Capacitor-grade biaxially oriented polyethylene terephthalate film is specified not by average thickness but by the statistical distribution of thickness across the web width and along the machine direction. On a commercial sequential stretching line, the melt is extruded through a slot die, pinned electrostatically to a cooled casting drum, then drawn in the machine direction by a series of heated rolls operating at draw ratios between 2.8:1 and 4.0:1, followed by transverse stretching in a tenter oven at ratios between 3.0:1 and 4.5:1. Thickness deviations generated in this sequence arise from die lip elastic deformation, asymmetric electrostatic pinning, drum temperature nonuniformity, and transverse stretch bowing. The resulting gauge profile is mapped by a scanning beta gauge or dual-sided capacitance sensor that traverses the web at 20–30 s intervals and reports cross-web thickness profiles with resolution near 0.01 µm. Under an applied direct-current or alternating-current test potential, the local electric field scales inversely with local thickness; a film with nominal thickness d_nom and local thickness d_actual experiences a field multiplier equal to d_nom/d_actual. A local thickness reduction of 5% therefore increases local dielectric stress by 5.3%, and a 10% reduction increases stress by 11.1%. Because dielectric breakdown is statistically initiated at the highest local field, the thinnest point on the web becomes the limiting dielectric weak point, regardless of average thickness.

Thickness measurement for capacitor-grade BOPET is governed by ASTM D374-16, which specifies dead-weight micrometer procedures but is supplemented on production lines by non-contacting beta gauges calibrated to reference films with known mass per unit area. The relation between mass per unit area and physical thickness requires an assumed density of 1.395 g/cm³ for oriented PET, and deviations from this density caused by crystallinity variations introduce linear errors in gauge mapping. A density error of 0.02 g/cm³ produces a thickness error of approximately 1.4% at constant mass per unit area. In practice, capacitor-grade film is often characterized by scanning beta gauges with a measurement spot size of 2 mm to 10 mm, so narrow striations below the spot size may be underestimated. This creates an additional breakdown risk because a 0.5 mm low strip can initiate breakdown while the scanning gauge reports acceptable average thickness. The usable web width for capacitor film is frequently narrower than the cast web width because edge bead and neck-in zones with thickness nonuniformity exceed tolerance limits.

Can Breakdown Strength Be Predicted Solely From Minimum Gauge in a Corona-Treated Film?

Breakdown strength values for BOPET are determined in accordance with IEC 60243-1 or ASTM D149-20 using cylindrical electrodes of 25 mm/75 mm geometry. The short-time test increases voltage at a controlled ramp, typically 0.5 kV/s to 2 kV/s, until failure occurs. Breakdown often occurs at the thinnest point, but only if the local field exceeds the intrinsic strength and no preexisting defect with a lower initiation threshold is present. The two-parameter Weibull distribution is used to model failure probability; the shape parameter, often designated β or m, reflects the narrowness of the breakdown distribution. Gauge variation creates a family of local field values whose relative spread corresponds to the thickness coefficient of variation. As gauge nonuniformity increases, the Weibull shape parameter declines because low-thickness regions dominate the low-voltage tail of the failure distribution. Published qualification studies report that reducing thickness tolerance from ±5% to ±2% can raise the Weibull shape parameter from below 15 to above 30 for metallized BOPET capacitors, though exact values depend on electrode area and voltage ramp. The presence of corona treatment complicates this prediction because surface polar groups and roughness alter charge injection and gas-discharge initiation, meaning breakdown strength is not a deterministic function of minimum gauge alone.

The field multiplier is a geometric consequence of the parallel-plate approximation and ignores electrode edge enhancement, but it demonstrates that gauge effects are nonlinear. Further, a local thickness defect does not need to span the full electrode area to reduce the observed breakdown voltage; a small-area thin region can initiate a partial discharge that erodes the metallized electrode and triggers clearing. The statistical interaction between gauge variation and defect density is therefore not additive but multiplicative when the electrode area is large enough to sample millions of potential weak points per capacitor element.

Across a sequential tenter line with working widths from 4.0 m to 8.5 m and line speeds from 150 m/min to 350 m/min, thickness uniformity is controlled by the interaction of extrusion pressure, die lip gap, casting drum speed, and preheat roll differentials. Die lip actuators with thermal expansion adjust local gap by as little as 0.01 mm, but the relaxation of the melt bank and edge bead cause the outer 100–200 mm of cast web to fall outside capacitor tolerance. In the machine-direction stretching unit, differential roll speeds are set between 2.8:1 and 4.0:1; local thickness reductions are amplified by neck-in. Transverse stretching in the tenter oven at 90–130 °C and draw ratios of 3.0:1–4.5:1 introduces further nonuniformity through clip spacing, rail bowing, and heat transfer differences. Heat setting at 200–230 °C under controlled relaxation of 2–8% modifies crystallinity and shrinks the film in width by 1–5%, and any nonuniformity in crystallinity across the web produces local thickness changes after shrinkage. The resulting orientation and crystallinity gradient means that gauge uniformity cannot be corrected after the tenter; subsequent slitting removes only the edge zones but cannot remove transverse bands. Published data for the exact scrap rate of capacitor-grade campaigns on specific tenter lines is limited, but production logs generally show that tight gauge tolerance is achieved at the cost of reduced line speed and more frequent die lip setpoint adjustments.

