Bisphenol A diglycidyl ether resin with an epoxide equivalent weight of 188 g/eq to 192 g/eq requires 86.5 phr to 89.4 phr of methyltetrahydrophthalic anhydride for a stoichiometric 1.00:1.00 anhydride-to-epoxide molar ratio, based on an anhydride equivalent weight of 166 g/eq to 168 g/eq for the commercial MTHPA isomer mixture. This calculation represents only a formal starting point because the actual reactive ratio is altered by tertiary amine-catalyzed esterification, residual water-driven anhydride hydrolysis, and temperature-dependent ether homopolymerization of excess epoxide. Industrial vacuum casting operations therefore compress the nominal mixing ratio into a band of 0.80:1.00 to 0.95:1.00, which corresponds to 70 phr to 83 phr for an EEW of 190 g/eq, while retaining sufficient anhydride for complete ester cross-link formation and limiting free anhydride exudation. Formulations below 0.75:1.00 generally exhibit a glass transition temperature below 90 °C according to ISO 11357-2:2020, a coefficient of linear thermal expansion greater than 75 × 10⁻⁶ K⁻¹ according to ISO 11359-2:2021, and tensile strength below 50 MPa according to ISO 527-2:2012. Above 1.00:1.00, the unreacted anhydride plasticizes the cured network, reduces volume resistivity, and increases equilibrium moisture absorption beyond 1.0% after 24 h water immersion according to ISO 62:2008. The lower process boundary at 0.80:1.00 is therefore driven by thermomechanical integrity, while the upper process boundary at 0.95:1.00 is driven by dielectric stability and surface quality in cast electrical components.

How Does Excess Anhydride Affect Thermal Oxidative Stability in Electrical Castings?

Accelerated aging of MTHPA-cured DGEBA castings under IEC 60068-2-78 conditions at 85 °C and 85% RH demonstrates that formulations containing an anhydride-to-epoxide ratio above 1.00:1.00 develop carboxylic acid species by anhydride hydrolysis at the casting surface, which increases the dissipation factor in the 50 Hz to 1 kHz range and reduces volume resistivity by one to two decades from an initial value near 1 × 10¹³ Ω·cm when measured according to IEC 60093. This degradation mechanism becomes more pronounced in unfilled castings because the absence of a mineral filler leaves a higher fraction of the free anhydride available for moisture uptake and subsequent acid formation; typical silica-filled formulations at 60 wt% filler partly mask the effect through barrier tortuosity but do not eliminate it when the mixing ratio exceeds 1.05:1.00. For high-voltage instrument transformers and busbar supports, process specifications commonly derate the upper ratio to 0.90:1.00 to maintain dielectric strength above 18 kV/mm following IEC 62631-3-1:2016 testing after 1000 h of humid aging. Thermal oxidative stability under air aging at 150 °C for 500 h shows an additional effect: ester carbonyl absorbance in Fourier-transform infrared spectroscopy at 1730 cm⁻¹ remains stable, but the carbonyl shoulder associated with free anhydride at 1785 cm⁻¹ increases with increasing ratio above 1.00:1.00, indicating residual unreacted species rather than decomposition of the cured network. The upper limit in electrical casting formulations is therefore not a single fixed point but a function of the aging environment, filler content, and end-use dielectric acceptance criterion.

On production-scale vacuum casting lines, the viscosity of the pre-reacted blend at 40 °C is a more immediate ratio-dependent variable than the final glass transition temperature. MTHPA has a neat viscosity of 50 mPa·s to 80 mPa·s at 25 °C, and when combined with DGEBA at a ratio of 0.90:1.00, the blend viscosity typically falls between 500 mPa·s and 1200 mPa·s depending on the EEW and isomer distribution of the anhydride; decreasing the ratio to 0.70:1.00 increases the blend viscosity by 10% to 20% because the unreacted DGEBA fraction is more viscous than the anhydride. In filled casting compounds containing 50 wt% to 65 wt% silane-treated silica, viscosity at 40 °C rises to 3000 mPa·s to 8000 mPa·s, and the processing window narrows further because vacuum degassing below 1 kPa requires viscosity below approximately 10,000 mPa·s to permit bubble removal from 25 mm thick molds within 15 min. Batch-to-batch variation in MTHPA isomer composition, particularly the ratio of 4-methyltetrahydrophthalic anhydride to 1-methyltetrahydrophthalic anhydride, can change the blend viscosity at a fixed mixing ratio by ±5% and has been observed to affect gel time measured by ASTM D2471-19. For such reasons, manufacturers of high-voltage cast-resin transformers control the anhydride-to-epoxide ratio not only by mass but also by acid value titration and near-infrared monitoring of the anhydride carbonyl band during inline mixing.

Accelerator-Facilitated Esterification Enables a Narrow Processing Window in Thick Sections

Addition of 0.5 phr to 2.0 phr of benzyldimethylamine or 0.5 phr to 1.5 phr of 1-methylimidazole shortens gel time and permits lower cure temperatures, but also narrows the allowable anhydride-to-epoxide ratio because the accelerator catalyzes both alternating esterification and ether homopolymerization of residual epoxide when the ratio is below 0.75:1.00. In thick casting sections exceeding 25 mm, the exotherm from a 50 kg batch mixed at 50 °C can raise the core temperature above 160 °C when accelerator concentration exceeds 1.5 phr at a near-stoichiometric ratio of 0.95:1.00, causing internal cracking, color darkening, and localized decomposition around embedded conductors. Production records from vacuum casting equipment with 80 L stainless steel mixing vessels indicate that the practical accelerator concentration is limited to 0.8 phr to 1.2 phr for ratio settings above 0.85:1.00 to maintain peak exotherm below 140 °C when mold temperature is controlled at 60 °C. At ratios below 0.80:1.00, the same accelerator concentration promotes excess epoxide homopolymerization, forming polyether segments that lower glass transition temperature and increase moisture absorption despite the apparent crosslink density; the measurable Tg may drop below 95 °C even though the gel time is shorter. Thus the processing window in accelerator-catalyzed MTHPA systems is constrained by a lower ratio boundary near 0.80:1.00 to suppress polyether formation and an upper boundary near 0.95:1.00 to limit exotherm and free-anhydride side reactions.

