Kettle Stability of Styrene Acrylic Latex during Semibatch Monomer Feed

In a 12,000 L glass-lined batch reactor equipped with a retreat-curve impeller having a tip speed of 2.9 m/s and four wall baffles, styrene acrylic latex is produced by semibatch monomer feed under monomer-starved conditions. The aqueous phase is charged at 78°C, the seed latex is present at 0.8 wt% solids on final monomer, and a pre-emulsion containing 50 wt% monomer is metered over 280 min. Kettle stability in this system is governed by the balance between particle aggregation and surface charge density, the time-dependent ratio of free surfactant to exposed interfacial area, and the local heat release near the sub-surface feed nozzle. A monomer mass flow of 18.5 kg/min for a final latex solids target of 50 wt% typically produces an instantaneous conversion of 85–90% for the styrene-rich feed, but excursions above 24 kg/min can create a flooded zone in which monomer droplets persist and the heat removal requirement exceeds the jacket’s capability. The resulting particle size distribution broadens from a target 150–180 nm z-average to values above 250 nm, and coagulum collected on a 150 µm screen increases from below 0.02 wt% to above 0.35 wt% based on latex solids. Such a transition is detectable not only by sieve residue following ASTM D1417-16 but also by a drop in the zeta potential from −48 mV to −32 mV when the surface coverage of carboxylate groups is insufficient to compensate for the increase in interfacial area during transient monomer swelling.

Semibatch feeding of styrene and acrylic ester as a pre-emulsion is chosen because it suppresses the compositional drift that would otherwise arise from the large difference in reactivity ratios; styrene and n-butyl acrylate have published reactivity ratios of approximately 0.75 and 0.18 in bulk copolymerisation, and the monomer-starved operating point keeps the instantaneous copolymer composition near the feed composition. In continuous aqueous-phase nucleation systems, however, the semibatch feed transfers mass from monomer droplets to polymer particles only when the aqueous-phase monomer concentration is below saturation. The kettle is therefore sensitive to any interruption in agitation during the feed, because a stop of more than 60 s can allow monomer droplets to coalesce and create a large monomer-rich layer that later reacts as a bolus. Agitator current, jacket return temperature, and reflux condenser load are monitored at 1 Hz; a drop in agitator current combined with a feed pump interlock is the primary control for avoiding a monomer pool. The use of a 50% monomer pre-emulsion with a droplet size of 250–450 nm reduces the tendency to form large monomer reservoirs, but the pre-emulsion itself must be stirred continuously at 60–80 rpm and kept below 30°C to prevent creaming and premature thermal polymerisation of acrylic monomers.

When the Monomer Feed Profile Exceeds the Jacket's Cooling Duty in a 20 m³ Reactor

For a 20,000 L glass-lined polymerisation vessel with a heat transfer area of approximately 42 m² and a measured overall heat transfer coefficient of 320–380 W/m²K under clean-wall conditions, the maximum steady-state cooling duty at a 12°C log-mean temperature difference is 160–190 kW. Because the copolymerisation of styrene with n-butyl acrylate releases 640–700 kJ/kg of monomer reacted, a semibatch monomer feed rate of 22 kg/min corresponds to an average heat release of 234–257 kW if conversion is instantaneous. This apparent imbalance is tolerable only because the monomer feed is intentionally starved and the actual exothermic load is distributed across the feed interval plus a short residual conversion period. Operation above 22 kg/min, or a loss of agitation that reduces the product-side film coefficient, produces a rise in kettle temperature from 80°C to 88–92°C within 15–25 min. The decomposition half-life of persulfate initiator in the aqueous phase drops sharply in this window, increasing the radical flux and accelerating conversion in a positive feedback loop. The resulting runaway is most severe when the pre-emulsion feed is directed into a quiescent zone behind a baffle or onto the thermowell, where local heat transfer is poor. Stable operation therefore requires a feed nozzle located 150–200 mm below the liquid surface and angled 30–45° into the impeller discharge stream, a maximum temperature alarm at 84°C, and an automatic feed shutoff interlocked with agitator current and jacket return temperature. The feed shutoff logic should be configured with a dead time of no more than 10 s and a safe-state that diverts the monomer pre-emulsion to a chilled hold tank.

