Polyvinyl alcohol (PVA) serves as the primary protective colloid in the free-radical emulsion polymerisation of vinyl acetate to produce polyvinyl acetate (PVAc) dispersions marketed as white glue or woodworking adhesive. The present article addresses PVA as protective colloid in PVAc emulsion (white glue), with emphasis on grade selection, reactor behaviour, and limit states. Partially hydrolysed grades with a degree of hydrolysis between 87 mol% and 89 mol% are preferred because they combine solubility at 20–25°C with sufficient interfacial activity to stabilise polymerising droplets and growing latex particles. Characterisation of the finished dispersion typically follows ISO 2555 for Brookfield viscosity, ISO 3251 for non-volatile matter, and EN 204:2016 for classification of non-structural wood adhesives.

PVA Grade Selection and Interfacial Phenomena in VAc Emulsion Polymerisation

The protective colloid performance of PVA in a PVAc emulsion is governed less by molecular weight alone than by the combination of degree of hydrolysis, blockiness of residual acetate groups, and 4% aqueous solution viscosity. Commercial PVA grades used in white glue production are commonly specified with a 4% solution viscosity of 3–70 mPa·s at 20°C by the Höppler method according to DIN 53015 or by ISO 2555. Low-viscosity grades below 10 mPa·s reduce reactor torque but require higher use levels to achieve adequate particle coverage; high-viscosity grades above 25 mPa·s increase final dispersion viscosity and can restrict heat transfer in the reactor. Because PVA with residual acetate below 90 mol% dissolves in water at 20–25°C, cold-water make-down is possible; fully hydrolysed grades require heating to 85–95°C and a hold time of 30–60 min. The degree of hydrolysis determines the balance between hydrophobic adsorption onto PVAc surfaces and hydrophilic hydration in the continuous aqueous phase. Partially hydrolysed PVA at 87–89 mol% adsorbs at the PVAc–water interface with a plateau surface coverage reported in published literature on the order of 1–3 mg/m². Residual acetate groups provide hydrophobic anchoring, while the hydroxyl-rich segments extend into water and create a steric barrier against coalescence. Vinyl acetate radicals abstract hydrogen from the PVA backbone during polymerisation, generating graft copolymer PVA-g-PVAc. This grafted layer improves colloidal stability at total solids up to 55 wt%, but excessive grafting raises low-shear viscosity and may delay drying of the applied adhesive film.

On a 5,000 L glass-lined batch reactor equipped with a pitched-turbine agitator operating at 45–60 rpm, the PVA protective colloid is charged as an aqueous solution before vinyl acetate monomer addition. The reactor jacket is maintained at 70–75°C while the internal dispersion temperature is controlled at 65–68°C during semi-batch monomer feed. A representative charge contains 3–6 phr of PVA per hundred parts of vinyl acetate, 0.3–0.6 phr of ammonium persulfate initiator, and 0.2–0.4 phr of sodium bicarbonate as buffer. A semi-batch monomer feed over 180–240 min with a 30 min initiator addition delay is common. Under these conditions, visible coagulum formation in the vessel and on the condenser dome is observed when the PVA dosage falls below 2 phr or when the monomer feed rate exceeds the heat-removal capability of the jacket. Production records from agitated batch lines indicate that the jacket heat-transfer coefficient drops sharply when the Brookfield viscosity of the reacting dispersion exceeds 8,000 mPa·s at 25°C, and the drive motor can shift from 60% to 90% of rated torque within a single monomer feed increment. Foaming is controlled with a silicone-based defoamer at 0.05–0.2 wt% of total batch, but excess defoamer can displace PVA from the particle surface and create fish eyes in the final film.

What Causes Viscosity Drift When PVA Degree of Hydrolysis Falls Below 87 mol%?

When the degree of hydrolysis falls below 87 mol%, the PVA chain carries a higher residual acetyl content, and the polymer develops a reduced cloud point in aqueous solution. In a PVAc white glue formulation, this can appear as irreversible viscosity increase after storage at 40°C for 28 days, measured by ISO 2555. The mechanism involves partial desorption of the protective colloid from the PVAc particle surface, followed by weak bridging flocculation between particles. The dispersion may remain optically homogeneous for several days, but the low-shear viscosity can rise by more than 20% relative to the initial value, while the high-shear viscosity remains comparatively stable. Rheological measurements in root-cause investigations typically include a low-shear Brookfield spindle 4 at 12 rpm and a high-shear cone-and-plate measurement at 1,000 s⁻¹. A shift in the low-shear to high-shear viscosity ratio above 6 indicates flocculation. PVA grades with 87–89 mol% hydrolysis and a blocky distribution of residual acetate groups are less prone to this drift because the hydrophobic blocks remain firmly anchored even at elevated temperature. Fully hydrolysed grades above 96 mol% present the opposite problem: they are highly water-resistant and crystalline, but their reduced interfacial activity results in larger mean particle sizes and higher coagulum generation during synthesis. Therefore, white glue manufacturers tend to limit PVA degree of hydrolysis to 87–92 mol% unless specific wet-tack or water-resistance requirements demand otherwise.

