The liquid monomer 2-propenenitrile, CAS 107-13-1, is a polar vinyl compound with a normal boiling point of 77.3 °C at 101.3 kPa and a water solubility of approximately 7.35% by mass at 20 °C. Commercial conversion to polyacrylonitrile proceeds by free-radical propagation because the electron-deficient double bond is highly susceptible to radical attack while anionic and coordination catalysts are deactivated by nitrile coordination and chain-transfer side reactions. The propagating radical is highly reactive; the polymer precipitates in water or forms viscous solutions in dipolar aprotic solvents, and this behaviour fixes reactor choice, heat removal, and isolation. In the evaluation of Acrylonitrile Polymerisation: Radical Routes, Control Points and PAN Quality, radical-route selection is the first process variable because it determines molecular weight distribution, head-to-head and branching defect concentration, and residual monomer entering the PAN precursor. Industrial PAN homopolymers and copolymers are used in wet-spun acrylic fiber, carbon-fiber precursor, and as the nitrile component in ABS and NBR elastomers. In each use, the radical route and its control points establish hot-drawability, thermal stabilisation exotherm, and gel-particle burden of the final product.

Which Radical Route Governs Molecular Weight Distribution in PAN Precursor?

Acrylonitrile can be polymerised by four radical routes with industrial significance: solution polymerisation in dipolar aprotic solvents, aqueous precipitation polymerisation, aqueous suspension polymerisation, and emulsion polymerisation for nitrile-containing copolymers. The route determines the locus of polymerisation, viscosity, termination mode, and therefore the kinetic chain length distribution. In solution polymerisation using dimethyl sulfoxide, dimethylformamide, or sodium thiocyanate brine, the monomer and initiator are initially homogeneous and the polymer remains dissolved or highly swollen. Thermal initiators such as azobis(isobutyronitrile) are decomposed at 50–65 °C with residence times of 8–24 h in continuous stirred-tank or tube reactors. Molecular weight is controlled by initiator concentration, chain transfer to solvent, and dissolved impurities. A commercial dope can be filtered and deaerated directly for wet or dry-jet wet spinning, eliminating polymer isolation and redissolution.

Aqueous precipitation polymerisation is the dominant route for carbon-fiber precursor because it yields a filterable slurry with reduced solvent recovery burden and high molecular weight. Redox initiation with potassium persulfate and sodium metabisulfite at 30–60 °C produces propagating radicals in the aqueous phase; PAN precipitates as primary particles that agglomerate into a filter cake. The kinetic chain length in this route is controlled by the redox ratio, iron or copper ion content, pH, and chain transfer agents such as iso-propyl alcohol or dodecyl mercaptan. Uncontrolled redox acceleration can raise local temperature and induce bimodal molecular weight distributions through compartmentalised termination. Conversion targets of 75–95% are used because complete conversion consumes transfer agent and generates high-viscosity slurries with poor filterability. Molecular weight control is also influenced by polymerisation locus. The ratio of aqueous-phase monomer to precipitated polymer determines local monomer concentration and therefore propagation rate. Because the propagation rate coefficient of acrylonitrile in water is higher than in dimethyl sulfoxide at the same temperature, precipitation routes can achieve higher molecular weight at lower initiator concentration. This kinetic advantage is offset by the difficulty of controlling particle size and the need for continuous slurry withdrawal from a well-mixed reactor.

The molecular weight distribution measured by size-exclusion chromatography according to ISO 16014-1:2019 typically ranges from 1.7 to 3.0 for industrial PAN, with carbon-fiber precursor specifications often requiring polydispersity below 2.5. Solution polymerisation can produce lower polydispersity if chain transfer is gradual, whereas precipitation polymerisation may develop high-molecular-weight shoulders due to gel formation or local initiator overfeed.

Control of radical polymerisation of acrylonitrile is achieved by fixing initiator feed rate, temperature, residence time, chain transfer agent ratio, and pH within narrow bands. The kinetic chain length is approximately inversely proportional to the square root of the initiator concentration under steady-state conditions, but deviations occur when primary radical termination or degradative chain transfer to impurities becomes significant. For redox polymerisation, the Fe²⁺/Fe³⁺ or Cu²⁺ background must be controlled below 1 mg/L in the aqueous phase; otherwise, uncontrolled redox cycles shorten chain length and raise the ash content of the final PAN. In continuous stirred-tank reactors, the residence time distribution broadens the molecular weight distribution and can produce low-molecular-weight fractions that cause fuming during thermal oxidation. A common industrial control point is the viscosity number measured by capillary viscometry according to ISO 1628-1:2021; acrylic fiber dope targets are often 1.4–2.4 dL/g in dimethylformamide at 25 °C, while carbon-fiber precursor may require a narrower band of 1.6–2.0 dL/g to balance spinnability and drawability.

Dissolved oxygen must be reduced below 0.1 mg/L by nitrogen sparging before initiator injection. Oxygen acts as a retarder and forms peroxide oligomers that alter pH and redox balance; even partial oxygen ingress during the first 15–30 min of reaction can broaden the molecular weight distribution. Residual monomer control is also a radical-route boundary condition. Unreacted acrylonitrile is stripped by vacuum or steam after polymerisation to below 0.1% by mass in most precursor specifications; headspace gas chromatography is used. The residual level is not only a toxicological issue but also a processing defect source, because monomer droplets act as chain-transfer agents and create low-molecular-weight domains during thermal stabilisation. Polymerisation temperature excursions above 65 °C in aqueous precipitation promote branching and colour bodies; therefore, reactor jackets and external heat exchangers are designed for a heat removal capacity derived from the polymerisation enthalpy of approximately -76.5 kJ/mol of monomer reacted.

