PAN Precursor for Carbon Fibre: Which Acrylonitrile Quality Actually Matters

Acrylonitrile (2-propenenitrile, CH₂=CH–C≡N) is the dominant monomer feedstock for polyacrylonitrile-based carbon fibre precursor. In carbon fibre conversion, the precursor filament is not pure polyacrylonitrile but a random terpolymer containing 92–98 wt% acrylonitrile and 1–5 wt% comonomers such as methyl acrylate, itaconic acid, or methacrylic acid. The link between monomer quality and carbon fibre performance is not bulk purity alone; impurities that alter radical polymerisation kinetics, molecular weight distribution, spinline morphology, and solid defect density are the controlling variables. Specifications written for acrylic fibre or ABS resin are generally insufficient for carbon fibre precursor because carbon fibre tensile strength is flaw-controlled and responds to foreign particles, gel defects, and voids at the 1–5 µm scale. The following sections examine which acrylonitrile quality parameters have direct technical consequences in continuous precursor manufacture.

What Impurity Classes Disrupt PAN Chain Architecture?

Radical copolymerisation of acrylonitrile with methyl acrylate or itaconic acid is sensitive to chain transfer agents present in the monomer feed. Acetonitrile, propionitrile, and acetone carry abstractable hydrogen atoms and can terminate growing chains, thereby reducing number-average molecular weight and broadening polydispersity. Published transfer constants for every impurity in acrylonitrile polymerisation are not tabulated in public standards; however, the operational specification for acetonitrile is typically maintained at ≤300 mg/kg because continuous reactor data show measurable molecular weight drift when recovered monomer fractions exceed this threshold. In a dimethyl sulfoxide-based dope system, the preferred precursor molecular weight range is 70,000–150,000 g/mol with a polydispersity below 2.5. Low molecular weight fractions reduce spinline drawability and increase the defect population after oxidation. Impurities with carbonyl functionality are particularly relevant because they can introduce carbonyl end-groups that alter the cyclization exotherm and may act as weak chain transfer agents. For this reason, total carbonyl control is more important than total non-acrylonitrile organic content.

Acrylonitrile quality also affects comonomer sequence distribution. Itaconic acid is deliberately added at 1–2 wt% to reduce cyclization onset temperature and broaden the stabilisation exotherm. If the monomer feed contains uncontrolled acidity from acrylic acid or acetic acid, the effective comonomer acid content shifts, and the stabilisation window narrows. This is measured by differential scanning calorimetry at 10°C/min under air, with comparison to ASTM D3418. A precursor with a broad, lower-temperature exotherm is preferred for oxidation at 200–300°C; a sharp exotherm can produce runaway heat release and core-skin heterogeneity in the carbonised tow. The monomer quality parameter that matters is therefore not the absence of all acidic species but the controlled ratio of deliberately added acidic comonomer to adventitious acidity.

Inhibitor Variability and Induction Period Drift in Continuous Feed

Commercial acrylonitrile is inhibited with monomethyl ether of hydroquinone at 35–50 mg/kg to prevent polymerisation during storage and transport. In continuous polymerisation, MEHQ is not an inert diluent; it consumes initiator radicals and delays the onset of conversion. A shift from 35 mg/kg to 50 mg/kg changes the induction period in azobisisobutyronitrile-initiated solution polymerisation, although published data for exact induction time in specific production reactors is limited. The technical response is to measure inhibitor in the feed by UV-Vis absorbance at 280–300 nm and adjust initiator feed rate in real time. Without this adjustment, the residence time distribution in a continuous stirred tank reactor shifts, producing lot-to-lot molecular weight variation that cannot be corrected downstream.

Dissolved oxygen acts as a second retardation variable. Oxygen reacts with carbon-centred radicals to form peroxy radicals and can generate peroxide species that later initiate unwanted branching or gelation. In carbon fibre precursor polymerisation, the monomer is sparged with nitrogen to reduce dissolved oxygen, and storage vessels are blanketed. Published data for specific oxygen thresholds in acrylonitrile feed is limited, but the standard practice is to maintain oxygen below the detection limit of a polarographic probe. The combined effect of inhibitor variability and oxygen ingress is most visible as instability in the first reactor stage. Process analytical technology with online gas chromatography and Karl Fischer titration is therefore used to close the gap between certificate-of-analysis values and actual feed quality at the reactor nozzle.

Water in the acrylonitrile feed is often treated as a bulk diluent, but its technical significance changes when the dope enters the air gap of a dry-jet wet spinning cell. PAN/dimethyl sulfoxide dopes are sensitive to non-solvent content; water is a non-solvent that lowers the cloud point and can force phase separation before the filament reaches the coagulation bath. This phase separation is fixed during coagulation at 5–15°C and can produce a radial skin-core morphology that survives carbonisation. Maintaining water at ≤0.30 wt% in the monomer feed is therefore not an arbitrary purity target; it is the feed-side boundary for maintaining a homogeneous dope at 20–25 wt% polymer solids. In plants using recovered DMSO and recycled monomer, water accumulates rapidly unless the recovery column is operated with a separate dehydration loop.

