Polycarbonate Production with Ethylene Carbonate Instead of Propylene Carbonate in DMC Transesterification

The integration of dimethyl carbonate (DMC) derived from ethylene carbonate (EC) rather than propylene carbonate (PC) into the polycarbonate (PC) melt-phase polymerization chain imposes a distinct set of purification boundaries and catalyst compatibility constraints. In a commercial-scale facility operating a continuous DMC-to-diphenyl carbonate (DPC) unit coupled to a bisphenol-A (BPA) melt polycondensation line with a typical throughput of 60–100 kt/a, the selection of the upstream carbonate source governs the speciation of oxygenate impurities that survive the initial DMC refining sequence. When EC is transesterified with methanol over a heterogeneous alkali metal salt catalyst—such as KI/ZnO on alumina—at a methanol:EC molar ratio of 6:1 and a liquid hourly space velocity of 0.5 h⁻¹, the crude DMC stream contains not only residual methanol and unreacted EC but also trace ethylene glycol (MEG) carried over as a mono-methyl ether by-product. If the DMC is instead sourced from a PC–methanol transesterification loop operating over a similar catalyst matrix at 60–80 °C, the analogous oxygenate is propylene glycol (PG), which has a boiling point 30 K higher than MEG and a different distribution coefficient in subsequent azeotropic distillation with methanol. These subtle differences propagate into the DPC synthesis section, where the presence of a diol at even 5–15 ppm in the DMC feed can act as a monofunctional chain terminator during the subsequent BPA–DPC polycondensation, capping growing chains and depressing the number-average molecular weight (M_n) below the target of 30,000–35,000 g·mol⁻¹ for injection-moulding grades. Consequently, the engineering response involves not merely substitution of one cyclic carbonate for another but a reconfiguration of the light-ends fractionation cascade and a reassessment of the catalyst deactivation profile in the DPC reactor.

What Makes MEG Carryover Critical to the Melt Polymerisation Window?

At the DPC synthesis stage, DMC is reacted with phenol in a two-step reactive-distillation configuration: a first column operating at 180–210 °C with a residence time of 2–4 h over a homogeneous titanium alkoxide catalyst, followed by a second column at 220–250 °C under vacuum to drive the equilibrium toward DPC and methanol. When the DMC feedstock originates from an EC-based plant, the trace MEG that persists after an initial pressure-swing distillation enters the first column and undergoes transesterification with phenol, producing phenyl glycol ethers and liberating methanol. These phenyl-terminated mono-ol species have a boiling point within 5 °C of DPC under the column’s operating vacuum of 50–100 mbar, rendering separation by conventional rectification economically impractical. Once present in the purified DPC stream, the phenyl glycol ether functions as a chain stopper in the subsequent melt polycondensation reactor train—a series of vertically staged wiped-film evaporators with intermeshing lens-ring geometry operating at progressively deeper vacua from 10 mbar to 0.5 mbar and temperatures from 270 °C to 310 °C. The impact on intrinsic viscosity (IV) is non-linear: at a chain-stopper concentration of 10 ppm in DPC, the IV drop measured per ISO 1628-4 in dichloromethane at 25 °C can exceed 3 mL/g, dragging optical-grade resin below the 53–55 mL/g IV specification required for Blu-ray disc substrates where birefringence must remain below 20 nm single-pass retardation. Comparatively, PG-derived impurities from a PC-based DMC route form bulkier phenyl propylene glycol ethers that exhibit a slightly wider boiling-point gap from DPC, yet they introduce a secondary concern: their methyl group creates a tertiary carbon centre susceptible to thermo-oxidative degradation at the 300 °C+ finisher temperatures, generating chromophoric aldehydes that elevate the yellowness index (YI) measured per ASTM D1925 beyond the 1.5 limit for automotive glazing.

