Polycarbonate (PC) is specified for electrical enclosures, battery pack components, LED optics, and medical device housings where a UL 94 V-0 rating at 1.5 mm or 0.8 mm is a condition of product acceptance. Drying and regrind: how both quietly break a V-0 rating on polycarbonate is not a formulation anomaly or a supplier lot defect. It is a process-induced shift in melt rheology and flame-retardant efficacy caused by residual moisture above 0.02 % and by cumulative thermal history in returned material. The part may hold dimensions, pass a durometer check, and carry the correct resin identification code, yet fail vertical burn because flaming drips ignite a cotton indicator. The failure is routinely attributed to the material certificate when the root cause lies in a desiccant dryer with insufficient airflow or a regrind stream contaminated with fine particles below 1 mm. These two variables interact in a way that lowers melt strength, changes droplet size, and disrupts the char front.

Flame-retarded PC achieves V-0 through a combination of condensed-phase char formation and anti-drip behavior. When molecular weight drops, elongational viscosity drops with it, and the melt no longer resists the surface tension that initiates drips. A hydrolyzed or regrind-degraded part can therefore pass the afterflame-time portion of the vertical burn test but fail the cotton-ignition portion. This characteristic pattern is used by process engineers to separate a true flame-retardant formulation problem from a drying or regrind problem. The vertical burn test does not report viscosity, but it is sensitive to viscosity indirectly through dripping.

What Does a V-0 Rating Actually Require Under UL 94?

Under ANSI/UL 94 and the harmonized vertical burning method IEC 60695-11-10:2020, five specimens of 125 mm × 13 mm × test thickness are conditioned at 23 °C and 50 % relative humidity for 48 h and exposed twice to a 20 mm methane or natural gas flame applied for 10 s, with the second flame applied after the first afterflame ceases. The equivalent ASTM method designation is ASTM D3801-20. A V-0 classification requires that no individual afterflame time exceeds 10 s, total afterflame time for all ten flame applications does not exceed 50 s, afterglow time does not exceed 30 s, and no flaming or glowing drips ignite the surgical cotton indicator placed 300 mm below the specimen. V-2 permits the same afterflame limits but allows flaming drips that ignite the cotton indicator. V-1 permits longer afterflame times without cotton ignition. The rating is thickness-specific; a V-0 result at 3.0 mm cannot be transferred to a 1.5 mm wall section. Compounds for thin-wall electronic housings rely on flame-retardant systems—potassium sulfonate salts, phosphorus compounds, or silicone anti-drip additives—that alter the char layer or suppress droplet detachment. A process condition that shifts melt viscosity or degrades that additive package can move a V-0 part into V-2 without changing the base resin type.

Below the hopper, residual moisture in PC acts as a hydrolysis agent at melt temperatures of 300 °C to 320 °C. The carbonate linkage undergoes chain scission, reducing number-average molecular weight. The plant-floor indicator is melt volume-flow rate measured according to ISO 1133-1:2022 at 300 °C and 1.2 kg. A dried virgin PC with an MVR of 10 cm³/10 min may shift to 12–14 cm³/10 min after a single pass through a 25 mm twin-screw extruder when granulate moisture exceeds 0.05 %. The same molecular weight reduction lowers elongational viscosity and produces smaller, faster-moving droplets during the vertical burn sequence. Those droplets transfer heat to the cotton indicator more efficiently than larger, slower drips from a higher-viscosity melt. Consequently, a molded specimen may have a short afterflame time and still fail because the cotton indicator ignites. This is a classic V-2 failure signature caused by viscosity loss rather than by insufficient flame retardant.

Desiccant Drying Thresholds and Hydrolytic Chain Scission

Material suppliers and desiccant dryer manufacturers specify a maximum residual moisture of 0.02 % for PC before injection molding or extrusion. Desiccant wheel or twin-tower dryers delivering a return-air dew point of −40 °C and a hopper temperature of 120 °C for 4 h are the baseline. Airflow must be maintained at the dryer manufacturer’s rated volume; many technical bulletins reference 3.7 m³/h per kg/h of throughput. The return-air dew point must be measured at the hopper inlet, not only at the dryer outlet. A system can show −40 °C at the dryer cabinet while the hopper cone near the feed throat remains above −20 °C because of heat soak from the injection barrel. Under that condition, granulate near the feed throat can carry 0.04–0.06 % moisture into the screw even when the controller displays a compliant dew point. This installed-system failure mode is common on machines with 110 t to 180 t clamp force and hoppers holding more than 200 kg of material. Wet granulate entering the feed throat hydrolyzes rapidly; the reaction does not require long residence time above 260 °C. The anti-drip or flame-retardant package can also lose function through hydrolysis or phase separation when moisture is present as a polar phase. A drop of 15–25 % in ISO 179/1eA notched impact strength frequently precedes the flame-rating shift because both properties track molecular weight.

Moisture verification is frequently performed by weight loss on a laboratory balance, but that method lacks the sensitivity needed for PC. Karl Fischer titration according to ISO 15512:2019 is more appropriate. Many production lines use an in-line dry-air moisture analyzer or a return-air dew point probe. The critical measurement point is the feed throat, not the dryer outlet. A hopper loaded with 200 kg of PC can show a return-air dew point of −35 °C at the dryer cabinet while the bottom cone near the feed throat remains above −15 °C because of heat soak from the machine barrel. The resulting moisture content can exceed 0.04 % even though the dryer controller reports a compliant process. This is a common installed-system failure mode, not a monitoring error. Insulating the feed throat and confirming air sweeps at the hopper base are direct countermeasures.