When Carbonyls and Nitriles Accumulate in Recycle Streams

Acrylonitrile purification by extractive distillation rejects light and heavy by-products, but monomer recovery from polymerisation vents and dope devolatilisation can return acetonitrile, propionitrile, and acetone to the feed tank. These impurities concentrate because their separation from acrylonitrile depends on column profile, reflux ratio, and azeotrope behaviour. Published data for specific recovery column configurations is limited; however, the failure mode is well known: a certificate of analysis may report acceptable acrylonitrile purity by ASTM D5507 while the total nitrile impurity burden remains high. Carbon fibre precursor producers therefore request separate acetonitrile and propionitrile values by ASTM D5508, not a single gas chromatographic area percent. In recycled monomer, acetonitrile can exceed 300 mg/kg even when fresh feed is below 100 mg/kg. The technical consequence is a slow reduction in molecular weight that may not be detected until the precursor draw ratio falls or the carbonised tow tensile strength shifts.

Acetone and other carbonyls behave differently from saturated nitriles. They are more effective chain transfer agents in radical systems and can introduce carbonyl end-groups that participate in early cyclization. In a stabilisation furnace, carbonyl-containing chain ends may initiate cyclization at lower temperature, but this can produce an uneven distribution of crosslink density along the tow. The practical specification is therefore a total carbonyl limit, typically ≤200 mg/kg as acetone, rather than a compound-specific sum. This limit is not derived from cytotoxicity or environmental compliance but from polymerisation kinetics. Producers relying on recovered monomer without carbonyl scrubbing may still meet ASTM D5507 purity while generating precursor with inconsistent thermal behaviour.

Metals, Non-Volatile Residue, and Spinneret Pack Life

Iron and copper in acrylonitrile are not ordinary specification items for ABS or acrylic fibre, but they are critical for carbon fibre precursor. These metals interact with residual peroxides and acidic comonomers to generate crosslinked microgel particles. Once formed, these gels are not removed by coarse filtration and accumulate in the spinneret pack. A carbon fibre precursor spin pack typically uses absolute filtration of 3–5 µm before capillaries of 50–80 µm diameter. Published data for exact gel formation threshold in production dopes is limited; nevertheless, polyacrylonitrile producers specify total iron and copper at ≤0.1 mg/kg because metal-catalysed gel formation reduces spin pack life and increases filament break frequency. The field signature is a rapid rise in filtration pressure across the spin pack. When pressure exceeds 10 MPa, the spinneret must be replaced, causing line stoppage and waste precursor.

Non-volatile residue includes oligomeric species, stabiliser residues, and high-boiling by-products. In a spinning line, non-volatile residue above 50 mg/kg is associated with reduced pack life and increased broken filament count. The direct technical measure is the pressure drop across the dope filter; a production-scale spinning line can maintain stable pressure for several days with clean monomer, whereas high residue shortens filter service to less than 24 h. Carbon fibre precursor producers therefore specify non-volatile residue separately from total purity. This parameter is not visible in a standard gas chromatographic purity report but determines process continuity and defect density in the final carbon fibre tow.

Carbon fibre tensile properties are measured on impregnated tows according to ASTM D4018 or ISO 11566. A precursor defect at the 1–5 µm scale can become a critical flaw after oxidation at 200–300°C and carbonisation at 1,200–1,500°C. Failure analysis of broken filaments commonly identifies a void or inclusion as the initiating defect, not a polymer-chain irregularity. This means that the acrylonitrile quality parameters most directly linked to carbon fibre strength are those that generate solid particles, microgels, and spinline instabilities. Chain transfer impurities matter for molecular weight control; particulate and gel precursors matter for translation of molecular weight into tensile strength. The two classes must be controlled together, but they are not interchangeable in specification design.

Does Higher Purity Always Translate to Higher Carbon Fibre Strength?

Ultra-high acrylonitrile purity does not compensate for incorrect comonomer distribution or unstable inhibitor control. A feedstock with 99.8% acrylonitrile but variable MEHQ and high acetone can produce worse precursor than a feedstock with 99.5% acrylonitrile but tight carbonyl and inhibitor limits. The reason is that carbon fibre precursor performance is governed by kinetic reproducibility and defect density, not by the nominal area percent of the main peak. If the stabilisation exotherm is too sharp because the acidic comonomer level is too low, the oxidation furnace cannot process the tow at the required line speed. If the molecular weight distribution drifts because acetonitrile accumulates in recycled monomer, the spinline draw ratio shifts and the resulting fibre modulus scatters. In this context, the most useful certificate of analysis is one that reports impurity classes separately and links each value to a polymerisation response.

Process analytical technology in acrylonitrile feed systems is used to prevent specification drift. Gas chromatographs with flame ionisation detection measure acrylonitrile, acetonitrile, and acetone every 10–15 min at the feed tank; Karl Fischer titrators report water; UV-Vis analysers track MEHQ. These measurements are only meaningful when the specification limits are tied to polymerisation kinetics. Without that link, a plant can hold ≥99.5% purity and still experience molecular weight drift, spinline breaks, and scatter in carbon fibre tensile modulus. The technical requirement is therefore not maximum purity but controlled impurity profile: water below 0.30 wt%, acetonitrile below 300 mg/kg, acetone below 200 mg/kg, acidity below 20 mg/kg, MEHQ in a narrow band, and total metals below 0.1 mg/kg. These limits are not arbitrary; they are the feed-side boundaries for stable radical polymerisation, spinnable dope rheology, and low-defect carbon fibre precursor.