The limiting heat removal rate is not static across the batch. During the first 15% of monomer feed, the reactor contains a low volume fraction of dispersed polymer and the latex viscosity is below 100 mPa·s; the product-side film coefficient may be as high as 1,500–2,500 W/m²K. As the solids content rises above 35 wt%, the viscosity at 20 rpm increases to 400–800 mPa·s, and the impeller pumping capacity suffers if the agitator is operated below 40 rpm. At 50 wt% solids the overall heat transfer coefficient can fall to 250–300 W/m²K because of increased wall deposition and lower Reynolds number. The jacket control valve must therefore be capable of maintaining a chilled water outlet temperature of 10–15°C while the return temperature is monitored; a return temperature above 45°C indicates fouling or a loss of recirculation. In one common failure mode, the jacket is connected to a cooling tower water loop at 25–30°C during summer months, and the available temperature difference falls to less than 5°C. The feed profile must then be reduced to 14–16 kg/min or the batch must be moved to a chiller loop at 5°C. Published data for the exact reduction factor in a high-solids styrene acrylic formulation is limited, but the effect is predictable from the ratio of log-mean temperature differences.

Mechanistically, the pre-emulsion droplet size does not transfer directly into latex particles; monomer is transported from droplets to growing particles by aqueous-phase diffusion. The droplet size, typically 250–450 nm when prepared with a high-shear disperser at 1,200 rpm, is nonetheless relevant to kettle stability because larger droplets can coalesce and form a monomer-rich layer above the aqueous phase. Kettle stability during the final 40 min of feed is dominated by the free surfactant concentration, which industrial conductivity and surface tension measurements place in the range of 0.15–0.30 g/L for sodium dodecylbenzene sulphonate when the latex is stable. If the free surfactant concentration falls below 0.05 g/L because of excessive particle surface area generated by secondary nucleation, the latex becomes shear-sensitive and wall deposits on the 80 mm baffle edges are observed. A secondary surfactant such as disodium ethoxylated alcohol half-ester of sulphosuccinic acid can be fed during the final 20% of monomer addition at 2.0–3.0 kg per tonne of monomer to improve stability without introducing alkylphenol ethoxylates, which are restricted under REACH Annex XVII. The secondary surfactant does not create new particles at this stage; instead it adsorbs on the exposed hydrophobic surface of particles that have outgrown their primary surfactant layer. The required secondary surfactant dose scales with the square of the z-average particle diameter and is typically twice as large for a 220 nm latex as for a 160 nm latex because the total interfacial area per unit mass of polymer increases as particle diameter decreases. Conductivity monitoring of the aqueous phase below 2,500 µS/cm during the entire feed is used to avoid electrolyte compression of the electrostatic double layer.

Pre-emulsion stabilisation is also affected by the monomer composition. Styrene has a water solubility of approximately 300 mg/L at 25°C, while n-butyl acrylate is approximately 1,500 mg/L; this difference means that butyl acrylate tends to transfer through the aqueous phase more rapidly and can generate secondary particles if the aqueous phase is undersaturated with surfactant. The ratio of styrene to n-butyl acrylate in the feed is commonly 55/45 wt/wt for architectural binders, but reducing the styrene fraction to 40 wt% increases the aqueous-phase propagation rate and the tendency to form secondary particles. Secondary nucleation at high conversion raises the particle number concentration from 1.0 × 10¹⁷ particles per litre to 2.5 × 10¹⁷ particles per litre and creates a bimodal particle size distribution that is often mistaken for insufficient detergent. Kettle stability in this case is improved not by adding more surfactant but by reducing the feed rate during the first 30 min and lowering the initiator concentration slightly from 0.35 wt% to 0.25 wt% on monomer. The initiator feed is also split, with 70% in the aqueous charge and 30% fed as a dilute solution during the feed, to avoid a burst of radicals when the pre-emulsion first contacts the hot aqueous phase.

What Limits Anionic Surfactant Efficiency at pH 4.8 with 0.5 wt% Acrylic Acid?