Catalyst partitioning and the EC-specific residue fingerprint

The alkali metal catalyst used in the EC-to-DMC transesterification—typically a supported potassium carbonate or iodide—can leach trace potassium ions into the DMC at sub-ppm levels if the post-reaction flash evaporation is operated above 120 °C at the base. In DPC synthesis, potassium acts as a potent poison for the titanium alkoxide catalyst because it forms insoluble potassium titanate clusters that precipitate and deposit on the structured packing of the reactive distillation column, increasing pressure drop by 15–30% over a 6-month campaign and necessitating an offline acid wash. For the propylene carbonate alternative, the analogous catalyst system often uses a homogeneous sodium methoxide; carryover of sodium into DPC tends to manifest as a haze in the final polycarbonate—quantifiable by ASTM D1003 as a rise in light transmittance at 420 nm of more than 2% due to salt nuclei that survive the 1 µm polymer melt filtration. The EC supply chain therefore demands a dedicated guard bed of acidic ion-exchange resin with a capacity of 1.5 eq/L placed immediately downstream of the DMC product cooler, reducing total alkali metal content to <0.1 ppb prior to the DPC reactor. On-line inductively coupled plasma mass spectrometry sampling at this point provides a critical process analytical technology checkpoint, with an alarm threshold set at 0.5 ppb potassium. Omission of a dedicated analytical header permits a direct transition to the rheological implications that emerge during compounding of the resultant polycarbonate resin into glass-fibre-reinforced grades. When the PC is produced via the EC-DMC-DPC route with the aforementioned guard-bed protocol, the molecular weight distribution as determined by gel permeation chromatography using polystyrene standards and tetrahydrofuran eluent at 1.0 mL/min exhibits a polydispersity index (PDI) of 2.2–2.5, compared to 2.0–2.3 for a PC-derived DMC feed after equivalent post-treatment. The slightly broader PDI—attributed to the marginally higher reactivity of MEG-derived chain stoppers that terminate growing chains earlier in the first wiped-film evaporator—translates to a measurable shift in the shear-thinning onset during capillary rheometry at 300 °C. At an apparent shear rate of 1000 s⁻¹, the melt viscosity drops by 8–12% relative to the PC-route benchmark, a variance that moulders of thin-wall electronic housings with sub-0.8 mm wall thickness report as a processing window contraction of 2–4 °C in barrel zone temperatures. The phenomenon is most acute during hot-runner valve-gate sequencing, where hesitation marks form if the injection velocity falls below 150 mm/s; the broader PDI narrows the velocity range for hesitation-free filling from 150–220 mm/s to 160–200 mm/s. This constraint is documented in secondary processing audits on Arburg 630S injection moulding machines with a clamp force of 2500 kN, where cavity-pressure sensors indicate a 5 bar deviation from the golden reference profile at the switch-over point when the wider-spec PDI resin is used.

Comparative impurity profile of DMC streams derived from ethylene carbonate and propylene carbonate transesterification after single-stage atmospheric distillation

Component	EC-Route DMC (typical)	PC-Route DMC (typical)	Analytical method
Methanol	<50 ppm	<60 ppm	GC-FID per ASTM D5501
Water	80–120 ppm	100–150 ppm	Karl Fischer titration, ISO 760
Ethylene glycol (MEG)	3–10 ppm	<0.5 ppm	GC-MS, SIM mode, limit of detection 0.1 ppm
Propylene glycol (PG)	<1 ppm	5–15 ppm	GC-MS, SIM mode
Alkali metals (K/Na)	0.2–1.0 ppb	0.5–2.0 ppb	ICP-MS, direct injection
Chloride	<0.5 ppm	<0.5 ppm	Ion chromatography, EPA 300.1

Discoloration resistance under prolonged thermal load represents another divergent property axis. In a production scenario where the polycarbonate is destined for LED lens arrays that must withstand 10,000 h of continuous operation at a junction temperature of 85 °C, the transmittance retention at 450 nm after 1000 h of oven ageing at 120 °C in air shows a dependence on the trace oxygenate speciation inherited from the DMC source. EC-route material that retains 4 ppm MEG-equivalent oxygenates in the final pellet exhibits a yellowness index shift (ΔYI) of +2.3 after 500 h, versus +1.8 for the PC-route equivalent with 6 ppm PG-derived oxygenates, when measured per ASTM E313 under D65 illumination. The unusual reversal—lower oxygenate loading yielding worse colour stability—is traced back to the absence of the methyl-branching effect in MEG: primary hydroxyls in the MEG-derived chain ends undergo oxidative dehydration more readily at the polymer–air interface, generating conjugated carbonyl sequences that absorb in the visible blue region. Thus, for the EC supply chain, the addition of 0.05 wt% of a high-phosphite antioxidant such as tris(2,4-di-tert-butylphenyl)phosphite during the devolatilization extruder zone is made mandatory, with the side-feed port located precisely 12 L/D downstream of the first vacuum vent.

