1 Introduction

Polymer latexes are waterborne polymer dispersions that are used to formulate all kind of paints, coatings, adhesives, and a myriad of additives for a wide range of applications.^[1-3] Particle size and particle size distribution (PSD) are key characteristics of polymer latexes,^[3] as they affect the rheology (e.g., viscosity) of the dispersion and its film formation, among others, which are important to control the process and the application properties of the latexes.^[3] Therefore, a robust and efficient synthetic process should ensure the reproducibility of these characteristics during the process to avoid off-spec products and economic loses. Hence, real-time monitoring of the particle size is necessary to be able to implement real-time (close-loop) control strategies. This is why a lot of effort has been devoted to develop real-time online or inline automatic particle size characterization techniques.^{[3, 4]}

Particle size characterization techniques can be clustered in three main groups: scattering techniques (e.g., dynamic light scattering (DLS) and static light scattering (SLS), respectively), laser diffraction, photon density wave spectroscopy (PDWS) or turbidity spectroscopy (TUS)); fractionation techniques (e.g., capillary hydrodynamic fractionation (CHDF), analytical centrifugation or field flow fractionation); and counting techniques (e.g., transmission electron microscopy (TEM), particle tracking analysis, optofluidic force induction (OF2i), or light obscuration). Most of them are offline analysis techniques, what means not only that samples need to be withdrawn and manually prepared for the analysis, but also that the time delay to get the results is large, going from few minutes (DLS case) to some hours (TEM case). Thus, online/inline particle size characterization techniques can be used to overcome these limitations, as time delays are substantially reduced and manual preparations are avoided.

Few technologies based on light scattering, ultrasonic properties, NIR of Raman spectroscopy, focus beam reflectance, or optofluidic force induction have been used to determine particle sizes in real time, but not all of them are able to reach the sub-micrometer particle size range.^[4-14] Moreover, many of them work in very diluted media; thus, there is a need to automatically dilute the sample before the analysis (online analysis), which cannot be acceptable for certain processes, due to the clogging that could occur inside the dilution system. Furthermore, the high dilutions needed could lead to large time delays between sampling and acquisition of the results.^[3]

The online TUS and the inline PDWS are two of the available real-time technologies to monitor the particle size of waterborne polymeric dispersions. Both are scattering-based techniques and have been previously assessed to online and inline monitor the particle size evolution during emulsion polymerizations.^[15-23] However, we are not aware that both techniques have ever been used simultaneously to monitor emulsion polymerization process and their performance compared with a common reference technique like DLS. Thus, in this work, both techniques will be used as online/inline technologies (TUS and PDWS, respectively) to monitor seeded semibatch emulsion polymerization processes of acrylic copolymers.

In the present study, a commercial calorimetric reactor is equipped with both spectroscopic technologies including an automatic dilution system for the turbidimetric technique. Thus, simultaneous real-time particle size measurements are gathered during the seeded semibatch emulsion copolymerization of methyl methacrylate (MMA)/n-butyl acrylate (BA)/methacrylic acid (MAA) reaction. In addition, the obtained results are compared to particle sizes obtained offline by DLS measurements. First, a brief description of the measurement principles of both techniques is given. Then, the novel reaction set-up and the experimental details of the emulsion polymerization carried out are presented, followed by the monitoring of the experiment and the comparison and discussion of the results.

2 Experimental Section

2.3 Description of the Set-Up and the Analysis Equipment

A 1 L calorimetry reactor (RC1e) from Mettler Toledo was used to perform all the reactions, equipped with a jacketed glass reactor fitted with the following fittings: a sampling device to withdraw the samples for TUS measurements, the inline PDWS probe, a nitrogen inlet, feeding inlets, a Pt-100 probe, and a stainless steel anchor type stirrer. iControl RC1e Software was used to control the reaction temperature and the inlet flow rates of semicontinuous feeds. Samples were withdrawn during the reaction to analyze the evolution of particle size offline by DLS (every 10–15 min during the first 60 min of the reaction and every 20–30 min after that time).

A PDWS equipment developed by PDW Analytics GmbH (Potsdam, Germany) was used for the particle size monitoring.^{[15, 22, 33, 34]} The PDWS was equipped with four lasers of 636, 778, 855, and 973 nm wavelengths, 27 possible distances from 3 to 25 mm, and 41 possible frequencies from 10 to 610 MHz. To calculate the particle size, a software developed by innoFSPEC Potsdam from the University of Potsdam was used, based on Mie theory and the theory of dependent scattering.

TUS equipment developed by IRIS Technology Solutions (Barcelona, Spain) was used, which consists of a continuous spectrometer and is composed of a probe, an electrical cabinet, and a computer with a dedicated software for data processing. As mentioned above, turbidity should be measured at conditions with no significant multiple scattering, what means that samples withdrawn from the emulsion polymerization experiments need to be diluted.^[18-21] The dilution system, developed by IRIS in collaboration with POLYMAT, used for this work is a dilution system with an automatic decision making software to select the appropriate dilution level by analyzing the transmission curves. A sketch of the setup and an image of the actual lab setup used for the simultaneous analysis are shown in Figure 1.