At pH 4.8, the carboxyl groups of acrylic acid repeat units are only partially ionised; with a pKa of 4.25, the fraction of carboxylate functionality is approximately 0.55, while methacrylic acid with a pKa of 4.66 contributes a slightly lower fraction. The latex surface charge therefore depends on both the copolymerised acid level and the counterion type. When 0.5 wt% acrylic acid on total monomer is copolymerised under a sodium persulfate initiator at 80°C, sulphate end-groups contribute an additional anionic surface charge estimated at 1–2 µC/cm², but this charge is progressively buried as the particle grows beyond 180 nm. In a semibatch feed with a delayed acid shot at 70–80% monomer addition, the carboxyl groups are concentrated in the outer shell, increasing the negative surface potential from −35 mV to −52 mV at 0.5 wt% acid content. Conversely, if all acid is present in the initial charge, the acid is buried and the latex shows poor electrolyte tolerance; addition of 0.5 mol/L calcium chloride causes visible coagulation below 100 µm in less than 5 min. Maintaining the aqueous phase conductivity below 2,500 µS/cm and dosing sodium bicarbonate at 0.2–0.3 wt% on monomer prevents the pH drift below 4.2 that otherwise accelerates acid protonation and reduces the electrostatic barrier. Potentiometric titration of cleaned latex serum distinguishes surface acid from buried acid; a surface acid content of at least 0.25 wt% on polymer is necessary for high-solids stability.

The anionic surfactant concentration required to prevent coagulation is not fixed by the monomer mass but by the total particle surface area. For a 50 wt% solids latex with a z-average particle size of 170 nm, the specific surface area is approximately 22 m²/g of polymer. The molecular area of sodium dodecylbenzene sulphonate at the polystyrene-water interface is roughly 0.55 nm² per molecule, so a full monolayer on 1 g of polymer would require about 6.6 × 10⁻⁵ mol of surfactant, corresponding to 2.3 wt% of the sodium salt on polymer mass. In practice, the total surfactant dose is between 1.0 and 2.5 wt% on monomer, approaching a full monolayer, because monomer droplets and empty micelles consume a substantial fraction. During the feed, the particle surface area increases linearly with polymer mass, but the free surfactant concentration decays nonlinearly because new monomer continuously introduces additional oil-water interface. The delayed acid shot reduces the demand for external surfactant by generating a dense outer carboxylate layer that swells at high pH and increases the effective thickness of the stabilising layer. However, the delayed acid shot also introduces a destabilising window between 60 and 75% monomer addition, when the existing particles have high surface area but the acid has not yet been incorporated into the shell. In this window, the feed rate should be reduced by 10–15% and the agitator tip speed maintained at 2.5–3.0 m/s to avoid shear-induced aggregation.

The practical surfactant profile for a stable semibatch feed is not constant. It consists of an initial charge of 0.5–0.8 wt% on monomer, followed by continuous addition of 0.05–0.10 wt% per hour during the monomer feed, and a final stabiliser boost of 0.2–0.3 wt% on monomer before the delayed acid shot. The exact split is adjusted using the surface tension of the latex as a feedback signal; surface tension below 35 mN/m indicates free surfactant, while surface tension above 50 mN/m indicates depletion. The addition of nonionic alcohol ethoxylates at 0.5–1.0 wt% on monomer improves calcium ion tolerance and freeze-thaw stability but lowers the cloud point and may reduce the effectiveness of anionic surfactants if the nonionic surfactant competitively adsorbs. The combined anionic-nonionic package should be selected by measuring the adsorption isotherm on the actual latex particles, not by solution surface tension alone.

Shear-Produced Coagulum on Baffle and Thermowell Surfaces

Latex particles of 150–180 nm diameter are generally stable against perikinetic aggregation, but in a 12 m³ reactor with a retreat-curve impeller operating at 45 rpm the maximum shear rate near the blade tip is on the order of 2,000–4,000 s⁻¹. At total solids above 55 wt%, the latex becomes pseudoplastic and the apparent viscosity at 20 rpm measured according to ISO 2555:2018 rises from 300 mPa·s at 50 wt% to 1,200–1,800 mPa·s at 60 wt%. Particles near the baffle edges experience extensional shear and temporary deformation; if the surface coverage is insufficient, hydrophobic patches on the particle interact and form micro-flocs. These micro-flocs deposit preferentially on the thermowell, the upper surface of the baffles, and the manway nozzle, where the wall shear rate is low but the residence time is long. Mechanical stability tests following ASTM D1417-16 use high-speed stirring and report coagulum on a 150 µm screen; a stable semibatch styrene acrylic latex exhibits less than 0.05 wt% residue after the specified shearing time. Industrial batches at the edge of stability reach 0.2–0.5 wt% residue and show a visible film on the reactor wall after a single batch.

