Extending the limits of emulsifier-free emulsion polymerization to achieve small uniform particles

Abstract

Conventional emulsifier-free emulsion polymerization often produces large particles, which are hardly uniform in the case of partially water-soluble monomers. In this research, preparation of uniform polymeric particles was attempted using monomer-starved semicontinuous emulsifier-free emulsion polymerization. Two monomers, with different water solubility, vinyl acetate (VA) and butyl acrylate (BA), were used as model monomers. The continuous fall in the number of particles (N_p) usually observed in the early stage of ab initio emulsifier-free emulsion polymerization transformed to a rapidly rising N_p for the semicontinuous operation. The semicontinuous process was found to be capable of producing uniform particles that are much smaller than those achievable via conventional process. This was attributed to the enhanced particle stability achieved by reducing the rate of particle growth and by prohibiting secondary nucleation via the aqueous phase being starved with the monomer. The formation of large number of small particles under monomer-starved conditions was found to be a feasible strategy to prevent secondary nucleation. This allows to apply this one-stage methodology to synthesize a wide range of uniform polymeric particles. The particle sizes decreased with decreasing monomer feed rate, while their uniformity was maintained. However, there is a minimum rate of monomer addition, depending on monomer solubility in water and polymerization conditions, below which uniform particles may not be achieved.

---

Introduction

Since 1947 when the first sample of uniform sized particles was accidentally discovered and found an instant application as internal standards for electron microscopy work, the market of uniform particles has grown immensely. Nowadays uniform particles are used in a wide variety of applications in the biomedical treatments, chromatography, information industries, electronic fields, and as building blocks for preparation of advanced materials via self-assembly. Uniform polymer particles in the micron size range are usually made by dispersion polymerization.^1,2 There are also a number of recent reports in the literature that have shown uniform microparticles can be produced using emulsifier-free emulsion polymerization with different initiator types.^3,4 In the submicron range, ab initio emulsifier-free emulsion polymerization is still the dominant method for the production of uniform polymeric particles. The method has been widely studied in detail by several investigators.^5–11 Emulsifier-free emulsion polymerization enjoys from simplicity in preparation protocol, purity of the resulting particles as usually no emulsifier or other ingredients than initiator and monomer are involved, and the low cost associated with the process. For the same reasons, emulsifier-free emulsion polymerization is in the forefront of advanced colloid research on new functional materials,¹² or new polymer chemistry.^13,14

Historically, most commercial applications of uniform particles involve polystyrene particles, which are made of a sparingly water-soluble monomer; styrene. Such monomers are easily dealt with because of their suppressed tendency for secondary nucleation. With the development of market for uniform particles, newer polymer materials with smaller particle size and narrower size distribution are becoming of interest. Polymerization of partially water-soluble monomers is usually associated with the formation of a second crop of particles that significantly affects the uniformity of latexes, as well as their properties. To produce monodisperse particles using such monomers, two or multiple-stage polymerization is usually required.¹⁵

Particles formed by emulsifier-free emulsion polymerization are usually smaller than 1.0 μm in diameter, but above 400 nm. One way to reduce the size of particles, while maintaining the uniformity, is to use a small amount of auxiliary substances including an emulsifier below its critical micellar concentration,¹⁶ an ionic or functional monomer such as acrylic acid,^17–19 or a solvent.^20–22 However, these supplementary substances can affect the purity as well as properties of the resulting particles, even if they are used in small quantities, or they may need costly post-polymerization purification. Using cationic initiators to produce positively charged nanoparticles for template applications has also been reported.²³ A practical method to reduce the size of particles is to withhold monomer from the reaction sites by the delayed addition of the monomer. Semicontinuous emulsion polymerization has been found as an attractive method to produce particles as small as 20 nm in the presence of emulsifiers.^24–27 Semicontinuous approach as a means to grow surfactant-free polymeric seed particles to the desired size range has already been well established.^4,28 However, the application of semicontinuous monomer feeding to the nucleation stage of emulsifier-free emulsion polymerization has been rarely studied in the literature.^29,30 In this research we aim to investigate the conditions under which uniform polymeric particles can be produced in one stage without the use of any auxiliary chemicals. Two model monomers of butyl acrylate (BA) and vinyl acetate (VA) with a wide water solubility difference were used.

Experimental

Chemicals

Butyl acrylate (BA) and vinyl acetate (VA) were obtained from Sigma-Aldrich (99+%). The monomers were distilled under vacuum and then stored at −18 °C. The initiator, potassium persulfate (KPS) and the buffer, sodium bicarbonate (SBc), were obtained from BDH and used as received.

Apparatus

Polymerizations were conducted in conventional 1 l jacketed glass reactor equipped with a four-bladed flat turbine type impeller with a width of 1/3 of vessel diameter, and a standard four baffle plates with the width of 1/10 of vessel diameter located at 90° interval. The temperature of the reactor content was controlled within ±0.50 °C of the reaction temperature by water with appropriate temperature being pumped through the jacket. All ingredients were separately purged with nitrogen for 30 min and then transferred to the reactor, which was kept under a blanket of nitrogen. The reaction was started with an aqueous solution of initiator and buffer in the vessel being gently stirred at 325 rpm, followed by the gradual addition of the monomer at a given rate. A delay period in the order of 0.5 to 5.0 min was observed for particle nucleation to start, depending on the conditions, which is likely to include the pre-nucleation stage. Emulsion polymerization reactions are usually preceded with a pre-nucleation stage during which radicals reach the critical size for precipitation, as stated by Fitch and Tsai.³¹ This pre-nucleation time has experimentally shown to depend on monomer and initiator concentration in the water phase.¹¹ The rate of polymerization during pre-nucleation stage is extremely low and may not be accurately measured by gravimetric methods.

Measurements

Conversions were measured gravimetrically. The instantaneous conversion at a given time, x_i, is defined as the weight ratio of the polymer formed in the reactor to the total amount of monomer fed into the reactor by that time plus the initial charge, if any. Overall conversion, x₀, is defined as the weight ratio of polymer in the reactor to the total monomer in the recipe. The steady-state rate of polymerization (R_p) was directly calculated from the slope of the overall conversion–time curves.