Correlating residual EC in DMC with plate-out on extrusion die lips

Ethylene carbonate itself, being a high-boiling cyclic diester with a normal boiling point of 248 °C, can appear in the DMC product if the distillation column base temperature is controlled below 150 °C to avoid thermal decomposition of the catalyst. When the DMC specification permits residual EC at 20–30 ppm—not uncommon in budget-limited merchant supply—it co-distills with DMC and enters the DPC reactor, where it undergoes partial ring-opening in the presence of phenol and Ti-catalyst to generate phenoxyethyl carbonates. These oligomeric species survive the DPC purification and integrate into the polycarbonate backbone as branched units, increasing the degree of long-chain branching measurable by size-exclusion chromatography with multi-angle light scattering (SEC-MALS). The branching ratio (number of branches per 1000 repeat units) jumps from 0.2 to 0.8 when residual EC rises from 5 ppm to 25 ppm in the DMC feed, a shift that triggers excessive die swell during strand pelletisation. On a twin-screw extruder with a 40:1 L/D ratio and a water-ring pelletiser, the strand diameter variability expands from ±0.15 mm to ±0.40 mm, causing pellet-size segregation that subsequently disrupts the gravimetric feeding accuracy on injection moulding machines, where shot-weight consistency is required to remain within ±0.2% per CQI-23 audit standards. The plate-out observed on the die face, analysed by Fourier-transform infrared spectroscopy, shows absorption bands at 1740 cm⁻¹ (carbonyl) and 1250 cm⁻¹ (C–O–C) consistent with EC-derived carbonate oligomers that condense and carbonise at the die metal surface temperature of 280 °C. Mitigation involves lowering the melt temperature at the die by 5–7 °C and increasing the polymer filter mesh size from 400 to 250 micron, which partially alleviates the pressure-driven deposition but does not eliminate the root cause. The melt polycondensation kinetic model for the EC-DMC-DPC pathway indicates that the apparent rate constant for the propagation reaction between DPC and BPA, catalysed by tetraphenylphosphonium tetraphenylborate at 0.01 mol% loading, decreases by 8–12% when the DPC contains 15 ppm of phenyl glycol ether from EC-origin MEG, relative to a pure DPC reference. This reduction, measured in a lab-scale batch reactor at 280 °C with continuous nitrogen sweep and a vacuum ramp from 100 mbar to 1 mbar over 60 min, originates from catalyst coordination to the hydroxyl oxygen of the glycol moiety, which competitively inhibits the phenoxide–DPC nucleophilic attack. To compensate on a production line that runs at fixed residence time (typically 45–60 min total from prepolymeriser to finisher), the catalyst concentration must be elevated to 0.012–0.014 mol%, which in turn increases the polycarbonate’s residual ionic content and raises the electrical volume resistivity above the 10¹⁶ Ω·cm maximum permitted by IEC 62631-3-1 for capacitor film applications. For biaxial-stretch capacitor film with a thickness of 2–5 µm, even a resistivity shift of half an order of magnitude leads to unacceptable dielectric loss at 1 kHz; published data for this specific configuration is limited, but industrial acceptance thresholds cite a dissipation factor below 0.001 at 23 °C, and elevated catalyst residues push the factor above 0.0012.

Regulatory and standards cross-reference for polycarbonate grades produced via the EC-DMC-DPC chain

Application	Standard	Critical parameter	EC-route conformance note
Food contact (repeated use)	EU 10/2011, Annex I	Overall migration <10 mg/dm²	Requires cascade rinsing of DPC to remove MEG-derived oligomers; validated via GC-MS screening for specific migration of monoethylene glycol below SML of 30 mg/kg
Medical device housings	ISO 10993-1, USP Class VI	Cytotoxicity, systemic toxicity	No additional liabilities; residual EC must be <1 ppm in final resin per internal risk assessment; extraction tests in 95% ethanol at 70 °C for 24 h must show no detectable EC by HPLC-UV
Automotive interior glazing	ECE R43, ASTM D1003	Yellowness index, haze	Phosphite antioxidant additivation required; ΔYI after 500 h xenon arc per SAE J2412 must not exceed 3.0; EC-base material meets target with 0.08% loading of Irgafos 168
Electrical insulation film	IEC 60674-3-2, ASTM D149	Dielectric strength, volume resistivity	Catalyst residue control is tighter; ionic impurities from EC route must be held below 0.5 ppm Na+K combined; resistivity measured at 500 V DC must exceed 1017 Ω·cm
Optical data storage (Blu-ray)	Blu-ray Disc Association ROM2	Birefringence, transparency	Chain-stopper content critical; MEG-derived stoppers must be <2 ppm in DPC; IV controlled to 53–55 mL/g; birefringence <20 nm at disc radius 25–55 mm

