Figure 1. Thermal camera imaging of thick-section potted power modules frequently reveals the potting layer as the dominant thermal resistance - a variable absent from most initial thermal models.
The thermal model showed junction temperature at 95°C under full load. The assembly runs at 118°C. Component returns begin at 14 months - IGBT gate threshold drift, electrolytic capacitor failure, solder joint fatigue concentrated around the high-dissipation zone. The engineering team investigates component quality. The PCB copper weight. The heatsink contact resistance. Nobody opens the thermal model and adds a line item for the epoxy potting compound between the component and the enclosure wall. That line item, had it been included, would have shown a thermal resistance contribution of 0.04–0.06 K/W per cm² at standard potting thickness - enough to account for most of the discrepancy between model and measurement.
Standard epoxy potting compounds at 0.5 W/m·K are not thermally neutral in thick-section designs. They are thermal insulators with a flame-retardant function. Treating them as thermally transparent in a power electronics thermal model is the cause, not the symptom, of the junction temperature problem.
The Thermal Resistance of a Potting Layer: A Quantitative Assessment
Thermal resistance through a planar layer is calculated as R = t / (k × A), where t is layer thickness, k is thermal conductivity, and A is cross-sectional area. For a standard potting compound at k = 0.5 W/m·K:
At 10 mm thickness, 1 cm² area: R = 0.010 / (0.5 × 0.0001) = 0.20 K/W
At 15 mm thickness, 1 cm² area: R = 0.015 / (0.5 × 0.0001) = 0.30 K/W
At 20 mm thickness, 1 cm² area: R = 0.020 / (0.5 × 0.0001) = 0.40 K/W
These are not negligible values. A power module dissipating 5 W through a 15 mm × 1 cm² potting section experiences a temperature rise of 1.5°C across the potting at 0.5 W/m·K - which sounds small until the cross-sectional area is 2 cm², the dissipation is 20 W, and the hot spot is concentrated. In dense power module layouts where multiple dissipating components share a potted volume, the cumulative thermal resistance of the potting layer contributes 15–30°C to the junction-to-ambient budget in designs where this contribution was not modeled.
At k = 1.5 W/m·K, the same geometry produces one-third the thermal resistance. Whether that reduction is meaningful depends on what the other resistances in the thermal path are - if the junction-to-case resistance of the component dominates, improving the potting compound provides little benefit. The potting layer thermal resistance is most consequential when it is the dominant term in the path, which occurs in thick-section designs with relatively low-resistance cooling paths on the outer surface.
Figure 2. In a 15 mm thick potting section, switching from 0.5 W/m·K to 1.5 W/m·K reduces the potting layer's thermal resistance by approximately two-thirds. Whether this reduction is significant depends on the relative magnitude of other resistances in the thermal path.
Where Thick-Section Potting Thermal Resistance Dominates
Not every potted assembly is sensitive to the thermal conductivity of the potting compound. The following design conditions identify cases where the potting layer is likely to be a dominant thermal resistance:
Potting section thickness above 8–10 mm. Below this range, the absolute thermal resistance of the potting layer is typically small relative to other resistances in the path. Above this range, particularly when the cooling surface is the outer enclosure wall, the potting layer frequently becomes the dominant term.
Power dissipation density above 1 W/cm² within the potted volume. At low dissipation density, the temperature differential across the potting layer stays within acceptable limits even at 0.5 W/m·K. As power density increases, the same thermal resistance produces proportionally larger temperature differentials.
Cooling path topology where heat must conduct through the potting layer to reach the cooling surface. In assemblies where a heatsink or enclosure wall is the primary cooling path and the potted volume separates the component from that surface, there is no bypass path - 100% of the component's dissipated heat must conduct through the potting. In assemblies where the component can cool through leads, a PCB copper plane, or direct contact with the housing, the potting contribution is reduced.
Continuous duty applications with no thermal cycling relief. A component that continuously runs near its junction temperature limit accumulates degradation linearly. A 15°C reduction in junction temperature - achievable through potting compound selection in some geometries - can double component service life under Arrhenius-model degradation.