Residual Monomer, Branching Defects and Chain-End Chemistry

The radical propagation of acrylonitrile is highly regioselective for head-to-tail addition, but the small fraction of head-to-head and tail-to-tail insertions creates chain irregularities that alter thermal stabilisation. ¹³C NMR data on radical PAN show a tacticity distribution close to 25% isotactic, 50% heterotactic, and 25% syndiotactic triads under conventional conditions; the atactic configuration is sufficient for solution spinning because nitrile dipole interactions dominate interchain cohesion. Branching occurs by intermolecular abstraction of the α-hydrogen or the nitrile-adjacent methylene hydrogen during propagation and, to a lesser extent, by backbiting. These branches reduce solubility and increase gel formation; polymerisation temperature and chain transfer agent concentration are therefore manipulated to suppress branch density.

Chain transfer to solvent, especially dimethylformamide and dimethylacetamide, introduces low levels of terminal amide or amine species; the chain-transfer constant for dimethylformamide is low but non-negligible at polymerisation temperatures. Chain ends derived from persulfate initiators carry sulfate or sulfonate groups, which contribute ash and affect thermal cyclisation. Some precursor specifications require post-polymerisation washing with demineralised water to reduce ionic end groups. Quantitative assignments for individual defect structures in industrial continuous polymerisation are less frequently disclosed; published data for specific head-to-head content in commercial precipitation reactors is limited. The nitrile conversion index determined by Fourier transform infrared spectroscopy at 2244 cm⁻¹ relative to 1450 cm⁻¹ is used to monitor residual nitrile groups during thermal stabilisation; this index is a direct quality check on radical-route chain regularity. Headspace gas chromatography can quantify residual acrylonitrile down to 0.1–10 mg/kg depending on sample preparation and detector. Industrial slurry is washed with heated demineralised water at 60–80 °C and then dried under vacuum at 50–60 °C to reduce residual monomer without initiating cyclisation. Residual water above 0.5% by mass can cause hydrolysis of nitrile groups during storage or downstream extrusion and can produce ammonia and carboxamide defects.

When Aqueous Suspension Polymerisation Replaces Solution Routes in DMSO

Process economics frequently force a shift from solution polymerisation in dipolar aprotic solvents to aqueous suspension or precipitation routes. The replacement changes molecular-weight distribution and purification sequence. In suspension polymerisation, acrylonitrile monomer droplets are stabilised with polyvinyl alcohol or cellulosic suspending agents; oil-soluble initiators such as AIBN are used at 40–70 °C, and the polymer is obtained as a dry powder or beads after filtration and drying. The absence of a polymer-solubilising medium permits higher polymerisation temperature, but droplet size and suspending agent concentration must be controlled to prevent coalescence and reactor fouling. Suspension grade PAN typically has higher residual suspending agent and coarser particle size than precipitation slurry, making it less suitable for carbon-fiber precursor unless additional purification is applied.

When the objective is PAN quality for carbon fiber, aqueous precipitation is generally preferred over suspension because ionic initiator residues and suspending agent can be removed by washing, and the primary particle size can be controlled below 150 µm. However, precipitation polymerisation in water has lower maximum monomer concentration due to viscosity and heat transfer; reactor productivity per unit volume is therefore lower than solution polymerisation. Published data for direct property comparisons between suspension and precipitation PAN at identical molecular weight and comonomer composition is limited, but industrial-grade suspension PAN typically shows higher gel-particle counts and broader particle size distribution. The filtration pressure rise across a 10 µm absolute filter is a practical control point for dope quality, with excessive pressure rise indicating gel particles or suspending agent residues.

For carbon-fiber precursor, the radical polymerisation route is only the first stage; subsequent dope filtration, wet spinning, coagulation, drawing, and oxidation stabilisation translate chain architecture into fiber quality. The key PAN quality attributes are molecular weight, polydispersity, comonomer content, residual solvent, gel particles, ash, and filament tensile properties. Comonomer selection is a decisive bridge between polymerisation and oxidative stabilisation. Methyl acrylate or vinyl acetate is added at 6–9% by mass in acrylic textile fiber to improve solubility and drawability; for carbon-fiber precursor, itaconic acid or methacrylic acid is used at 0.5–5.0 mol% because the carboxylic acid groups lower the cyclisation exotherm onset and broaden the exotherm, reducing runaway heating during stabilisation. The comonomer distribution must be uniform; batch addition of itaconic acid faster than its consumption can generate composition drift and uneven stabilisation in the tow.

Thermal analysis by differential scanning calorimetry shows that PAN homopolymer cyclisation exotherm peak temperature is commonly near 290–320 °C at heating rates of 5–10 °C/min, but the incorporation of 1–3 mol% itaconic acid can reduce the onset to approximately 220–260 °C. The exact shift depends on comonomer type, concentration, and heating rate; published data for specific heating rates and precursor formulations is limited. Oxygen permeability and fiber diameter also modulate the exotherm, so precursor specifications include filament denier, cross-sectional shape, and coefficient of variation. Single-fiber tensile testing according to ASTM D3822-14 on precursor filaments typically reports tenacity of 0.4–0.8 GPa and elongation at break of 10–18% for wet-spun acrylic fibers; carbon-fiber precursor may be drawn to higher orientation with lower elongation.

Gel-particle burden is measured by filtration pressure rise or particle counting on filtered dope. A 10 µm absolute filter is typically used; an excessive pressure rise above a specified threshold before a fixed throughput indicates unacceptable gel content. Gel content can originate from reactor wall deposits, local overheating, or insufficient wash removal of suspending agents. Radical-route control points for gel control include reactor cleanliness, initiator addition point, and post-polymerisation washing; their accuracy is measured simultaneously as viscosity number, residual monomer, ash content, and filtration pressure rise.