of polymerization reactions (ACOMP) includes an SLS particle size analyzer(e.g.,Malvern,Shimadzu)to measure particle size distribution [69]. Some scholars used SLS to measure particle diameters or hydrodynamic radius [70,71]. In the past ten years,few people reported the use of SLS for measuring the particle size distribution of latex, mainly because the SLS technology has a low resolution,and it is an indirect measurement in the fitting process.For particles with a diameter less than 100 nm, the measurement accuracy cannot be guaranteed. On the contrary, DLS becomes more and more popular, and the relevant report is increasing year by year.Some scholars argued that the technique based on SLS was not satisfactory in estimating PSD [29].

There are many reasons for the success of static light scattering measurement of particle size distribution: relatively short measurement time,high reproducibility,and wide measurement range.Besides,it can be used not only for solid-liquid systems but also for aerosols.However,SLS has inherent disadvantages.First of all,static light scattering is an optical fitting process,and the mathematical model is quite complicated. In this method, it is assumed that the measurement object is spherical, and non-spherical particles will not be provided with any information on sphericity. At the same time, it is quite noise sensitive.There are significant statistical errors in static light scattering, especially for very widely distributed populations [72].

3.2. Microscopy

The method of measuring the particle size distribution of latex based on microscopy is usually limited to scanning electron microscopy(SEM)and transmission electron microscopy(TEM).General optical microscopy (OM) does not have sufficient magnification and resolution,so they are not suitable for the observation of latex materials. At present, the resolution limit of mature industrial high-resolution optical microscopes (such as confocal optical microscopes) can reach a sub-micron level (≥100 nm), so they can replace electron microscopes for off-line PSD measurement within a certain measurement range. As for other microscopes such as atomic force microscopy(AFM)and confocal laser scanning microscopy(CLSM),they can also observe latex and record particle size distribution in general,but there are few reports in the literature for various reasons. The imaging principles of transmission electron microscopy and scanning electron microscopy are fundamentally different. TEM imaging is based on the fact that electron beams scatter with atoms in the sample when they pass through the sample,and these electron beams carry structural information inside the sample.The SEM imaging is based on the backscattered electrons and secondary electrons that excite the sample when the electron beam reaches the sample.The principle of electron microscopic imaging can refer to specific literature [73].

Electron microscopy is a well-known and mature technology for measuring the PSD of latex systems[40].The particle size distribution of poly(vinyl chloride)latex was measured by SEM[74].Other literature on the comparative study of SEM and DLS can be found elsewhere [42,45,57]. Liuet al. [75,76] used TEM to measure the PSD of latex particles. Other literature on the comparative study of TEM and DLS can be found elsewhere [41,47].

Based on the above,the steps to obtain the PSD of latex by SEM and TEM are summarized.The general steps of TEM for PSD are as follows[47,75,76]:first,the latex sample was directly diluted with deionized water to a suitable concentration;then the diluted latex was transferred using a pipette to a copper mesh covered with a supporting film such as a carbon film.If the support film has insufficient strength and rigidity, it is necessary to plate a reinforcing film of carbon on the support film to improve the strength and heat resistance. Then comes the drying process, which can be room temperature drying or freeze drying. Finally, the sample image was taken by TEM,and the TEM micrograph was analyzed by ImageJ software to obtain PSD of the latex.The software can automatically detect the particle size, but it needs to manually adjust the error detection or missing particles [45]. It should be noted that the number of particles detected should be at least statistically significant.

For TEM, the electron beam passes through the sample, so the thickness of the sample should not exceed 100 nm. However,SEM sample preparation is not as demanding as TEM sample preparation.SEM samples can be tested as long as they do not contain water. For a latex sample that has been diluted, the sample needs to be dropped on a specific substrate, then dried, and sprayed with carbon or gold for SEM observation. If the sample itself is conductive, carbon spraying is not necessary. The image processing is the same as that for TEM.