The location of shear-induced coagulum is influenced by impeller selection. A retreat-curve impeller with a blade tip radius of 800 mm running at 40–50 rpm produces a tip speed of 3.3–4.2 m/s, which is sufficiently high to generate local energy dissipation rates above 10 W/kg near the blade. The impeller discharge stream impinges on the glass-lined wall at a velocity proportional to the tip speed; wall shear rates above 1,000 s⁻¹ can remove weakly adhered polymer deposits, but shear rates below 100 s⁻¹ in the shadow zones behind baffles permit deposition. Baffles in a glass-lined reactor are typically fabricated from glass-coated steel and are 80–120 mm wide. The gap between the baffle and the vessel wall is 25–40 mm to avoid dead zones. If the gap is obstructed by polymer scale, the baffle no longer functions as an effective vortex breaker, and the vortex depth increases from 100 mm to more than 500 mm, drawing air into the latex. Air bubbles destabilise the latex at the air-water-polymer interface and increase coagulum by a factor of 2–5. The vortex depth is monitored by radar or by an upper impeller vibration sensor; a stable semibatch latex batch typically maintains a vortex depth below 150 mm at 45 rpm and 55 wt% solids.

Scale-up from a 2 L laboratory reactor to a 12 m³ production kettle changes the shear environment by more than an order of magnitude. A laboratory flask with a magnetic stir bar at 400 rpm has a maximum energy dissipation rate of less than 1 W/kg, whereas the production kettle impeller zone may exceed 10 W/kg. Consequently, a latex that is stable in the laboratory can form coagulum in production if the laboratory formulation does not include a high-shear mechanical stability check. The scale-up rule for agitator tip speed is not to keep the same tip speed but to keep the maximum energy dissipation rate below a critical value that is specific to the particle size and surface charge; for a 170 nm styrene-acrylic latex with a zeta potential of −48 mV, the critical energy dissipation rate is approximately 5–8 W/kg. The production impeller speed is therefore set at 38–42 rpm for a 1.6 m diameter retreat-curve impeller, which corresponds to a tip speed of 3.2–3.5 m/s. A higher tip speed can be tolerated only after the latex has been diluted to 45 wt% solids or after a post-feed stabiliser boost has been fully adsorbed.

Optimising the Delayed Acid Feed Position without Compromising Kettle Stability

The position of the acrylic or methacrylic acid feed during the semibatch monomer addition determines whether the acid is incorporated into the particle core or the particle shell. When the acid is fed as a separate aqueous stream during the last 25% of monomer addition, the carboxyl groups are located primarily in the outer 20–30 nm of the particle and are accessible to the continuous phase after neutralisation. This raises the zeta potential at pH 7.5 from −42 mV to −55 mV and improves mechanical stability. However, the same delayed acid feed creates a period between 60 and 75% monomer addition when the acid content of the aqueous phase is low and the existing particles have grown to a size at which their electrostatic stabilisation is weakest. The free surfactant concentration during this period must be maintained above 0.10 g/L, and the monomer feed rate should be reduced by 10–15% from the nominal 18.5 kg/min. A practical control strategy is to compensate for the delayed acid feed by adding a small portion of the anionic surfactant during this interval at 0.5–1.0 kg per tonne of remaining monomer. The acid stream itself should be diluted to 10 wt% and fed at a rate that does not cause a local pH below 3.0 at the drop point, because localised acidification can protonate the surfactant and create a transient hydrophobic zone. The acid feed line is therefore jacketed and angled into the impeller discharge, and the pH of the reactor contents is measured every 5 min according to ISO 976:2013.

In high-solids formulations containing 1.0 wt% methacrylic acid, the delayed acid feed can increase the neutralised viscosity from 1,500 mPa·s to 3,500 mPa·s at 20 rpm. This increase is not linear with acid content; it scales with the concentration of ionised carboxylate groups on the particle surface, and industrial formulations often use a combination of methacrylic acid and acrylamide-based functional monomers to limit the viscosity response. The operational boundary is reached when the carboxylic acid functional monomer content exceeds 1.5 wt% on monomer, because at that level the latex cannot be filtered through a 200 µm in-line screen at a pressure drop below 0.8 bar without dilution to 45 wt% solids. The use of methacrylic acid rather than acrylic acid reduces the degree of aqueous-phase polymerisation because methacrylic acid has a lower water-phase solubility; however, it also alters the reactivity ratio and can produce a more hydrophilic core-shell structure when fed late. Published data for the exact boundary in styrene-butyl acrylate formulations with a feed ratio of 55/45 styrene to butyl acrylate is limited, and the safe operating window should be established by a factorial experiment over 0.25–1.25 wt% acid and 50–60 wt% solids.