Property data for unfilled DGEBA/MTHPA castings cured for 2 h at 100 °C followed by 4 h at 150 °C show non-linear degradation at the edges of the permitted ratio band. The steepest cliff occurs between 0.75:1.00 and 0.70:1.00, where the maximum attainable glass transition temperature falls from 105 °C to below 80 °C and flexural strength according to ISO 178:2019 falls by more than 30%. The comparative data in the following table are representative ranges compiled from publicly available supplier data for unfilled castings; specific lots vary with resin EEW, accelerator type, and cure schedule.

The pronounced drop below 0.75:1.00 is consistent with a transition from a network dominated by diester cross-links to a less regular structure containing unreacted epoxide clusters and polyether domains. Above 1.00:1.00, the decline in strength and Tg is less abrupt but the increase in water absorption is directly relevant to outdoor electrical insulation. Published data for the specific configuration of 0.70:1.00 with high-temperature post-cure is limited, and end-users should verify strength retention under their own cure schedule before adopting such a ratio.

When Moisture Uptake Pushes the Upper Ratio Boundary Toward Surface Blushing

Surface blushing in MTHPA-cured castings is an observable failure mode that places an upper practical limit on the anhydride-to-epoxide ratio in humid manufacturing environments. Blushing develops when residual cyclic anhydride at the part surface reacts with atmospheric moisture to produce a tacky layer of methyltetrahydrophthalic acid half-esters, which then crystallizes as a white haze after solvent wiping or temperature cycling. In formulations above 1.00:1.00, the free anhydride concentration is sufficient to produce visible haze after 24 h at 25 °C and 60% RH, while formulations below 0.90:1.00 generally do not blush under the same exposure. The issue is intensified when molds are filled at low temperatures below 40 °C, because phase separation of unconsumed MTHPA at the resin-air interface occurs before gelation; this behavior has been observed in production-scale vacuum casting of bushings with 30 mm wall thickness. Infrared analysis of blushed surfaces shows a strong carbonyl doublet at 1785 cm⁻¹ and 1730 cm⁻¹, corresponding to free anhydride and ester, whereas the interior material shows only the ester carbonyl. The standard mitigation is to maintain the ratio at or below 0.90:1.00 for unfilled or lightly filled castings exposed to uncontrolled humidity, and to post-cure at 120 °C to 150 °C for at least 2 h to consume residual anhydride before the part leaves the heated mold. If the end-use requires reduced water absorption below 0.8% per ISO 62:2008, experience from supplier bulletins indicates that the upper ratio should not exceed 0.85:1.00 for silica-filled systems because the filler increases tortuosity but does not chemically bind free anhydride.

What Limits the Lower Anhydride Ratio in Alumina-Filled Casting Compounds?

Alumina-filled casting compounds present a different lower-ratio boundary because the high filler content creates an interfacial demand for polar anhydride species. In DGEBA/MTHPA formulations loaded with 60 wt% to 70 wt% alumina trihydrate or fused alumina, the effective anhydride-to-epoxide ratio at the resin-filler interface is lower than the nominal bulk ratio due to preferential adsorption of anhydride onto hydrated alumina surfaces and partial reaction with surface hydroxyl groups. At a nominal ratio of 0.80:1.00, differential scanning calorimetry according to ISO 11357-2:2020 often shows a reduced cure enthalpy and a Tg that is 5 °C to 10 °C below the unfilled control, indicating that the true ratio in the matrix is closer to 0.70:1.00 or lower. The practical lower bound for alumina-filled castings is therefore raised to 0.85:1.00 unless the filler surface is first silane-modified with a hydrophobic epoxysilane, which reduces hydroxyl interaction and permits operation at 0.80:1.00 without loss of Tg. Underfilled or non-silane-treated formulations below 0.80:1.00 exhibit incomplete filler wetting, increased viscosity above 12,000 mPa·s at 40 °C, and entrapped air defects when degassed under 1 kPa vacuum for 20 min. Published data for the specific configuration of fumed alumina at 70 wt% loading in MTHPA-cured DGEBA is limited, and the stated boundary should be validated by cure rheology and dielectric strength testing on the actual filler batch.

Compliance Matrix for MTHPA-Cured Casting Resins Under International Standards

Regulatory and performance compliance for MTHPA-cured casting resins is assessed against a matrix of mechanical, thermal, electrical, and environmental standards. The following checklist table represents the most commonly cited test methods and typical acceptance criteria for unfilled or silica-filled DGEBA/MTHPA formulations used in indoor electrical insulation; the exact limits are set by the end-use specification, and the table should not be read as a universal formulation target.

Compliance testing is normally performed on cured specimens machined from degassed castings, and the ratio-dependent variability in properties means that a formulation with an anhydride-to-epoxide ratio below 0.80:1.00 may still meet tensile strength requirements while failing the 105 °C Tg criterion; conversely, a formulation above 1.00:1.00 may pass initial electrical tests but fail humid aging limits. Test values in the table should be treated as internal screening criteria, and the specific test report must state the exact batch ratio, filler treatment, cure schedule, and specimen conditioning history because these variables affect the measured result more than the standard designation alone.