Image processing technology based on computer vision can be used to process electron microscope images to obtain particle size distribution. It will be time-consuming and painstaking to implement similar output manually.Momotaet al.[77]reported a digital image processing technique for measuring the size and size distribution of round particles. At present, the most commonly used image processing tool is commercial image processing software(such as Image Pro, National Instruments-Vision Builder, and EPIX-XCAP) for particle morphological feature extraction and quantification. The above commercial software cannot directly obtain particle size distribution from electron microscopy images.An attractive solution is the development of the ImageJ plug-in,which is a machine vision application for calculating particle size distribution. ImageJ is an open-source image processing and analysis program based on the Java programming language.In order to facilitate the matching of area and circumference,ImageJ maps the actual particles to the equivalent ellipse, so the measurement results will have a little deviation compared with the actual size.After determining the particle shape,the ellipse size should be corrected with an appropriate correction factor for accurate particle size measurement [78].

An obvious advantage of electron microscopy is that it may be the most accurate technique for statistical particle size distribution if a sufficient number of representative particles are counted [79].To a certain extent,we can think that the particle size distribution observed and recorded by the electron microscope is the absolute values because they are statistics and analysis through the images of actual particles. Therefore, this technique was often used as a reference.

Electron microscopy has two unavoidable disadvantages.Firstly, electron microscopy technology is expensive and timeconsuming. Due to the cost and time required to analyze samples,it is basically impossible to analyze a large number of samples in a short period of time by using SEM or TEM. Secondly, the sample preparation of soft polymer latex, especially the TEM sample, is still very difficult. The preparation of samples for electron microscopy analysis not only needs time-consuming preparation steps such as dilution, drying, and fixation but also takes into account the influence of environmental factors in the preparation process on the properties of latex particles.For example,some copolymers exhibit low glass transition temperature, and due to this and the low liquid fluidity of polymers at room temperature, SEM is not suitable to characterize the particle size distribution [43,80].

3.3. Separative techniques

The measurement of PSD based on separation technology is generally completed in two steps.First,the particles are separated from each other using inertia-based methods; then, the particle size distribution is measured. This section introduces three standard separation techniques, DCP, CHDF, and FFF.

3.3.1. Capillary hydrodynamic fractionation

CHDF is a simple technology that is often used for estimating the PSD of latex based on the separation of particles in a long capillary tube [81,82], and detailed information on how CHDF works and the theory of particle size detection can be found elsewhere[83,84].A highly diluted sample is injected into a long tube(capillary) where a parabolic velocity profile develops under laminar flow conditions,and the elution rate of large particles is faster than that of small particles due to the flow characteristics, the largest particles therefore leave the column first. The schematic diagram of the separation principle of latex particles is shown in Fig. 4.Then,the number of particles in each elution fraction is estimated using a turbidity detector[84].Finally,after complex mathematical processing,the PSD of the latex can be obtained.The measurement range of CHDF is 5 nm to 3 μm[85].It is worth noting that the electric double layer effects and van der Waals attraction are also influencing factors of the separation process.

Fig. 4. The separation principle of capillary hydrodynamic fractionation.

In the past twenty years or so, some researches on PSD measurement with CHDF technology have been published. Leeet al.[86] used CHDF to measure the particle size distribution of the doped polypyrrole nanocomplexes. Choi and Lee [87] determined the PSD of latex prepared by the emulsion polymerization of vinyl acetate monomer,ethylene,andN-methylol acrylamide in the final emulsion by using CHDF.These also include a study by Zeaiteret al.[88] that used CHDF to determine the particle size distribution of the final product of a laboratory scale styrene polymerization,and this off-line measurement was incorporated into the control strategy of the on-line control of the PSD from an emulsion polymerization reactor. Other examples that used CHDF to determine the PSD of latex produced by emulsion polymerization can be found elsewhere [89-91].

The main advantage of CHDF is a simple operation and high efficiency for identifying multiple modes [84]. However, CHDF also has some disadvantages: (1) the typical analysis time of CHDF is about 10-20 min [9,88]; (2) CHDF must calculate the calibration curve using a standard of known particle size; (3) the particle refractive index(PRI)and instrumental broadening (IB)may influence the final result [84].