Laser light scattering was used for the z-average particle size (d_z) measurement. The dynamic light scattering (DLS; Malvern) measurements were carried out at 90 degree angle. Particle size distribution (PSD) of latexes were measured by using a JEOL 1000X transmission electron microscope (TEM). For constructing the PSD, plots were made by using a diameter increment of 20 nm. For PSD presentation, all particles in the range of a diameter increment were assigned to the higher bound of the size increment for the bin. The polydispersity index (PDI) of particles was calculated as the weight-average particle diameter over the number-average diameter. UV crosslinking was used for PBA polymer hardening.⁷ Surface tension of latexes in the course of reactions was measured using Du-Nouy ring method within 3 min after removal of the samples.

The number of particles, N_p, was calculated using the following equation:

N_p = (6R_atx₀/πρ_pd_v³) (1)

where R_at is the weight of monomer in the reactor at time t (with R_a being the rate of monomer addition in g h⁻¹), ρ_p is the polymer density, and d_v is the volume-average diameter of particles. In this work, we used d_z obtained by DLS, instead of d_v, to calculate the number of particles, which may lead to an underestimation of N_p.

Zeta potentials were measured using zetasizer (Malvern). A couple of methods were used for sample preparation. They will be explained in more detail where appropriate. Briefly, in one method latexes were centrifuged to precipitate particles. The top clear layer, serum, was then used to dilute the original latex for zeta potential measurement. In the other method latexes were dialyzed in cellulose tubing for a week with frequent changes of distilled de-ionised water (DDI) prior to zeta potential measurement. Measurements showed insignificant changes in the size of particles after dialysis.

In order to clean latexes for titration, they were first diluted to around 2.0 wt% solids content with DDI water and then cleaned using a mixture of ion-exchange resins, Dowex 1 and Dowex 50W, according to the procedure explained elsewhere.³² All DDI water used was purged with high purity nitrogen for 30 min to remove any CO₂ dissolved in the water phase, which could adversely affect the measurements.

The sulfate charge density of particles was measured by titration as follows; the cleaned latexes were diluted with DDI water to give 1.0 wt% solids content. The sulfuric acid groups on the surface of particles were titrated by the slow addition of sodium hydroxide (0.01 N) as the conductivity was measured. Once the hydrogen ions are neutralized, further addition of base will lead to a rise in conductance as more electrolytes enter the solution. The number of moles of the base at the minimum conductance was used to calculate the sulfate charge density.

Results and discussion

The formulation recipe and the schematic presentation of this research are given in Fig. 1. The starting transparent aqueous solution of initiator quickly turned into a bluish one and then possibly into an opaque one with further addition of monomer, depending on the polymerization conditions. Fig. 2 shows the time evolution of (a) conversions, (b) z-average particle diameter (d_z), (c) number of particles (N_p), (d) monomer concentration in the water phase ([M]_w), (e) surface tension (σ), and (f) particle size distribution (PSD) for the semicontinuous polymerization of BA and VA with different R_a and also for their batch emulsion polymerization in which all monomer was initially placed in the reactor.

Fig. 1 Top left: Schematic for the polymerization system. Top right: Recipe for the polymerizations. ^aMost experiments were started with the aqueous solution of KPS and SBc, followed by the gradual addition of monomer at flow rates of 20, 40, or 80 g h⁻¹. In the batch polymerizations, the total amount of monomer (80 g) was initially placed in the reactor. *Experiment with the initial VA monomer charge was started with 20 g monomer followed by the addition of the remainder of the monomer (60 g) at the rate of 20 g h⁻¹; indicated by R^*_a = 20 g h⁻¹ **Experiment started with the aqueous solution containing [KPS] = 0.48 g (2 mmol) and [SBc] = 0 g, followed by the addition of VA monomer (80 g) at the rate of 20 g h⁻¹; indicated by R^**_a = 20 g h⁻¹. Middle and bottom row: Schematic presentation of monomer-starved semicontinuous process and conventional monomer-saturated batch process, respectively.

Fig. 2 Time evolution of (a) conversions, (b) average particle size (d_z), (c) number of particles (N_p), (d) monomer concentration in water ([M]_w), (e) surface tension (σ), and (f) N_p × d at different rates of monomer addition: left; BA and right; VA emulsion polymerization. Open and closed symbols in (a) represent instantaneous and overall conversion, respectively. Data for batch operations are also provided for comparison.

Rate of reaction

The rate of polymerization decreased with decreasing rate of monomer addition for both monomers in such a way that R_p ≈ R_a, suggesting that the rate of polymerization was tightly controlled by the rate of monomer addition. The instantaneous conversion was higher when a lower rate of addition was used for both monomers, as expected (Fig. 2a).

Particle size

For BA monomer, the size of particles increased with increasing R_a, reaching the maximum for the batch process. The application of semicontinuous process to BA monomer led to a significant reduction in the size of particles, and a large increase in N_p (Fig. 2b and c). Interestingly, for VA monomer, the average size of final particles remained unaffected by alteration in R_a, within the range studied (Fig. 2b). From a previous report, we know that PVA particle size is insensitive to VA concentration in water ([M]_w) (i.e. low R_a) if it is lower than the saturation value (i.e., no monomer droplets exist in the reaction medium).³³ However, particle size started to increase at higher monomer concentrations,³³ as seen for the batch operation in Fig. 2c.