A Zetasizer Nano Series (Malvern Instrument) DLS was used as reference technique in this work. For the analysis, the latex was diluted in DI water until an attenuation number of 6–7 was achieved at 4.65 mm position to ensure single scattering conditions. DLS measurements were carried out at room temperature, and three repeated measurements were carried out for each sample. All reference particle sizes shown in this work were both intensity- and number-average particle sizes measured by DLS.

3 Results and Discussion

A seeded semibatch emulsion copolymerization (composition MMA/BA/MAA: 51/47/2 wt%) reaction (SB1) was used to simultaneously assess the performance of both TUS and PDWS technologies to monitor particle size growth. Reaction SB1 aims to produce a latex with an average particle size of 225 nm and a solids’ content of 45 wt% starting from a seed (S1, see the Supporting Information) with a particle size of 67 nm (intensity-average size) and 3% of solids’ content. Figure 2 presents the transmission curves obtained for the samples online analyzed by TUS during SB1 emulsion polymerization reaction.

From the transmission curves presented in Figure 2, the wavelength exponent was calculated as explained in the previous section, getting to the following results (Figure 3).

Figure 3 shows that the wavelength exponent decreases during the reaction, which, according to the calibration curve used, predicts an increase in the particle size, as it is expected in this seeded semibatch emulsion polymerization. From the wavelength exponent values presented in Figure 3 and by means of the calibration curve explained above (which can also be found in Figure SI1 in the Supporting Information), the particle size was retrieved (see Figure 5).

Figure 4 presents the time evolution of the reduced scattering coefficient obtained by inline PDWS analysis during SB1 emulsion polymerization reaction.

Figure 4 shows that the reduced scattering coefficient increases during the synthesis for all the laser wavelengths. At the beginning of the reaction, the solids’ content is very low; thus, the multiple scattering conditions are not fully accomplished, and the changes in the reduced scattering coefficient can hardly be detected, but after 20 min, the sample is turbid enough (both particle size and solids’ content increased) and the entire process can be monitored. Moreover, it can be seen that for the analyzed particle size range, the smaller the wavelength of the laser light, the higher the sensitivity to the changes occurring in the reactor.^[22]

The particle size evolution was retrieved, from the above presented wavelength exponent (Figure 3) and reduced scattering coefficient evolutions (Figure 4), using the wavelength exponent method and the radiation transport theory, Mie theory, and dependent scattering approaches, respectively, as discussed in further detail in other works.^{[16, 22, 23, 25, 30]} The results obtained by the simultaneous analysis and their comparison to offline DLS reference measurements are displayed in Figure 5.

All the techniques, real-time online TUS and inline PDWS and offline DLS, captured the growth of the particle size with reaction time, which matches the theoretical evolution of the particle growth in the absence of secondary nucleation or aggregations (see Figure SI3 in the Supporting Information). However, there are some differences that need a correct interpretation of the plotted data. Note that for the offline DLS analysis two averages are provided; the intensity-average, which is the natural average obtained from the DLS and the number average that is calculated from the cumulant method assuming a unimodal Gaussian distribution for the dispersed particle sizes (DLS particle size and PDI values can be found in Table SI3 in the Supporting Information). Interestingly, the evolution of the particle sizes calculated from the real-time spectroscopic techniques is closer to different average values; TUS particle sizes are in reasonable agreement with intensity-average sizes of DLS, while PDW particle sizes are closer to the number-average sizes of DLS. The matching of the TUS particle size and intensity-average particles sizes of DLS was expected, because the calibration curve that relates the wavelength exponent and particle size (see Figure SI1 in the Supporting Information) was built using intensity-average particle sizes measured by DLS. For the PDWS, there is no calibration curve to retrieve the particle size. The particle size is obtained from the theoretical dependence of the reduced scattering with the particle size as explained elsewhere.^{[30, 35]} The only input parameters are the physical properties of the polymer particles (in this work the wavelength-dependent refractive index and the density of the MMA/BA/MAA copolymer) and the solids’ content of the dispersion. Therefore, the particle size measured can be considered as the PDWS average. Interestingly, this average (obtained from the reduced scattering obtained with the laser with lowest wavelength of the probe; 636 nm, the most sensitive one) seems to be in excellent agreement with the number-average particle size measured by DLS. In literature, different comparisons can be found, but there is not a consensus about which average is provided by PDWS.^{[15-17, 22, 35]}

4 Conclusions

In the present work, an emulsion polymerization process to produce an MMA/BA/MAA copolymer latex was monitored by simultaneously using TUS and PDWS equipments. A calorimeter reactor was equipped with both spectroscopic instruments. For PDWS, a probe was introduced into the reactor medium allowing inline measurements, while, for TUS, an automatic dilution system was built-in that was able to extract and dilute the sample before the analysis in the measurement cell.

Thus, real-time inline and online particle size measurements were obtained during the seeded semibatch emulsion polymerization reaction. The particle size of the monitored process changed from the initial 67 nm of the seed (and 3 wt% solids’ content) to around 200 nm (and 45 wt% solids’ content) of the final latex. Both techniques predicted well the growth of the seed particles. However, the absolute value calculated by each technique matches better with different average values calculated offline by DLS.

Conflict of Interest

The authors declare no conflict of interest.