3.3.2. Field flow fractionation

Strictly speaking, FFF is a kind of separation technology based on chromatographic-like principles [92]. It contains many subcategories such as:flow FFF,asymmetrical flow field-flow fractionation (AF4) [93], hollow-fiber flow field-flow fractionation (HF5)[94], sedimentation field-flow fractionation (SdFFF) [92], gravitational field-flow fractionation (GrFFF) [95] and so on. Flow FFF is a kind of separation technique based on the force field, which is perpendicular to the direction of the main flow (laminar flow) in a narrow channel[96].And the forcing flow field is usually realized by a cross flow stream,which is used to trap the particles injected into the channel.The larger particles are then pushed closer to the wall, where the velocity is lower than the centerline velocity.Therefore,the larger particles leave the channel behind the smaller particles.Detailed information on how FFF works and the theory of particle size detection can be found elsewhere[97].The schematic diagram of the separation principle of latex particles is shown in Fig. 5. The particles of latex can be measured and recorded using detectors placed at the end of the channel as each particle leaves the device. The measurement range of FFF is 1 nm to 50 μm [96].

Fig. 5. The separation principle of field flow fractionation.

There are many reports that the PSD of latex prepared by emulsion polymerization can be determined using the FFF technique.Hanet al. [98] determined the particle size distribution of silica nanoparticles synthesized by emulsion polymerization by using AF4, and the results were compared with those of DLS and TEM.The results from TEM and AF4 are in good agreement.Other examples that used AF4 to determine the PSD of latex produced by emulsion polymerization can be found elsewhere [92,99-102].Xiaoet al. [103] developed a method for characterizing the PSD of poly(vinyl acetate) nanoparticles by combining HF5 with a UV-vis detection, and the results were in good agreement with those obtained by TEM.The size distribution of silica nanoparticles synthesized by emulsion polymerization was determined using SdFFF and DLS by Leeet al. [104]. Makanet al.[92] used SdFFF to investigate the evolution of particle sizes and particle size distributions as the polymerization progressed. Moreover, the emulsion polymerized polystyrene was analyzed using GrFFF to obtain the PSD by Parket al. [95].

Because FFF is a large category of separation techniques, this article cannot summarize the advantages and disadvantages of all of them. The main advantages of FFF are simple instrumentation,convenient operation,and low operating cost.Similarly,their analysis time is also significantly different.For example,the typical analysis time of flow-FFF is 20-30 min[9],while the analysis time of GrFFF is less than 10 min [95].

3.3.3. Disc centrifuge photosedimentometer

DCP is a technology that separates particles radially along the entire rotating disk based on their size and density caused by centrifugal force,and detailed information on the DCP procedures and its theory can be found elsewhere[105,106].The sample enters the DCP from near the center axis of centrifuge chamber that is defined by a few parallel concentric transparent disks several millimeters apart, and the particles of latex will move to the end of the outer edge at different speeds depending on their sizes [29], then the detector placed somewhere near the disk records the particle size and number according to the transmitted light technique when the particles pass. The separation of particles in the centrifugal force field is based on Stokes law [40]:

whereDis the particle size of the latex,kis a constant,η is the viscosity of the spin fluid,Rdis the radius where the detector is located,Rmis the radius of the meniscus where the sample enters,Ω is the rotational speed,Δρ is the relative density (the difference between the particle and the solvent),andtis the time required for the particle to reach the detector or the sedimentation time. The particle size distribution is derived from the detector response using the Lorenz Mie theory, and the information about this theory can be found in the previous Section 3.1.1.It is worth noting that the density of the polymer must be greater than the density of the solution to achieve separation.

The particle size and its distributions of polyacid-stabilized latexes were obtained by using DLS,DCP,and SEM analysis by Yang and Armeset al.[107].Fieldinget al.[108]reported a method that a corrected DCP technique was used to determine the PSD of polystyrene/silica nanocomposite particles,which possess a core-shell morphology that affects the accuracy of DCP;and the PSD obtained by the corrected DCP technique shows that it is consistent with other measurement techniques such as TEM data. Other examples that used DCP to obtain the weight-average PSD of latex produced by emulsion polymerization can be found elsewhere [109-111].

The analysis time of DCP depends on the sample;smaller particles require longer centrifugation time,and a multimodal distribution requires even longer time[40].Therefore,DCP is very effective for large particles and unimodal distribution.

It should be noted that CHDF, FFF and DCP theoretically need detectors (such as UV-vis detector), which measure and record the particle size profile based on optical principles. The optical principle of these detectors is similar to that of the light scattering technique,so they are subject to the same limitations and difficulties,especially when confronted with a wide range of latex or multimodal PSD [29]. For polydisperse systems, they can be divided into several monodisperse groups by using separation technology,and then DLS technology can be used for accurate measurement.