Particle nucleation

In a typical homogeneous nucleation, radicals formed in the water phase, by decomposition of initiator, propagate with the monomer dissolved in the water phase until they reach a critical size beyond which become insoluble in the water phase and precipitate to form precursor particles. These oligomeric chains, defined as primary precursor particles,³⁴ are extremely unstable and so they coagulate with each other, and also with clusters formed by merging oligomeric chains, to form stable primary particles. The rate of particle formation is decided by the competition between that of generation and coagulation of primary precursor particles as follows:⁸where R_gen is the rate of primary precursor particle formation by the sum of homogeneous and micellar nucleation aswhere k_pw is the propagation rate constant in the water phase, [IM_jcr−1] is the concentration of initiator derived radical with the chain length of J_cr−1 (J_cr is the critical chain length for precipitation). R_gen also includes the rate of micellar nucleation, via in situ emulsifiers, given by ρ_micelle[M]_micelle, where [M]_micelle is the concentration of in situ emulsifier micelles in the aqueous phase. These oligomeric emulsifiers are dead radicals resulting from radical termination in the water phase. As eqn (4) suggests R_gen is a function of the monomer concentration in water; [M]_w. However, this is not a linear relationship as [IM_jcr−1] is also an implicit function of [M]_w. In order to track the changes in [M]_w with time in the course of polymerization, the following equations were simultaneously solved.

R_at(1 − x_i) = [M]_pV_p + [M]_wV_w (8)

where [M]_p is the monomer concentration in the polymer particles, [M]_p,sat and [M]_w,sat are the saturation monomer concentration in the polymer and in the aqueous phase, and V_p and V_w are the volumes of the polymer and water phase, respectively. M_mon and ρ_m are the monomer molecular weight and density, respectively. Eqn (5) and (6) relate the monomer concentration in the polymer phase to the monomer concentration in the water phase and to the polymer weight ratio in the particles (w_p), respectively, under equilibrium conditions.³⁵ Eqn (7) and (8) are mass balances across the polymer phase and latex that relate R_a with the monomer concentration in both phases. We assume that the amount of monomer solubilised inside in situ surfactant micelles is negligible, and monomer droplets do not form under monomer-starved conditions.

The rate of particle coagulation/aggregation (R_coag) is function of surface charge density of the particles. The surface charge density of polymer particles is composed of the covalently bound ionic end groups and physically bound in situ generated emulsifiers. Precursor particles coagulate with each other, while simultaneously grow by absorbing monomer until they reach a size that become colloidally stable. The rate of particle coagulation, R_coag, increases with the size of precursor particles for a given number of charged groups attached.

Ab initio emulsifier-free emulsion polymerizations are marked by an instant particle formation and a significant coagulation afterwards. Goodall et al.³⁶ reported a continuous decrease in the number of particles for an ab initio emulsifier-free emulsion polymerization of styrene, which has been confirmed by many other investigators. Because of the presence of a large quantity of monomer in a typical batch process, the growth rate of particles is so rapid that their stability, acquired by a given number of initiator-derived chains, is not sufficient to protect them against coagulation, leading to a large drop in the number of particles in the early stage of reactions. Fig. 2 shows an opposite trend for the semicontinuous operation. When polymerization occurs in the presence of small quantity of monomer, the growth rate of particles is reduced so the rate of particle coagulation. Fig. 2b clearly shows that particles are smaller at a given time when the monomer is gradually added to the reactor. As a result, the particle surface charge density is higher; evidence for this is presented later. This leads to less sever coagulation and a larger number of particles. This interesting behavior resembles that of micellar nucleation; that is a gradual, but rapid, increase in the number of particles in a short span of time. This suggests that semicontinuous process (i.e., monomer-starved nucleation) can significantly shift the balance represented by eqn (3) towards the nucleation by reducing the rate of particle coagulation.

We adopt the common terminology and divide nucleation into primary and secondary nucleation. What is meant by the latter is the nucleation in the course of monomer addition and after the number of primary seed particles has been established even though sometimes the distinction between these two is not easy. We note that the rate of (secondary) particle formation is related with that of radical entry into existing particles because the two processes compete. New particles can only form if existing particles cannot sweep the propagating radicals from the water phase, before they precipitate.^6,8 The conditions for the onset of secondary nucleation is usually assessed by the product of particle size and number (N_p × d). We mentioned that the number of particles is decided by the competition between particle formation and coagulation. We now analyze the conditions under which primary nucleation occurred for these monomers.

BA monomer represents monomers with relatively low water solubility, though its solubility is still much larger than that of styrene. The solubility of BA in water is small; around [M]_w,sat = 1.4 g L⁻¹, implying that practically the initiation in all BA semicontinuous polymerizations started when the water phase was already saturated with the monomer, taking into account the rate of monomer addition and the delay time, as seen in Fig. 2d. The critical polymer weight ratio (w_p), at which PBA particles become saturated with the monomer, is around 0.40.³⁷ Below this ratio, which can be taken equal to x_i as a rough measure, excess monomer will exist as monomer droplets. The [M]_w–time data for BA (Fig. 2d) clearly point to the survival of monomer droplets in the first few minutes of the reaction when [M]_w = [M]_w,sat and x_i < w_p = 0.40, which lasted longer with increasing R_a. We may conclude from this figure that for sparingly water-soluble monomers such as BA under practical conditions (except for extremely low rates of monomer addition) primary nucleation occurs when the water phase is equal or close to the saturation level (i.e., [M]_w ≈ [M]_w,sat). Generally for a monomer such as BA with a high propagation rate constant, the propagation of radicals in the water phase is very quick, making R_gen less dependent on [M]_w. So overall the variations in N_p with R_a for BA monomer may be explained by particle coagulation (and possibly by secondary nucleation as discussed later). For this reason, we attribute the increase in N_p with decreasing R_a to the enhanced stability and reduced coagulation of PBA particles under the conditions of this study.