If the Reactor Temperature Excursions Above 70°C During Semi-Batch Monomer Addition

Maintaining the reactor jacket at 70–75°C is normal for vinyl acetate polymerisation, but a dispersion-phase excursion above 70°C during the final 30% of monomer addition can destabilise PVA-stabilised particles. The adsorbed PVA layer contracts as hydrogen bonding with water weakens, and the refractive index difference between the particle and serum can increase as grafted PVA–PVAc migrates away from the surface. In practice, viscous fouling on the agitator shaft and thermometer pockets becomes visible within 15–30 min of an overshoot to 75°C in a reactor batch with 55 wt% target solids. Corrective action includes reducing the initiator feed rate to 50% of the programmed value, increasing the jacket cooling-water flow by 20–30%, and interrupting the monomer feed until the internal temperature returns to 65°C. Published data for the exact heat-transfer coefficients in PVA-stabilised PVAc reactors is limited, but production-scale observations indicate that recoverable dispersion viscosity changes after a short thermal excursion remain below 10% if the excursion lasts less than 10 min. Longer excursions above 75°C cannot be reliably reversed by post-addition of PVA solution because the desorbed colloid is no longer positioned at the particle–water interface.

In continuous stirred tank reactor trains, the relationship between PVA feed ratio and particle size distribution becomes nonlinear because grafted PVA–PVAc species accumulate in the first reactor and alter the effective stabiliser concentration. A two-stage continuous reactor with 3 h total mean residence time will generally produce a stable dispersion only when the PVA concentration in the first stage remains above the surface-coverage limit of roughly 1 mg/m². Below that threshold, large particles and bimodal distributions develop, and downstream filters plug with coagulum. The first reactor of a two-stage continuous train is usually operated at 25–35% conversion to avoid excursive viscosity build, while the second reactor completes conversion above 90%. If the first reactor temperature is not controlled, PVA-g-PVAc accumulates on the cooling coil and causes the pressure drop across the transfer pump to increase. Table 1 presents representative pilot-scale data for a vinyl acetate emulsion formulated at 50 wt% solids with ammonium persulfate initiation at 65°C; particle size was measured by laser diffraction according to ISO 13320:2020, and coagulum was determined as dry residue on a 100 µm sieve according to ISO 4576:1996.

The data show that a modest increase in PVA dosage from 2 phr to 4 phr reduces mean particle size from 1.8 µm to 1.1 µm and lowers coagulum by 87%, but it raises viscosity by more than 150%. A fully hydrolysed PVA at 98 mol% is a less efficient stabiliser under these conditions, producing a larger mean particle size and a higher sieve residue than the corresponding 88 mol% grade. These trade-offs explain why formulators do not simply maximise PVA content; the target is the minimum dosage that yields acceptable coagulum and storage stability while keeping viscosity below the limit of the let-down mixer and pumping system.

White glue formulation beyond the reactor involves let-down of the raw PVAc dispersion with additional PVA solution, plasticiser, defoamer, and preservative under low-shear mixing. In a 500 kg atmospheric mixer with an anchor agitator running at 30–40 rpm, a 10 wt% borax solution is sometimes added to raise tack and build viscosity through reversible PVA–borate crosslinking. Addition beyond 0.3 wt% borax on total formulation frequently causes a rapid viscosity increase above 50,000 mPa·s at 25°C, and the batch will not recover after dilution with water because the gel network is only partially reversible. Plasticiser selection is constrained by migration limits and by compatibility with PVA; dibutyl phthalate was historically used at 5–12 phr, but many current formulations use benzoate or citrate esters to meet regulatory restrictions. Filled white glues may include 10–25 wt% ground calcium carbonate with median particle size 2–10 µm; the presence of filler raises apparent viscosity and reduces clarity. When the adhesive is applied to beech wood at a spread rate of 150–200 g/m², open time depends on PVA molecular weight, ambient relative humidity, and the presence of extending starch. The final bonded assembly may be classified under EN 204:2016 according to the D1–D4 durability classes; most PVA-stabilised white glues without crosslinkers do not exceed D3 because water immersion at 20°C or 40°C plasticises the PVA-rich interphase.

Compliance of a PVA-stabilised PVAc white glue for packaging and wood bonding requires documentation against the appropriate end-use standard. For indirect food-contact adhesive applications, the formulation may be evaluated under FDA 21 CFR 175.105, which addresses adhesives used in contact with dry food or with food separated by a functional barrier. In the European Union, the PVA and PVAc components must be registered under REACH and must not contain restricted phthalates above the limits of REACH Annex XVII entry 51. For woodworking adhesives, EN 204:2016 clause 5.2 defines the minimum shear strength values for D1, D2, D3, and D4 classes after specified conditioning sequences. D3-grade white glue must withstand storage in a humid atmosphere and repeated water exposure without falling below the standard’s shear-strength threshold; D4 requires a resistant crosslinking system that PVA alone does not provide. A compliance matrix typically lists the adhesive’s Brookfield viscosity, solids content, pH, minimum film formation temperature, and the shear-strength value on beech after the relevant conditioning sequence. When using PVA as protective colloid in PVAc emulsion (white glue), the technical data sheet should state the PVA grade, degree of hydrolysis, and residual monomer content because these variables influence both the final bond performance and the environmental release profile.