3.4. Comparison among off-line measurements

Among all PSD measurement methods,off-line measurement is the most widely used technology,but it cannot reflect the particle size distribution of latex in the reactor in real-time. Therefore, the long detection time is an inevitable limitation of off-line measurement,and its practical application is greatly restricted.The analysis time of off-line measuring equipment only takes a few minutes or less,but its preparation(for example,sampling,dilution,mixing)is very time-consuming. The comparison of off-line measurement methods is shown in Table 4.

Table 4 Comparative analysis of off-line measurement methods.

Besides,some practical guidelines of off-line methods are given to help researchers in the field to use these measurement methods effectively and avoid common mistakes in measuring PSD. This includes, for instance (this is a non-exhaustive list):

(1) Dynamic light scattering measurement technology is based on the principle of single scattering, so the DLS measuring instrument has special requirements for the concentration of the test sample. In general, the latex sample is diluted with deionized water under mechanical stirring to a solids concentration of 5% (mass).

(2) DLS technology is not suitable for measuring a wide range or very polydisperse particle size distribution. In this case, try to use the multi-angle dynamic light scattering measurement technology, or first use the separation technology to separate the samples, and then perform DLS measurement for each separated sample, and finally perform statistics.

(3) Electron microscope measurement of PSD is the only absolute measurement technique, though the sample preparation process is complicated and the test cost is expensive.In addition, when using image processing software to analyze electron microscope photos, the number of particles detected must be high enough.

(4) The light scattering technology uses fitting and Laplace inverse transformation when calculating PSD, so the measurement result should be compared with the measurement result of the electron microscope measurement technique to verify the accuracy of the result when it is necessary. Similarly,the results of other measurement methods should also be verified using electron microscopy techniques.

(5) CHDF is essentially a separation technique, and it must use the standard substance with a known particle size to create a calibration curve.

4. On-line and In-line Measurement Methods

Unlike the off-line measurement, which only needs a single analyzer, the on-line and in-line measurement methods usually need a set of devices to form the analysis system. For example,for the on-line measurement, the most basic device consists of the automatic sampler, auto-diluter, auto-sampler and analysis equipment. This section will outline the latest on-line and in-line technologies and highlight areas where progress needs to be made,either reach a consensus on the conflict of views or just deepen our current understanding. Besides, the difficulties and challenges of these so-called in-line methods are discussed in detail.

4.1. On-line measurement methods

In order to reduce the measurement time, some scholars reported that off-line measurement instruments were connected to the reactor to complete automatic sampling, dilution and other preparations. This method is called an on-line measurement method. This section mainly summarizes some reported on-line measurement systems and their advantages and disadvantages.

Craver and Provder[112]reported a system for determining the particle size distribution of latex prepared by emulsion polymerization: DLS is connected to a reactor with automatic sample acquisition and dilution steps. The system controlled by a computer consists of three main parts: sampler-prediluter, autodiluter and DLS instrument. The sampling cycle begins by capturing a small amount of concentrated sample from the latex reactor.After pre-dilution, the partially diluted sample is sent to the automatic diluter, where the dilution continues until the sample concentration meets the requirements of the instrument. Then, the DLS instrument is used to analyze the diluted sample. The computer controller then waits for the start of the next measurement cycle.A cycle time of 15 min or less is feasible,and this on-line test time is much shorter than off-line measurements. The schematic diagram of this system is shown in Fig. 6(a).

There are several reports on on-line systems based on CHDF to determine the PSD of latex produced by emulsion polymerization.Liottaet al.[113]built an automated reactor control facility with a sampling loop that took a sample every ten minutes and then transported the sample to the on-line particle size sensor (CHDF)to measure the PSD of polystyrene particle. Parket al. [114] set up an on-line measurement platform, which used a diaphragm pump to circulate the material through a circulation pipeline,where a portion of the reactants was diverted to CHDF to measure the PSD of latex prepared by the semibatch vinyl acetate/butyl acrylate emulsion copolymerization system.It is worth noting that the emulsion polymerization reaction generally requires stirring,the use of inert gas to ensure the stability of the reaction environment and controlling the reaction temperature.So many restrictive conditions and sensor monitoring need to be controlled by a computer. The schematic diagram of this system is shown in Fig. 6(b).

Fig. 6. (a) On-line with DLS [112]; (b) On-line with CHDF [113]; (c)ACOMP [69].