The solubility of VA in water is [M]_w,sat = 25 g L⁻¹, much more than that of BA, representing monomers with moderate water solubility. For the runs with R_a = 20, 40 and 80 g h⁻¹ in the absence of reaction, it takes 60, 30, and 15 min, respectively, for the water phase to reach the saturation value (the amount of water in the recipe is 800 g). The gravimetric measurement of the samples taken at 5 min detected polymer chains in the latex. Therefore, polymerizations undoubtedly started under monomer-starved conditions for theses runs to a degree depending on the rate of monomer addition. A higher concentration of monomer in water, as shown in Fig. 2d (right), allows more precursor particles to form, but at the same time it enhances the rate of particle growth, as more of the monomer accumulates in the particles. Therefore, the net effect between particle nucleation and coagulation will decide the particles number for water-soluble monomers such as VA. We find from the results that the net effect was always close to zero for VA, as the size of particles did not change with the rate of monomer addition within the range studied. We were not surprised by this result as we have previously observed a similar behavior for ab initio emulsifier-free emulsion polymerization of VA under monomer-starved conditions.³³

One general observation from Fig. 2c was that primary nucleation period (i.e., or secondary nucleation) became longer with decreasing monomer feed rate. This is plausible because in the first couple of minutes particle size and number may not be sufficient to capture all growing radicals in the water phase and as a result radicals can continue to propagate in the water phase to form new particles. This indicates that the generation of more surface-bounds and in situ surfactants during longer nucleation times encountered in semicontinuous process can be another contributing factor to the stability and thus small size of resulting particles, an advantage that may come at a cost; namely a broader particle size distribution at low R_a. Fig. 2c points to a significant secondary nucleation for some VA semicontinuous runs, in comparison with those for BA monomer which was little.³⁸ The real extent of secondary nucleation, however, cannot be easily judged by the number trajectory of particles measured by DLS, because of inaccuracies involved in detecting small particles by this technique.

The significance of in situ surfactants. For emulsifier-free emulsion polymerization, a decrease in surface tension can reflect the formation of in situ emulsifiers in the aqueous phase. Here we attempt to investigate how σ–t data can be interpreted along with nucleation. Surface-active oligomers, in situ emulsifiers, can theoretically aggregate to form micelles, which can be the nuclei for particle formation.^39,41 Kuhn and Tauer also reported aggregation of oligomeric radicals in the aqueous phase, but assumed a combination of dead and live oligomeric radicals in the aggregates.¹¹ One can see from Fig. 2e that the surface tension drops to a value around 34 mN m⁻¹ in the early reaction. This is a surface tension much lower than that can be accounted for by the presence of the monomer in water, and is close to that of micellar solution of typical anionic emulsifiers such as sodium dodecylsulfate,²⁵ suggesting that the critical micellar concentration (CMC) was reached, and the resulting micelles contributed to the particle nucleation/stabilization. Rudin et al. showed that they were able to produce particles by using serum, containing in situ emulsifiers, obtained from emulsifier-free emulsion polymerizations.⁴⁰

A continuous rise in N_p during the time that surface tension was low (i.e. when micelles are likely to be present) is observed in Fig. 2c and e. However, this was not a condition as primary nucleation also occurred when surface tension was high and above that at the supposed CMC. See Fig. 2e for BA run with R_a = 40 g h⁻¹ and VA run with R_a = 20 g h⁻¹, for examples. Apparently, homogeneous nucleation, via oligomer precipitation, was the main mechanism that contributed to the number of particles under such conditions.

Generally for BA monomer, surface tension (Fig. 2e) decreased more significantly for the lower R_a. The surface tension dropped to a value as small as 34 mN m⁻¹ for R_a = 20 g h⁻¹, and then rose to a steady-state value. For the higher feed rate, R_a = 40 g h⁻¹, the surface tension only dropped to 47.0 mN m⁻¹ during polymerization. For BA system, the rate of generation of in situ emulsifiers during primary nucleation is not expected to have greatly been affected by R_a as the monomer concentration in the water phase was at the saturation level, as shown in Fig. 2d. However, a high rate of particle growth, caused by the higher rate of monomer addition, can sweep in situ emulsifiers from the water phase, leading to a quick recovery in the surface tension. Furthermore, the monomer droplets formed in the early reaction can also extract in situ emulsifiers from the aqueous phase by adsorption. So overall, the emerging minimum in the surface tension for the lower R_a can be attributed to the limited growth of particles and fewer monomer droplets formed in the system. This suggests that in situ emulsifiers can be more effectively involved in particle nucleation, rather than stabilization, at lower R_a. Hergeth et al. also reported a significant drop in the surface tension of latexes with the addition of monomers at the conditions similar to this work.⁴¹ A surface tension as small as 42 mN m⁻¹ was reported for VA and styrene monomers, which appears to be larger than the values obtained in this work. A persistent low surface tension for BA polymerization (≈50 mN m⁻¹) in this work may indicate a possible secondary nucleation in the course of monomer addition too,⁴² but its extent, as we will see later, was limited.

For VA monomer, the fall in surface tension was greater with the higher R_a, an opposite trend to that observed for BA monomer. A higher rate of VA monomer addition led to a higher monomer concentration in the water phase ([M]_w), as seen in Fig. 2d. While the relation of homogenous nucleation with [M]_w is clearly understood, that of generation of in situ emulsifier has not been well elaborated in the literature yet. Simulation results, which can conceptually be worked out too, indicate that the rate of nucleation of primary precursor particles, as well as radical entry, increases with [M]_w, while the rate of radical termination in water (i.e. generation of potentially in situ emulsifiers) will decrease.³⁸ This is because more radicals can reach the critical chain length for entry or precipitation (nucleation) with increasing monomer concentration in water. This truly suggests that the concentration of terminated radicals or oligomers in water increases with a decrease in the monomer concentration in water. One would expect to see a more decrease in surface tension with decreasing [M]_w, if all terminated radicals were surface active. However, oligomeric radicals propagate more slowly at a low [M]_w before they terminate, and as a result they are shorter and less surface active. This is probably the reason that a lower minimum in interfacial tension was reached with increasing R_a (i.e., increasing [M]_w), as shown in Fig. 2e. It is apparent from Fig. 2c that primary particle formation occurred during the primary minimum in σ–t for all runs. This minimum in surface tension was recovered soon to a value around 45 mN m⁻¹, and sometimes followed by a secondary minimum during which secondary nucleation could have occurred but at a drastically lower rate.