Regrind does not enter the melt as a homogeneous stream. Sprues and runners from cold-runner tools contain particles below 1 mm, granulator dust, and occasional burnt slugs from hot-runner stalls. When a 25 % regrind fraction is blended with virgin PC, bulk density can vary by ±10 %, and hopper segregation delivers alternating melt viscosities to the screw. Low-viscosity regions created by fines or degraded polymer flow more readily and become drip sources during burning. The flame-retardant additive distribution is also disrupted: polar sulfonate salts can concentrate in the finer fraction or adhere to metal fines, leaving some cavities without the intended char promoter. A molding shop that returns crushed sprues and runners without screening below 3 mm or without controlling regrind at 10–20 % may observe a slow drift from V-0 to V-2 on the same mold and the same virgin resin lot. The mechanism is invisible in shot weight, cushion position, or cycle time. It first appears as sporadic cotton ignition in the vertical burn test.

Granulator blade condition also controls flame-rating stability. Dull blades or wide gaps generate fines rather than clean chips. Regrind with more than 10 % particle mass below 1 mm can stratify during conveying and feed early into the screw. Fine PC wets and plasticates faster than virgin pellets, producing alternating melt temperatures and viscosity. Metal fines from granulator wear can accelerate oxidative chain scission at processing temperatures above 290 °C. Magnetic traps and screen packs placed before the feed throat reduce this variable. The screening target is a regrind particle size between 1 mm and 3 mm, with fines below 1 mm kept to a minimum.

When Regrind Heat History Exceeds the Additive’s Thermal Stability Window

PC regrind that has passed through a 30 mm twin-screw extruder or a 120 t injection molding machine carries oxidative degradation from melt temperatures above 310 °C. The base resin may tolerate one or two re-melts with acceptable MVR shift, but the flame-retardant package may not. Phosphorus-based flame retardants can hydrolyze or volatilize above their stability limit; silicone anti-drip modifiers can coalesce under repeated shear; potassium sulfonate salts can deactivate through localized overheating in the compression zone. Each additional heat history lowers melt viscosity and shifts the UL 94 failure mode from afterflame time to flaming drip. A specimen from a 100 % regrind run may show afterflame times below 5 s and still ignite the cotton indicator. In production this is misinterpreted as a flame-retardant concentration problem, when the additive remains at the listed concentration but its morphology or chemical state has been altered. Regrind ratio, granulator screen size, and melt residence time are the controlling variables. Material supplier processing guides commonly limit PC regrind to 20 % or 30 % for V-0 applications. Above that threshold, published data for this specific configuration is limited because tool geometry, hot-runner volume, and regrind particle shape dominate the burn result.

Drying and regrind do not operate independently. Regrind particles have greater surface area-to-volume ratio than virgin pellets, absorb atmospheric moisture faster, and can carry moisture into an otherwise dry feed stream. A desiccant dryer sized for virgin pellets may not deliver equivalent drying to a regrind blend because bulk density and air channeling differ. High-fines regrind compacts in the hopper, reduces airflow, and creates pockets where the return-air dew point remains above −20 °C. The dryer outlet may read −40 °C while material in the lower cone stays wet. When a blend containing 30 % regrind reaches 0.06 % moisture, the melt viscosity loss is non-uniform. Specimens cut from the same shot can produce V-0, V-2, or HB because the cavity filled first with hydrolyzed low-viscosity material and later with less-degraded melt. This batch-to-batch variance is a hallmark of drying-regrind interaction, not a raw-material lot change. The corrective action is not to add more flame retardant; it is to re-dry to 0.02 % and re-screen the regrind through a 3 mm screen before dosing at a controlled fraction.

Flame-retardant additives in PC are not always chemically grafted to the chain. Sulfonate salts and certain phosphate esters can undergo diffusion-controlled migration toward the surface during slow cooling and during repeated heating. When regrind is dried at 120 °C for extended periods, polar flame retardants may enrich the surface layer. During the first flame application, the surface-rich layer chars quickly. The second flame application then reaches the depleted layer and produces a flaming drip. This two-phase burn response explains why a part can pass the first flame exposure and fail the second even though total flame-retardant content is unchanged. It is a migration kinetics problem in the polymer matrix, not a weighing error in the formulation.

The transition from V-0 to V-2 is controlled by the Weissenberg number at the burning drip front. As molecular weight decreases, the characteristic relaxation time shortens, the melt behaves more like a Newtonian fluid, and droplet breakup moves into a regime that ejects small particles. Small particles entrain more oxygen and transfer more thermal energy to the cotton indicator. Because the UL 94 test ignites the indicator when flaming drips fall, even a single hydrolytically thinned region in a multicavity part can dominate the result. Viscosity measurements at low shear may not capture this difference; oscillatory shear rheology from 0.1 rad/s to 100 rad/s shows the loss in storage modulus at low frequency more clearly. The crossover frequency between G′ and G″ moves to higher frequencies as molecular weight drops, and the shift is measurable in parts that have been dried improperly or molded from high regrind. This rheological signature is a stronger predictor of dripping than a single MVR value.

In injection molding with 110 t clamp force and a 30 mm screw, the screw cushion can hide viscosity loss until the cushion length begins to shorten. Shot weight may remain stable while the recovery time drops because low-viscosity melt wets the screw less efficiently. The result is unmelted granules at the end of the stroke, localized overheating in the barrel, and additive degradation. Parts then pass a dimensional check but lack the designed flame response. To maintain a V-0 rating, processors should not rely only on the material certificate; they should verify MVR per ISO 1133-1:2022 before molding, measure moisture by Karl Fischer titration according to ISO 15512:2019 or an equivalent dry-air moisture analyzer at the feed throat, and retain UL 94 specimens from start-up, mid-run, and end-of-run periods. The specimen set must include parts produced from the first ten shots after a hot-runner stall or brief shutdown because those shots carry the longest residence time and the highest probability of flame-retardant degradation.