A different on-line measurement system called automatic continuous on-line monitoring of polymerization reactions (ACOMP),including ‘‘the Front-End” and ‘‘the Detector Train” subsystem,was established by Albet al. to measure PSD [69,115]. The frontend is the heart of the ACOMP system,and it includes the ensemble of extraction devices(sampler),pumps,mixing chambers,and conditioning steps. The ACOMP Front-End automatically takes a high concentration of latex from the reactor, dilutes it to the right concentration,and then sends it to a detector to determine the particle size distribution. Any number of detectors can be integrated into the ACOMP Detector Train, and the samples conditioned from the ACOMP Front-End are analyzed to determine the PSD of latex by using light scattering technology(DLS and SLS).The schematic diagram of this system is shown in Fig. 6(c).

It can be seen from the above literature that this on-line measurement method is essentially an off-line measurement technology, so the on-line measurement method inevitably has the inherent shortcomings of the off-line measurement methods. The analysis instruments usually need to be connected by a hose,wherein the sampled residue remains for the last cycle measurement will affect the accuracy of the next measurement.Therefore,cleaning for assuring a representative sample being taken is essential for on-line measurement accuracy. The advantages of the online measurement system include its general purpose applicability:it can provide rich data from multiple independent PSD detectors for comparative analysis,and it can provide various characteristics of polymerization reactions.The mechanical complexity of preparing the sample often consumes most of the entire analysis time,which is inevitable, but the on-line analysis method does shorten the measurement time compared to off-line measurement.

4.2. In-line measurement methods

Another solution to solve the time-consuming problem of the off-line measurement is to directly use optical principles to monitor the emulsion polymerization process in the reactor to avoid tedious sample preparation such as sampling. As mentioned earlier, this article uses the term ’in-line’ to refer to this method,and the in-line measurements provide continuous data from the reactor of emulsion polymerization using the analytical probes,sensors or soft-sensor, which constitute a permanent direct connection with the process flow. It is worth emphasizing that no sampling is performed when using the in-line method to measure the PSD during emulsion polymerization. However, this measurement method has many limitations to be solved, such as the limit on the concentration of emulsion polymerization particles.At present, there are only a few reports that have tentatively explored this method for specific emulsion polymerization systems, and no mature commercial application instruments are now available.

Chemtobet al. [116] reported an in-line measurement to monitor the particle size evolution of the mini-emulsion photopolymerization throughout UV irradiation by using DLS. For these measurements, the very small reactor was inserted into the DLS cell holder without a thermal cap,and the Xe-Hg lamp was placed vertically 44 mm from the sample surface, and then DLS analysis and UV illumination were triggered simultaneously for measurement. The object of measurement is a photo-induced emulsion polymerization system, not a thermally initiated emulsion polymerization system, so photo-initiator and light are indispensable.The photo-induced emulsion polymerization was carried out in a sealed quartz cell without nitrogen bubbling and stirring, and the volume of the sample was 10 mm × 10 mm × 25 mm. It is noted that this in-line measurement method does not require sampling,mainly because the organic phase concentration is only 5% (mass)and does not affect the backscatter detection (there is no multiple scattering problem). Besides, the time resolution of this in-line measurement method runs every 10 s, which is much faster than that of off-line and on-line. However, this experimental polymerization system is not a conventional thermally initiated emulsion polymerization system. In addition, the concentration of the organic phase and the solid content of the final product of this so-called mini-emulsion photopolymerization system is required to be less than or equal to 5% (mass), which obviously does not meet the requirements of industrial production.

Another worthwhile study is to measure the droplet size distribution in emulsion polymerization. Hadziret al. [117] used a specially designed dynamic light scattering instrument to measure droplet/particle size in-line under CO2atmosphere. This special DLS consists of a high-pressure syringe pump, a magnetic stirrer,a remote DLS probe with a single photon detector, a data processing linear correlator, a backscattering detector, and an external sample holder (serving as an emulsion polymerization reactor).The in-line measurement is carried out by focusing the laser beam on the sample center inside the high-pressure reactor without stirring under the specified pressure and temperature. The schematic diagram of this system is shown in Fig. 7. However, this in-line measurement method is not used to measure PSD but to measure the droplet size distributions of emulsion.

Fig. 7. Schematic diagram of experimental device for online measurement of droplet size distribution rather than PSD [117].