If the cook-down redox initiator is added as a concentrated aqueous solution, localised radical bursts can destabilise the latex surface and produce specks that only appear after storage. The cook-down stage following semibatch monomer feed is designed to reduce residual styrene from a typical feed-end concentration of 200–500 mg/kg to below 100 mg/kg before cooling and neutralisation. A redox couple of tert-butyl hydroperoxide and sodium formaldehyde sulphoxylate is fed as a 5.0 wt% solution through a dip tube at 0.5 kg/min over 20 min, with the kettle maintained at 80°C and the pH held above 5.5 by aqueous ammonia. The residual styrene concentration is determined by capillary gas chromatography according to ISO 13741-1:1998. Styrene depletion from the latex particle follows first-order kinetics with an apparent rate constant that depends on the copolymer glass transition temperature; at 80°C and a copolymer Tg of 25°C, the residual monomer falls from 300 mg/kg to below 50 mg/kg within 60 min. If the copolymer Tg is above 45°C, the diffusion coefficient of styrene in the particle drops by one to two orders of magnitude, and the residual monomer remains above 200 mg/kg unless the cook temperature is raised to 85°C or a steam-stripping step is used. The redox addition must be conducted under agitation; a temporary agitator failure during the redox feed can cause a monomer-rich layer to rise to the surface and later form a skin.

Redox reduction of residual styrene in a semibatch kettle is more than a post-polymerisation formality; the residual monomer acts as a plasticiser and can swell the latex particles, reducing the glass transition temperature of the particle surface and increasing tackiness. Tacky particles are more likely to deposit on the reactor wall and on the undersides of nozzles. The cook-down temperature is selected as a compromise between the kinetic demand for higher radical flux and the approach of the latex to its cloud point. For a 55 wt% solids styrene-acrylic latex, the cloud point of nonionic surfactant segments in the stabiliser package may be as low as 35–50°C in the presence of electrolytes, and heating above 85°C during cook-down can reduce steric stabilisation and produce gel particles. The operational limit is therefore 83–85°C for most anionic-nonionic surfactant systems, and the batch is cooled to 40°C before neutralisation to minimise the viscosity rise and avoid thermal ageing of the latex. A condenser load above 200 kW during this stage indicates that unreacted monomer is being refluxed rather than polymerised, which is a sign of insufficient initiator or low agitation.

Electrolyte Tolerance and Post-Neutralisation Viscosity Rise Are Nonlinearly Linked to Surface Carboxyl Distribution

Neutralisation of the latex from pH 4.8 to pH 8.0 with aqueous ammonia or sodium hydroxide increases the degree of carboxylate ionisation to above 0.95 and raises the latex viscosity from 400 mPa·s to 2,500–3,500 mPa·s for a 55 wt% solids copolymer containing 1.0 wt% methacrylic acid. The viscosity rise follows a non-linear relationship with the square of the surface carboxylate concentration and is more pronounced when ammonia is used because ammonium carboxylate groups hydrogen-bond with the polyether segments of the nonionic stabiliser. In a 15 m³ post-reactor blending tank equipped with a propeller agitator at 90 rpm, a viscosity above 2,000 mPa·s produces cavern formation around the impeller and poor mixing of defoamer and biocide dispersions. The addition of a low-shear acrylic thickener at 0.1–0.3 wt% based on latex can increase the viscosity further to 5,000–10,000 mPa·s, which is measured by ISO 2555:2018 at 20 rpm. Processing through a 200 µm in-line filter becomes difficult above 4,000 mPa·s, and the pressure differential across the filter rises from 0.2 bar to 1.2 bar. Batch-to-batch variance in the acid monomer feed position changes the surface acid density and therefore the neutralised viscosity; this is a common cause of downstream formulation nonconformance in architectural coatings when the low-shear rheology target is 105–110 Krebs units measured by ASTM D562-10.