Particle size distribution. TEM micrographs of final PVA latexes and their constructed PSDs are shown in Fig. 3. The time evolution of PSDs, and their polydispersity indexes are given in Fig. 4. For the batch polymerization, the PSDs of both monomers appeared to be bimodal, but to a degree increasing with the monomer solubility in water. This suggests that the conventional batch emulsifier-free emulsion polymerization may not be an ideal system for producing uniform particles of monomers that are even slightly water soluble. As a general statement, the semicontinuous runs produced sharper PSD than batch operations (Fig. 3e and f). These results might appear to be surprising because of small mass of the polymer (low solids content) produced at low feed rates during primary nucleation, which hardly appears to be sufficient to prevent secondary nucleation. The term N_p × d, which represents the likelihood of radical entry into particles and a measure by which the extent of secondary nucleation can be compared,³⁸ is found to be greater for the latexes produced by batch operation in the first few minutes of reactions, as shown in Fig. 2f. However, this quickly changes because of the formation of a large number of small particles at low R_a, which boosts the N_p × d. Furthermore, the monomer concentration in the water phase, as seen in Fig. 2d, is small for the semicontinuous runs. This is an additional reason that secondary nucleation did not occur for some semicontinuous runs. The PDI of the final latexes made by semicontinuous polymerization broadened with decreasing rate of monomer addition, as listed in Fig. 4d, because of longer primary (i.e., secondary) nucleation. For BA monomer, the PDI increased from 1.021 to 1.032 with decreasing R_a from 40 to 20 g h⁻¹.

Fig. 3 TEM micrographs of final PVA latexes made with (a) R_a = 20 g h⁻¹, (b) R_a = 40 g h⁻¹, (c) R_a = 80 g h⁻¹, (d) batch, and PSDs of (e) PBA and (f) PVA latexes. See caption of Fig. 1 for the details of experiments (scale bar is 1000 nm).

Fig. 4 Time evolution of PSDs of PVA latexes for (a) batch process, (b) semicontinuous run with R_a = 40 g h⁻¹, and (c) R_a = 20 g h⁻¹, and (d) the polydispersity index (PDI) of the final PVA and PBA latexes obtained at different conditions as specified in the caption of Fig. 1.

We now look more closely at the time evolution of PSD for VA monomer, as shown in Fig. 4a–c. The size distribution of PVA particles at early conversions became sharper with increasing R_a, becoming the sharpest for the batch operation. However, the PSDs became progressively broader with time and decreasing rate of monomer addition. For VA batch operation, a secondary peak appeared at later conversions (x₀ ≈ 0.78 as seen in Fig. 4a). The latex produced with R_a = 80 g h⁻¹ was found to be quite uniform. With this rate of monomer addition, the primary nucleation was relatively fast, and the PSD did not broaden in the course of reaction as the secondary nucleation was insignificant. This is supported by the large N_p × d value obtained for this run, as shown in Fig. 2f. For the run with R_a = 40 g h⁻¹, the PSD was initially sharp, but a small crop of new particles appeared at later conversions (Fig. 4b). At R_a = 20 g h⁻¹, the primary nucleation did not really stop and a skewed PSD evolved with time, as seen in Fig. 4c. The product N_p × d decreased with decreasing R_a (Fig. 2f) implying that the likelihood of continued nucleation (or secondary nucleation) is more significant at lower R_a.

We can conclude that, although particle formation in the semicontinuous runs was not instantaneous, as in the case of ab initio polymerization, was fast enough to result in particles with narrow size distributions for both monomers. Such sharp PSDs also imply that secondary nucleation can effectively be controlled in the semicontinuous polymerization process by the generation of a large surface area of small particles, even at monomer-starved conditions where the growth of particles is limited. The results also indicate that there is a lower end to the extent by which R_a can be reduced while the particles uniformity being maintained. This critical value was found to be around 80 g h⁻¹ for VA and roughly 40 g h⁻¹ for BA monomer at the conditions of this study. These critical R_a values vary for different polymerization systems as other influential parameters such as solubility and the propagation rate constant of monomers in water are also important.

Policies to produce narrow particle size distribution

The reason for sharp particle size distributions usually encountered in ab initio emulsifier-free emulsion polymerization is a fast nucleation period during which particles grow equally and simultaneously, in contrast to emulsion polymerization with rather long nucleation times. The results clearly indicate sharp PSDs can also be obtained via semicontinuous emulsifier-free processes, but they quickly degrade if continuous (secondary) nucleation is predominant. In this section, we seek possible scenarios that can serve to shorten primary nucleation and/or prevent secondary nucleation in a typical semicontinuous polymerization.

Maximization of particle size (d) by the use of an initial monomer charge. We find from the discussion so far that the main reasons for PSD broadening in VA polymerization are a relatively long primary nucleation period in the start of reaction and a sustained secondary nucleation in the course of reaction. So there is a trade-off; while increasing R_a will shorten the primary nucleation and produce sharp initial PSDs, it can also increase the possibility of secondary nucleation and PSD broadening, because of a higher [M]_w in the water phase. One possible solution towards this is to conduct the polymerization reaction with an amount of monomer in the reactor charge in order to increase d, and possibly N_p, and thus their product N_p × d. It is understood that the instantaneous rate of particle nucleation is maximized when the aqueous phase is saturated with a monomer,⁶ though this may not lead to a maximum N_p.³³ A monomer concentration greater than the saturation level would not significantly affect the rate of nucleation but could substantially enhance the rate of particle coagulation, and thus reduce N_p. For some monomers such as methyl acrylate, N_p is maximized at [M]_w ≈ [M]_w,sat, but N_p × d (at the end of primary nucleation) reaches its maximum at the monomer saturation level of around [M]_w/[M]_w,sat = 75%.³³ For methyl methacrylate (MMA) monomer, interestingly, N_p was minimum when the water phase was saturated with MMA due to the gel effect.³³ As a result, maximization of N_p × d for this monomer could only be achieved at a monomer concentration much above the saturation level. The preceding discussion clarifies that there might be different scenarios applicable to different monomers. For VA monomer, as previously explained, N_p was found to be independent of [M]_w if [M]_w ≤ [M]_w,sat. As a result, the maximization of the term N_p × d can be achieved by enlarging the size of particles via using the water phase being fully saturated with VA monomer; [M]_w = [M]_w,sat.³³ Therefore, we conducted a polymerization run that started batchwise with the water phase saturated with VA monomer (monomer ratio in the reactor charge to that in the feed is 25%), followed by the addition of the remainder of monomer at R^*_a = 20 g h⁻¹ (* is used as a symbol for this run). A pre-period time of 10 min was allowed before monomer addition started in order to achieve high conversions.