It is worth noting that the above two experimental cases do not realize in-line measurement in a real sense during a thermally initiated emulsion polymerization system,but they provide ideas and directions for exploring in-line measurement.The in-line measurement method does not sample but directly connects with the reactor for PSD real-time measurement and has advantages and disadvantages. The main advantage of the in-line measurement method is to further save measurement time compared to offline and on-line ones. The disadvantages of in-line measurement based on light scattering include special requirements for samples(low solid content, high surfactant concentration); meanwhile, no mature commercial test equipment is available.

At present,it is not feasible to determine the particle size distribution of the latex produced by emulsion polymerization using inline technology [118]. The main reason is that the existing instruments, lacking sufficient measurement accuracy, cannot process overlapping signals from different particle phases in a highly concentrated multiphase system or cannot capture images of moving particles. For average particle size measurements, a few in-line technologies are available, such as electrical impedance spectroscopy [119], near infrared spectroscopy [120], Raman spectroscopy [118], and multi-angle laser light scattering (MALLS)[121]. However, this review mainly focuses on the PSD measurement methods rather than average particle size measurement.

A soft sensor can be used to estimate PSD based on the relevant spectral data,and soft sensors include mathematical and statistical programs as well as physical and chemical models that use available sensor data to calculate target characteristics that cannot even be directly measured[122].The soft sensor is also called a process model [123]. The sensors can be analytical equipment such as an in-line spectrometer or non-analytical sensors such as temperature and pressure. Gomes [124] proposed an effective control framework that incorporates a model-based algorithm with software engines for in-line feedback of properties, and this algorithm is a verified dynamic model that can be used as a soft sensor. The developed system can be used to control conversion rate, molar mass,polymer composition,particle size,and particle size distribution of the latex prepared by emulsion polymerization using model predictive control (MPC). After the experimental test, it is found that the developed control scheme can achieve the expected goal well. Some scholars believed that the basic measurement in inline polymerization monitoring and control aimed to avoid empirical models and inference models in data interpretation[69].Thus,whether the soft sensor can be used as an in-line measurement tool is a question worthy of in-depth exploration.

4.3. Comparison analysis

The purpose of developing and improving on-line and in-line measurement methods is to shorten the measurement time for particle size distribution and even realize real-time monitoring and control of the growth process of polymer particles.On-line and inline measurement methods eliminate the time-consuming manual sample preparation procedures compared with off-line measurement methods.The on-line measurement method can be regarded as an automated off-line measurement method in a sense so as to eliminate the time-consuming sample preparation steps. At present, there are only a few reports on a tentative exploration of the in-line measurement for specific emulsion polymerization systems.Through comparative assessment of related works,one finding is that the in-line measurement of particle size distribution during the process of emulsion polymerization is not a ‘‘finished”measurement technology.

Different from the off-line measurement, which only needs a single analyzer,the on-line and in-line measurement methods usually need a set of devices to form the analysis system.The sources of error are more extensive,so there are some practical guidelines for on-line and in-line measurement methods.

(1) In the off-line and on-line measurement methods, sampling or dispersion is the main error source rather than the measurement of particle size distribution itself, so please fully master the relevant knowledge and skills before conducting the measurement experiment. For example, low temperature treatment is often carried out after sampling to prevent further polymerization.

(2) In the off-line and on-line measurement methods, the sampling operation should not interfere with the emulsion polymerization system,especially for the laboratory scale study.Researchers need to deal with the relationship between sampling volume and effective reactor volume.

(3) For on-line measurement,one sampling can be used for multiple instrument analysis,but the sampling hose needs to be cleaned and flushed before each sampling to avoid the influence of the previous sample analysis.

(4) It is currently not feasible to determine the particle size distribution of the latex produced by conventional emulsion polymerization using an in-line technology. Although some scholars think that in-line measurement should avoid empirical models, soft sensor technology is a promising inline measurement technology. So whether the soft-sensor can be used as an in-line measurement tool is a question worthy of in-depth exploration.