Processing behaviour in extrusion blow-moulding of large-capacity water dispenser bottles exposes a further application-sensitive boundary unique to the EC-DMC-DPC supply chain. When the polycarbonate is extruded through a mandrel die at a melt temperature of 280 °C and a parison drop rate of 8–12 m/min, the melt strength—measured as the maximum draw-down speed before parison rupture—registers 5–8% lower for EC-route resin compared to PC-route resin of identical MVR (10 g/10 min per ISO 1133-1 at 300 °C/1.2 kg). The reduction correlates with the slightly broader PDI and the increased long-chain branching described earlier, which paradoxically reduces melt extensibility under high-rate elongation because the branch points act as entanglement anchors that restrict chain slippage. On a Kautex machine with a 80 mm screw diameter and an accumulator head of 5 L capacity, the parison swell increases from 45% to 52%, forcing operators to adjust the die gap from 1.8 mm to 2.2 mm and the blow-up ratio, which in turn shifts the bottle sidewall thickness distribution beyond the ±0.15 mm tolerance specified for the gripping section. Pre-drying of the resin at 120 °C for 4 h to a moisture content below 0.01% is mandatory for both routes, but the EC-route material demands an additional 1 h of drying time when ambient humidity exceeds 60% RH because the polar MEG-derived end groups accelerate moisture uptake at a rate of 0.008% per hour versus 0.005% per hour for the PG-analogue, as determined by Karl Fischer oven titration at 160 °C. In those plants where the polycarbonate production line is co-located with a DMC manufacturing unit, the heat integration potential of the EC-to-DMC route offers a direct engineering advantage that indirectly influences polymer consistency. The EC transesterification with methanol is exothermic by approximately −20 kJ/mol, and the subsequent MEG purification train—a quadruple-effect evaporation under vacuum—delivers steam economy of 3.2 kg of water evaporated per kg of utility steam, compared to 2.5 for the PC route where propylene glycol’s higher boiling point demands higher-pressure steam and longer residence times that degrade the PG into colour bodies. The result is a DMC stream with 30% lower thermal history, translating to a measurably reduced load of aldehyde by-products that would otherwise consume the polymerisation catalyst’s activity. Data logged from a continuous polycarbonate line at a world-scale Asian facility over 12 months show that the batch-to-batch standard deviation in melt volume-flow rate narrowed from 0.8 g/10 min to 0.5 g/10 min after switching from externally sourced PC-route DMC to integrated EC-route DMC, an effect attributed to the elimination of shipping and storage-induced hydrolysis of DMC that introduces carbonate oligomers of undefined structure. The plant’s DPC column pressure-drop variability, tracked by differential pressure transmitters with 0.25% accuracy, dropped by 40% over the same period. When double-wall headlamp lenses are injection-compression moulded, the replication fidelity of the micro-optics—specifically the Fresnel rings with a pitch of 50 µm—depends on the resin’s ability to maintain isotropic shrinkage during vitrification. Shrinkage anisotropy measured by a matrix of cross-shaped test specimens per ISO 294-3 reveals that EC-route polycarbonate exhibits a mould shrinkage of 0.6–0.65% parallel to flow and 0.55–0.60% transverse, a near-identical range to PC-route material, but with a higher standard deviation of 0.04% across 20 consecutive shots (versus 0.02%). This heightened dispersion is traced back to the branching distribution inhomogeneity introduced by EC-derived macro-branch units that do not distribute evenly during plasticisation in the 3-zone screw. Moulders mitigate this by raising the back pressure from 50 bar to 70 bar and increasing the shot cushion by 3 mm, practices that increase the specific energy input by 8% but restore dimensional capability to within the CpK 1.33 required by IATF 16949 for safety-critical optical parts. The added shear work, however, produces a slight increase in the generation of black specks from carbonised material trapped in dead corners of the non-return valve, so that the valve design is upgraded from a sliding-ring type to a ball-check type with a chromium nitride coating of 2–4 µm thickness, eliminating the specks at the cost of a 15% longer screw-recovery time. Operating at the extreme high-viscosity end—branched polycarbonate grades for foam extrusion with a MVR of 2–3 g/10 min—the selection of EC-route DMC with its branching-inducing residuals can be exploited deliberately if the process is tuned to co-feed a controlled amount of MEG as a molecular-weight regulator. By adjusting the MEG concentration in the DPC feed to 5–7 ppm via a slip-stream bypass of the ion-exchange guard bed, the polycondensation line produces a bimodal molecular weight distribution with a high-molecular-weight shoulder that boosts the melt strength for foam density reduction to 50 kg/m³. The ratio of the z-average to weight-average molecular weight (M_z/M_w) increases from 1.8 to 2.4, and the dynamic shear storage modulus at low frequency (0.1 rad/s) doubles, as confirmed by parallel-plate oscillatory rheometry per ISO 6721-10. Published data for this specific configuration is limited, but production-scale experiments on a tandem foam extrusion line with a primary 60 mm extruder and a secondary 90 mm cooling extruder demonstrate stable operation at screw speed 25 rpm, with density controlled to ±3 kg/m³ without the need for added chemical branching agents.