The results, in comparison with those of the control experiment (no seeding stage) are shown in Fig. 5. As seen in this figure, a high conversion was achieved quickly (Fig. 5a), the nucleation was very fast, and particles were stabilized very quickly without any fall in their number (Fig. 5b and c). The monomer concentration in water started from the saturation value in the early reaction and quickly reduced in line with the progress of reaction (Fig. 5d), while surface tension only fluctuated around 45 mN m⁻¹ (Fig. 5e) and the N_p × d (Fig. 5f) was the highest in the early stage among all runs, indicating the lowest possibility for secondary nucleation. An enhanced primary nucleation combined with a suppressed secondary nucleation are the key features of this experiment, which led to the production of uniform particles, as shown in Fig. 6b.

Fig. 5 Time evolution of (a) conversions, (b) d_z, (c) N_p, and (d) monomer concentration in the water phase ([M]_w), (e) surface tension (σ), and (f) N_p × d_z of the three PVA semicontinuous runs at R_a = 20 g h⁻¹. Open and closed symbols in (a) represent instantaneous and overall conversion, respectively. R_a = 20 indicates the base (control) experiment and is provided for the sake of comparison, R^*_a = 20 g h⁻¹ is the run with the initial VA monomer charge, and R^**_a = 20 g h⁻¹ is the experiment with low electrolyte concentration. See caption of Fig. 1 for all details.

Fig. 6 TEM micrographs for final particles for runs with (a) R_a = 20 g h⁻¹, (b) R^*_a = 20 g h⁻¹, and (c) R^**_a = 20 g h⁻¹. (d) Shows the final PSDS for the runs. See caption of Fig. 1 for all details (scale bar is 1000 nm).

It should be noted processes that use initial monomer charge have already been established as “in situ seeded emulsion polymerization”. This is a widely used technique to produce large uniform polymer particles. In this section, we intended to show that by a meticulous design of the initial charge conditions one can limit the extent of secondary nucleation. One point worthy of note is that monomer-starved semicontinuous polymerizations are susceptible to the delay time, comprised of pre-nucleation stage and possible inhibition time, during which monomers can accumulate in the reactor. An exhaustive method was used to eliminate oxygen content in the aqueous phase. However, the pre-nucleation time, which is to do with the kinetics of the aqueous-phase polymerization, cannot be eliminated. As a result, the semicontinuous runs reported in this work may have included a short semi seeding stage.

Maximization of N_p by enhancing the rate of nucleation or suppressing the rate of coagulation. We have noticed that secondary nucleation occurred even in the high-solids content batch emulsion polymerization of VA in the presence of large particles (see Fig. 3d and f). This confirms that particle number is also important in the prevention of secondary nucleation. In this section, we explore other means by which particle number can be increased, even though this may be accompanied by a decrease in the solids content prior to the end of nucleation (taken equal to the polymer phase volume V_p). The batch operation and also operations that use monomer initial charge take advantage of a large V_p by the end of primary nucleation, which can serve to limit secondary nucleation. However, it can be shown that particle size is more important than the solids content in controlling secondary nucleation. The product N_p × d is related to the solids content and particle diameter as follows:

The above equation suggests that a significant enhancement in particle number via monomer-starved nucleation may be sufficient to hinder secondary nucleation at right conditions. One obvious choice to enhance nucleation is to increase the reaction temperature, by which the rate of radical generation increases so the number of particles. Alternatively, N_p can be boosted by suppressing the rate of particle coagulation. Here, we report the preparation of emulsifier-free PVA latexes at a low electrolyte concentration. The particles made using the recipe given in Fig. 1, with a high concentration of electrolytes (i.e., initiator and buffer) were susceptible to coagulation during nucleation, as the electrical double layer is collapsed in the presence of too much electrolytes. We conducted a semicontinuous run at a reduced amount of initiator, 0.43 g or 2 mmol l⁻¹, in comparison to 6.0 mmol l⁻¹ used previously, in the absence of buffer, at the monomer addition rate of R^**_a = 20 g h⁻¹ (** is used as the symbol for this run). The results for this run are also shown in Fig. 5.

The final particle size decreased drastically using a reduced concentration of electrolyte, in comparison with other runs (see Fig. 5b). Note that usually particle size is increased with decreasing initiator concentration at a constant electrolyte concentration. However, in this experiment the decrease in the initiator concentration was accompanied by a drastic decrease in electrolyte concentration too. The nucleation did not start within the first 5 min. The relatively longer delay time in this case can be attributed to the longer pre-nucleation time at the lower KPS concentration used⁴³ and possibly to the lower concentration of initiator derived radicals, which can react with traces of the dissolved oxygen or any impurities in the water phase during the delay time. The particles in the early stage of monomer addition were as small as 50 nm in diameter but reached 330 nm by the end of reaction. One can see from Fig. 5c that almost 5 times increase in the number of particles was resulted, when the initiator concentration was reduced by one third while the rate of reaction did not change significantly as it was tightly controlled by the rate of monomer addition (see Fig. 5a).

Furthermore, the formation of a large N_p caused the instantaneous conversion to reach higher values, opposite to the effect one would expect to see by decreasing initiator concentration. Unlike the run carried out at the high electrolyte concentration, nucleation was completed rather early and N_p remained almost constant in the course of reaction, as seen from Fig. 5c. The product N_p × d for this run was found to be the second highest in the early stage, with the run with initial monomer charge being the first, but the largest in the later stage, among all three runs, indicating less likelihood for secondary nucleation to occur. The surface tension was within the same range as those in the other runs, as shown in Fig. 5d, but slightly higher. This is probably because of smaller amount of in situ surfactants formed, as a result of reduced amount of initiator used. The large particle surface area generated could sweep growing radicals and in situ oligomeric emulsifiers from the water phase to prevent secondary nucleation. As a result, uniform particles, as seen in Fig. 6c, were produced. However, the size distribution of particles was slightly broader than the run with the initial charge, as seen in Fig. 6d and indicated in Fig. 4d. The results suggest that particle uniformity can be tuned without a need to use seeding stage. This requires that conditions are optimized to produce large number of small particles, which is more easily achieved under monomer-starved conditions. This also suggests that the optimum R_a can be further reduced by optimization of the recipe and reaction conditions.