5. Conclusions and Outlook

A comprehensive overview of the particle size distribution measurement methods in the process of emulsion polymerization has been deliberated, including the existing off-line, on-line and inline methods.Among all PSD measurement methods,off-line technology is the most widely used despite its time-consuming analysis preparation, such as sampling, but it cannot reflect the particle size distribution of latex in real-time. Electron microscope measurement as a typical off-line measurement method is quite expensive, but its measurement results are absolute and accurate, so it can be used as a reference standard. At present, the resolution of mature industrial optical microscopes (such as a confocal optical microscope) can reach the submicron scale (≥100 nm), which may replace the electron microscope for off-line PSD measurement in a specific range of measurement.The light scattering technology represented by DLS and SLS is a relatively convenient technology with the shortest measurement time among off-line measurement methods, but it is not suitable for measuring a wide range or strongly polydisperse particle size distribution. Separation technology is an off-line measurement method with relatively simple operation and suitable for almost all polydisperse systems, but some separation measurement techniques must use the standard substance with known particle size to get the calibration curve.For polydisperse systems, they can be first divided into several monodisperse groups using separation technology, and then DLS technology can be used for accurate measurement.

On-line measurement technology only links off-line measurement equipment with the reactor to complete automatic sampling,preparation and analysis, which saves time. But, it still cannot avoid sampling and other preparatory steps.The real on-line analysis should not disturb the reacting system, especially for laboratory scale investigations. For example, sampling should try to avoid interfering with the original reaction state. Although inline measurement technology does not require sampling, it uses sensors or soft sensors containing mathematical models to predict PSD. However, the basic measurement for in-line monitoring the evolution of particle size distribution aims to avoid empirical models and inference models in data interpretation.When researchers need to understand the precise particle size distribution, technology that takes a few minutes will not provide users with actual images in the reactor.Therefore,the in-line measurement technology, which does not need sampling and empirical model, is very important. Nevertheless, it is currently not feasible to determine the particle size distribution of the latex produced by emulsion polymerization using the up-to-date in-line technology. The main reason is that the existing instruments lack measurement accuracy, and they cannot process overlapping signals from different particle phases in a highly concentrated multiphase system or cannot capture clear images of moving particles.

Since the in-line measurement methods of PSD can reflect the change of particle size distribution in real-time during the emulsion polymerization process, it is a major stumbling block for the understanding of emulsion polymerization mechanisms (such as particle nucleation)and the development of mathematical models.Therefore,this review attempts to give several theoretical possible solutions of in-line PSD measurement for reference.

(1) The first solution worth trying to measure PSD in-line is to use optical principles directly, such as light scattering technology. The laser light emitted by a light source directly interacts with the polymer particles in the stirred reactor,then the detector receives the optical signal after the action and completes the photoelectric conversion, and finally a specific algorithm is used to analyze the photoelectric signals in order to obtain the particle size distribution.The difficulty is that the principle of light scattering technology is based on single scattering theory, so there are special requirements for particle concentration. If using this technology to monitor the particle size distribution of polymer particles in real-time, it is necessary to modify the reaction formulation to reduce the concentration of polymer particles in order to eliminate overlapping signals from different particles.

(2) Another solution that can be tried is to use an optical microscope to directly image the latex in the reactor and shoot with a high-speed camera,then use image analysis technology for real-time analysis to realize the real-time monitoring of the evolution of particle size distribution. Because the sample cannot contain water during electron microscope analysis, in-line measurement is basically impossible by using electron microscopes. High-resolution optical microscopes, such as confocal microscopes, have fewer requirements for samples than electron microscopes, are possibly hopeful for in-line measurement. This measurement scheme’s difficulty is whether the high-speed camera can quickly capture the high-speed moving nano-scale polymer particles. Simultaneously, the limitation of this scheme is that it can only monitor the polymer particles in the focal plane in real-time and have high requirements (e.g., high light transmittance) on the reactor.

(3) Although some scholars think that in-line measurement should avoid empirical models, the soft sensor technology is a promising in-line measurement one. However, this approach’s difficulty is that the model cannot be verified due to the lack of accurate in-line measurement equipment.A feasible method is to thoroughly study an emulsion polymerization reaction system to obtain enough PSD data,then combine it with big data or artificial intelligence technology to predict or calculate the particle size distribution under new working conditions.

Acknowledgements

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The National Key Research and Development Program(2020YFA0906804), the National Natural Science Foundation of China (22078325, 22035007, 91934301), the NSFC-EU project(31961133018)and the Special Project of Strategic Leading Science and Technology,CAS(XDC06010302)are gratefully acknowledged.