Capacitor Winding Stress, Metallized Electrode Margins, and Thermal Runaway Thresholds

Capacitor winding imposes additional requirements on thickness uniformity because winding tension interacts with local thickness to create hard and soft bands. A film with ±10% gauge variation can wind with alternating tight and loose layers, causing air gaps, pressure points, and local dielectric stress concentration at the edges of metallized electrodes. In metallized BOPET, the electrode is vacuum-deposited aluminum or zinc/aluminum with thickness between 20 nm and 60 nm. A thin spot in the base film may not itself fail, but when the film is wound, adjacent metallized layers become closer, increasing capacitance per unit area at that location and creating a local hot spot during operation or clearing. The self-healing mechanism clears the fault by evaporating electrode material around the breakdown channel, but repeated clearing at low spots increases capacitance loss and dissipation factor. Capacitor manufacturers therefore specify both total thickness tolerance and thickness profile across the wound width, and many qualification programs use segmented electrode patterns to localize clearing rather than allowing a single weak point to degrade the full capacitance. The dielectric loss power density in a capacitor film is given by P = 2πfε₀ε_rtanδE². A 5% thickness reduction increases the local field by 5.3%, which increases power density by approximately 10.9% because of the squared field dependence. Above 100 °C, the dissipation factor of PET rises due to increased dipole mobility and ionic conduction, and local thin regions can enter a thermal runaway condition in which heat generation exceeds heat dissipation through the film stack.

Test method selection changes the numerical breakdown value more than many formulation adjustments. The same roll of capacitor-grade BOPET tested under IEC 60243-1 with recessed spherical electrodes can produce a different breakdown strength than a flat 25 mm/75 mm cylinder system because the stressed volume and edge shielding differ. In addition, varying the voltage ramp from 0.5 kV/s to 2 kV/s shifts the breakdown value because partial discharge damage accumulates at higher ramps. The role of gauge uniformity cannot be separated from the test electrode area; a larger electrode area samples more thickness minima and increases the probability of encountering a low-strength defect. This is why capacitor-grade qualification often requires breakdown testing on multiple specimens with a fixed electrode area and a specified relative humidity below 50%, because absorbed moisture lowers surface resistivity and can reduce the observed breakdown voltage by more than 10%.

These acceptance bands are representative rather than universal; individual capacitor manufacturers and film suppliers may use different limits depending on capacitor voltage class, segmentation pattern, and operating temperature range. The thickness profile measurement itself should be executed after conditioning at 23 °C and 50% relative humidity for at least 24 h because PET absorbs moisture and swells slightly, shifting thickness readings by a fraction of the tolerance. Failure to standardize conditioning can lead to batch-to-batch false rejections or false acceptances on a production line.

When Thickness Tolerance Is Tightened Below ±2%, the Economic and Physical Constraints Change

Tightening total gauge variation to ±2% or better forces the extrusion die into a narrower gap band, which may require automated die lip actuators with response times below 5 s and position repeatability of 0.002 mm. The casting drum speed and temperature uniformity must be maintained within ±1.0 °C across the web width. In the machine-direction stretching unit, roll speed differentials must be held within ±0.1% to avoid long-wavelength thickness modulation. Transverse stretching rails may require convex or concave profile adjustments to compensate for bowing. The cost rises because the line speed is often reduced by 10–30% to allow control loops to stabilize, and because the frequency of profile scans is increased. Scrap generated during transitions can increase from below 5% to above 15% of output. The resulting film may have a slightly different crystallinity profile because lower draw rates or higher heat setting temperatures are used. Therefore, tightening gauge tolerance is not a simple specification change; it alters the orientation state, dielectric loss tangent, and thermal shrinkage. Published data for the exact cost multiplier of achieving ±1% gauge tolerance on a given tenter line is limited because such data is held as proprietary by film producers, but equipment supplier technical bulletins indicate that automatic profile control and higher-resolution gauges are mandatory below ±2%.

Capacitor-grade BOPET with an absolute gauge variation below 0.5 µm is not generally specified for wound capacitors with metallized electrodes unless the application uses a liquid dielectric or thick segmented electrode with current-limiting fusing. At relative humidity above 60%, PET absorbs moisture and hydrolyzes under thermal load; operational boundaries for high-humidity service are therefore set by the need to avoid accelerated molecular weight reduction that lowers both dielectric strength and tensile properties. If the film is pretreated with corona for metallization adhesion, the treatment intensity must be limited because excessive surface oxidation can increase surface charge injection and reduce breakdown strength at thin spots. Avoid combination with amine-based slip additives or plasticizers that can migrate to the interface and reduce the self-healing clearing resistance. No additional surface coating should be applied to capacitor-grade BOPET without requalification under IEC 60243-1 and ASTM D149-20, because coatings alter the discharge inception voltage and can mask the true breakdown location. These operational boundaries define the practical tradeoff between gauge uniformity and dielectric withstand: thinner, tighter-gauge film raises energy density, but process control, winding mechanics, and thermal stability impose hard limits that cannot be overcome by average thickness specifications alone.