Electrolyte tolerance is assessed by adding a 0.5 mol/L calcium chloride solution to the latex at 25°C and measuring the amount of coagulum formed after 24 h on a 150 µm screen. A latex with 0.25 wt% surface acid and a zeta potential of −48 mV can tolerate 0.2–0.3 mL of the calcium chloride solution per 100 g of latex before visible flocculation; a latex with buried acid tolerates less than 0.05 mL under the same conditions. The difference is critical because waterborne coatings are frequently let down with hard tap water containing 120–180 mg/L calcium carbonate equivalents, and the latex must remain stable during tinting with pigment dispersions containing anionic dispersants. If the latex is to be used in a formulation requiring 0.5 wt% of a cationic biocide, the anionic surface charge will be screened and a compatibility test with the exact biocide concentration is necessary; mixtures can form grit that passes through a 50 µm filter but later plugs spray tips of 0.28 mm diameter. The safe operating boundary for this styrene-acrylic latex is a neutralised pH of 7.5–8.5, a conductivity below 4,000 µS/cm, and a carboxyl surface distribution with at least 0.25 wt% surface acid on polymer.

Formulation compatibility tests are conducted with the actual letdown water quality, because the electrolyte tolerance measured in deionised water overestimates stability in hard water by a factor of 2–4. The use of 0.2 wt% sodium bicarbonate in the kettle charge is beneficial for pH buffering but contributes 1,200–1,800 µS/cm to the aqueous phase conductivity. If the latex is intended for a formulation containing high levels of anionic pigment dispersants, the order of addition is critical: the pigment dispersion should be added to the latex under agitation after the latex has been neutralised, and the coalescent should be added slowly at 0.1–0.2 kg/min per 100 kg latex to avoid solvent shock. The coalescent, typically a low-viscosity ester alcohol at 3–5 wt% on latex solids, can penetrate the surfactant layer and reduce the glass transition temperature of the surface region; if added too quickly it produces local particle swelling and the formation of micro-gel.

Batch-to-batch wall fouling data from production campaigns producing 55 wt% solids styrene-acrylic latex show a progressive decline in the clean-wall heat transfer coefficient from 360 W/m²K to 220 W/m²K after 8–12 batches when the monomer feed profile and acid shot are executed within the stable window. The fouling layer is composed primarily of polymer that has been destabilised by local shear, electrolyte concentration, and monomer swelling; it is not removed by a hot water rinse at 60°C but requires an alkaline cleaning cycle with 2–3 wt% sodium hydroxide at 80°C for 4–6 h. The glass-lined surface must then be passivated with a dilute acid rinse to restore the surface smoothness below 0.4 µm Ra. If the reactor wall is allowed to accumulate polymer scale thicker than 0.5 mm, the risk of glass-lining damage from thermal stress during cleaning increases, and the batch-to-batch variance in latex particle size can broaden by 10–20 nm. Operators should monitor the pressure drop across a 100 µm in-line filter during transfer to the blending tank; a pressure drop above 0.8 bar indicates that coagulum is present and that the latex should be diverted through a spare filter without stopping the transfer pump.

The final transfer of latex from the kettle to storage is a processing step that is often neglected in kettle stability studies but can induce significant shear. The use of a positive displacement pump with a gear-type head at 1,500 rpm can raise the latex temperature by 3–5°C and create a pressure drop of 2–4 bar across a 100 µm filter; this level of shear can destabilise a latex that was stable in the reactor. Transfer pumps should therefore be selected with a low-shear lobe or diaphragm configuration and operated at a tip speed below 3.0 m/s. Storage tanks should be fitted with a slow-speed anchor agitator at 20–30 rpm and maintained at 25–35°C; storage above 40°C accelerates hydrolysis of residual methacrylic acid and growth of bacteria, while storage below 5°C requires freeze-thaw stabilisation with 5–10 wt% ethylene glycol or propylene glycol on latex. The addition of freeze-thaw stabiliser must be made after neutralisation; if it is added to the reactor before the cook-down is complete, the glycol can partition into the polymer and alter the film formation behaviour. The latex should be filtered through a 150 µm screen before leaving the kettle, and the screen residue should be recorded for every batch as the primary kettle-stability metric.