Colloidal stability

The stability of particles in emulsifier-free emulsion polymerization is mainly due to surface-bound and physically absorbed ions, which are in fact in situ emulsifiers made by termination of propagating radicals in the aqueous phase. Cleaned latexes are stripped off the physically adsorbed ions, but do maintain their chemically surface bound ions, which are transformed to hydrogen ions after cleaning with ion-exchange resins.¹

Fig. 7 (top) shows the variations in the strong acid sulfate surface charge density of particles with R_a. Note this surface charge density does not include the contribution of weak acid groups, such as carboxylic acid groups that are formed via hydrolysis of strong acid groups. The difference in the surface charge density of PVA and PBA particles from the batch process was negligible, but widened with increasing R_a (longer addition time) for the semicontinuous runs. The longer polymerization time, as well as the lower molecular weight of the resulting polymers from semicontinuous operations,²⁴ means that more initiator radicals contribute to the particle surface charge density. This occurs despite the fact that the total surface area of particles generated by the semicontinuous operation is several times greater than that of batch operation and also that some sulfate ends may get buried in the particles during the growth stage as more monomer is polymerized with time at the surface of particles.

Fig. 7 Top: Variations in the sulfate charge density of final latexes with R_a for VA and BA monomers. Bottom: Zeta potential of final latexes (see Fig. 1 for the full description of experiments). *This represents the run with (initial monomer charge. **This represents the run with [KPS] = 2 mmol. ^aThis zeta potential represents surface-bound ions. ^bThis zeta potential represents surface-bound ions and in situ adsorbed emulsifiers.

The electrostatic stability of latex particles can also be reflected in zeta potential, which is in fact the difference in potential between the water phase and the stationary layer of water attached to the particles. Particles with a high zeta potential are electrically stabilized while particles with a low zeta potential tend to flocculate or coagulate. Both initiator-derived radicals and in situ emulsifiers can contribute to zeta potential. We prepared two different sample latexes for zeta potential measurements as follows. (1) Cleaned particles: latexes were cleaned by dialysis with DDI water for a week, and then diluted with DDI water by 1/50 for measurements. This method will only represent anchored emulsifier as in situ emulsifiers are removed during dialysis. (2) Uncleaned particles: latexes were diluted with the serum, obtained by the centrifugation of the same latex, by 1/50 (particles are dispersed in the same medium as during the reaction). This method represents contribution from both surface-bound and physically absorbed ions.

Fig. 7 (bottom) shows that the zeta potential of the cleaned and uncleaned particles made by semicontinuous polymerization is much higher than those made by batch operation for both monomers. Also, the zeta potential of particles increased with decreasing monomer feed rate, in agreement with the statement already made based on the sulfate charge density of cleaned particles. The run R^**_a = 20 g h⁻¹, has the minimum zeta potential among the similar runs because of lower concentration of KPS used and a larger surface area of particles developed. Interestingly, the uncleaned particles (produced by the 2^nd method) had a higher zeta potential than the cleaned ones, for both monomers, despite the presence of electrolyte in the serum (initiator and buffer). This re-emphasizes the importance of in situ emulsifiers in stabilizing polymer particles. These all clearly demonstrate why particles made via semicontinuous mode are more stable than those made by batch process.

Another interesting observation worthy of comment is that the same trend as that obtained for the sulfate charge density of particles is observed for zeta potential of the cleaned particles (including only chemically-bound ions) (see Fig. 7); a higher zeta potential for PVA particles than for PBA particles. This may suggest that particles made of monomers with high water solubility have a higher surface charge density because of larger radical entry efficiency. Whereas, the zeta potential of uncleaned particles, containing both chemically and physically bound ions, is significantly larger for PBA than for PVA particles under the same conditions, opposite to the trend observed for chemically surface-bound ions. This latter may complete the previous statement, that is, because of the relatively higher rate of radical termination in the water phase (smaller radical entry efficiency) and also the fact that [M]_w remains close to [M]_w,sat during nucleation (in semicontinuous process) for less water-soluble monomers, more in situ surfactants are likely to form and contribute to particle stability, as indicated by a larger drop in the surface tension for BA. This suggests that in situ emulsifiers play a more decisive role in the stability of particles made of sparingly water soluble monomers than more water-soluble monomers.

Conclusion

Conventional emulsifier-free batch emulsion polymerization of sparingly water-soluble monomers often produces large uniform polymer particles because of the rapid particle coagulation associated with primary (homogenous) nucleation. We report application of semicontinuous operation as a means to produce small uniform particles. The main conclusions are:

• Batch emulsifier-free emulsion polymerization of monomers with modest water solubility such as VA and BA may produce large particles with rather broad particle size distributions because of secondary nucleation.

• A semicontinuous operation will decrease the size of particles via enhancing their stability by controlling their growth and thus increasing their surface charge density.

• An enhanced stability of particles is represented by an increasing N_p in the early stage of reaction, unlike conventional emulsifier-free systems where a sharp fall in N_p occurs.

• A fast, but not necessarily an instantaneous nucleation as often believed, can be considered as a pre-requisite to the formation of uniform particles.

• Numerous small particles with sharp PSDs can be produced at low R_a, which can prevent secondary nucleation. However, a threshold value for R_a may be found below which uniformity of particles start to degrade due to continuous nucleation. This threshold value depends on the monomer solubility in water and polymerization conditions.

• An in situ seeded semicontinuous approach, in which the primary nucleation (particle size times particle number) can be boosted by the presence of small amount of monomer in the reactor initial charge was found as an alternative way to produce small uniform particles.

References

1. C. M. Tseng, Y. Y. Lu, M. S. Elaasser and J. W. Vanderhoff, J. Polym. Sci., Polym. Chem. Ed., 1986, 24, 2995–3007.
2. R. Arshady, Colloid Polym. Sci., 1992, 270, 717–732.
3. D. Nagao, T. Sakamoto, H. Konno, S. C. Gu and M. Konno, Langmuir, 2006, 22, 10958–10962.
4. Y. Yamada, T. Sakamoto, S. C. Gu and M. Konno, J. Colloid Interface Sci., 2005, 281, 249–252.
5. J. W. Goodwin, J. Hearn, C. C. Ho and R. H. Ottewill, Colloid Polym. Sci., 1974, 252(6), 464–471.
6. F. K. Hansen and J. Ugelstad, J. Polym. Sci., Polym. Chem. Ed., 1978, 16, 1953–1979.
7. F. K. Hansen and J. Ugelstad, J. Polym. Sci., Polym. Chem. Ed., 1979, 17, 3033–3045.
8. R. M. Fitch, B. Prenosil and K. J. Sprick, J. Polym. Sci., Part C: Polym. Symp., 1969, 27PC, 95–118.
9. B. W. Brooks and J. Wang, Polymer, 1993, 34, 119–123.
10. Z. Q. Song and G. W. Poehlein, J. Colloid Interface Sci., 1989, 128, 486–500.
11. I. Kuhn and K. Tauer, Macromolecules, 1995, 28, 8122–8128.
12. C. E. Reese and S. A. Asher, J. Colloid Interface Sci., 2002, 248, 41–46.
13. Y. Kitayama, K. Kishida and M. Okubo, J. Polym. Sci., Polym. Chem. Ed., 2013, 51, 716–723.
14. Y. J. Chen, W. Luo, Y. F. Wang, C. Sun, M. Han and C. C. Zhang, J. Colloid Interface Sci., 2012, 369, 46–51.
15. E. D. Sudol, M. S. Elaasser and J. W. Vanderhoff, J. Polym. Sci., Polym. Chem. Ed., 1986, 24, 3499–3513.
16. H. Ishii, M. Ishii, D. Nagao and M. Konno, Polymer, 2014, 55, 2772–2779.
17. J. H. Kim, M. Chainey, M. S. Elaasser and J. W. Vanderhoff, J. Polym. Sci., Part A: Polym. Chem., 1992, 30, 171–183.
18. M. S. D. Juang and I. M. Krieger, J. Polym. Sci., Polym. Chem. Ed., 1976, 14, 2089–2107.
19. S. Jiang, A. Aun, J. P. Deng, W. T. Yang and Q. Yu, J. Macromol. Sci., Part A: Pure Appl. Chem., 2011, 48, 846–850.
20. M. Okubo, A. Yamada, S. Shibao, N. Katsuhiko and T. Matsumoto, J. Appl. Polym. Sci., 1981, 26, 1675–1679.
21. T. Ngai and W. Chi, Langmuir, 2005, 21, 8520–8525.
22. S. T. Camli, F. Buyukserin, O. Balci and G. G. Budak, J. Colloid Interface Sci., 2010, 344, 528–532.
23. A. B. D. Nandiyanto, A. Suhendi, T. Ogi, T. Iwaki and K. Okuyama, Colloids Surf., A, 2012, 396, 96–105.
24. Y. Chen, F. Jahanzad and S. Sajjadi, Langmuir, 2013, 29, 5650–5658.
25. S. Sajjadi, Langmuir, 2007, 23, 1018–1024.
26. J. S. Nunes and J. M. Asua, Langmuir, 2012, 28, 7333.
27. G. W. He, Q. M. Pan and G. L. Rempel, Macromol. Rapid Commun., 2003, 24, 585–588.
28. C. S. Chern, C. H. Lin and T. J. Chen, Polym. J., 1997, 29, 249–254.
29. C. S. Chern and C. H. Lin, Polym. J., 1995, 27, 1094–1103.
30. C. S. Chern and C. H. Lin, Polymer Int, 1997, 42, 409–421.
31. R. M. Fitch and C. H. Tsai, in Polymer Colloids, ed. R. M. Fitch, Plenum, New York, 1971, p. 103.
32. J. W. Vandehoff, H. J. Van den Hul, R. J. M. Tausk and J. T. G. Overbeek, Clean Surfaces, ed. G. Goldfinger, Dekker, Newyork, 1970, p. 15.
33. Y. Chen and S. Sajjadi, Polymer, 2009, 50, 357–365.
34. P. J. Feeney, D. H. Napper and R. G. Gilbert, Macromolecules, 1987, 20, 2922–2930.
35. R. G. Gilbert, Emulsion polymerisation: A mechanistic approach, Academic press, London, 1995.
36. A. R. Goodall, M. C. Wilkinson and J. Hearn, J. Polym. Sci., Polym. Chem. Ed., 1977, 15, 2193.
37. S. Sajjadi and B. W. Brooks, J. Polym. Sci., Polym. Chem. Ed., 1999, 37, 3957–3972.
38. C. J. Ferguson, G. T. Russell and R. G. Gilbert, Polymer, 2002, 43, 4557–4570.
39. Z. Q. Song and G. W. Poehlein, J. Polym. Sci., Polym. Chem. Ed., 1990, 28, 2359–2392.
40. Z. Y. Wang, A. J. Paine and A. Rudin, J. Polym. Sci., Polym. Chem. Ed., 1990, 33, 1597–1606.
41. W. D. Hergeth, W. Lebek, R. Kakuschke and K. Schmutzler, Macromol. Chem. Phys., 1991, 192, 2265–2275.
42. A. Sajjadi, J. Polym. Sci., Polym. Chem. Ed., 2000, 38, 3612–3630.
43. K. Tauer, H. Hernandez, S. Kozempel, O. Lazareva and P. Nazaran, Colloid Polym. Sci., 2008, 286, 499–515.

---