Why Standard Epoxy Thermal Conductivity Is Low and What Raises It
Unfilled and lightly filled epoxy resins have thermal conductivity in the range of 0.15–0.25 W/m·K. This is inherent to the cross-linked polymer matrix - polymer chains are poor thermal conductors because heat transfer in amorphous polymers is primarily through vibrational energy transfer along chains, which is inefficient compared to crystalline materials. The 0.5–0.7 W/m·K values typical of standard flame-retardant epoxy potting compounds represent some filler content - usually the same inorganic fillers that contribute to the flame-retardant function - but at filler loadings optimized for processability and flame performance, not for thermal conductivity.
Reaching 1.5 W/m·K requires significantly higher filler loading with thermally conductive inorganic particles - typically aluminum hydroxide, alumina, or boron nitride at volume fractions above 50%. The trade-off is a steep increase in base component viscosity: a formulation delivering 1.5 W/m·K will typically have a base viscosity in the range of 500,000–1,500,000 cps at 25°C, compared to 4,000–10,000 cps for a standard flame-retardant system. This viscosity range requires mechanical pre-mixing, and preferably heated dispensing at 50°C, to achieve void-free fill in confined potting cavities. The thermal conductivity gain is real, but it comes with a process discipline requirement that is not present in standard epoxy potting.
A critical but frequently overlooked point: the thermal conductivity of a highly filled system is only achieved when the filler is uniformly distributed in the cured section. Filler settling in the base component during storage - which is significant in systems with particle densities substantially above the resin carrier - produces a cured section with variable filler distribution, and therefore variable thermal conductivity. Thermal conductivity measured at one location in the cured part may not represent the bulk average, and it will not represent sections where the filler-depleted upper material was poured. This is not a material defect - it is a handling defect. Pre-mixing the base component in its original container before weighing is not optional in high-filler systems.
Figure 3. Filler settling in the E533 base component is significant enough during storage to produce measurable non-uniformity in cured thermal conductivity if the container is not mechanically re-mixed before weighing.
The Voids Problem: Why Degassing Is More Critical in Thermally Conductive Systems
In a standard 0.5 W/m·K epoxy potting compound, entrapped voids reduce local dielectric strength and create stress concentration sites. In a thermally conductive compound designed to conduct heat, voids have an additional and more severe consequence: they are thermal insulators embedded in a thermally conductive matrix.
The thermal conductivity of air at ambient conditions is approximately 0.026 W/m·K - roughly 1/58th of the surrounding 1.5 W/m·K matrix. A spherical void in a thermally conductive matrix creates a local thermal resistance that is orders of magnitude higher than the surrounding material. In a thick-section power module where the design intent is to conduct heat through the potting to the enclosure wall, a cluster of voids at a critical location can create a local thermal bottleneck that defeats the purpose of specifying the higher-conductivity compound.
Vacuum degassing is therefore more consequential in thermally conductive systems than in standard systems. The argument for degassing a standard system is primarily dielectric - voids reduce effective dielectric strength. The argument for degassing a thermally conductive system is both dielectric and thermal. Whether a given application requires degassing depends on cavity geometry and void content achievable through careful dispensing, but in high-power-density potted modules the safe assumption is that degassing is required unless cavity fill quality has been validated on representative specimens.
Glass Transition Temperature and Its Relationship to Thermal Performance
A thermally conductive potting compound is being used in a hot environment by definition - that is the application condition that motivated the selection. The glass transition temperature (Tg) of the cured system determines at what temperature the mechanical form of the potting begins to change. Below Tg, the compound is glassy, rigid, and dimensionally stable. Above Tg, the polymer network transitions to a rubbery state with significantly reduced modulus and rapidly increasing CTE.
For a potted power assembly running at elevated temperature, the Tg of the compound establishes the upper bound of reliable dimensional stability - not the continuous service temperature maximum, which requires a thermal margin below Tg. If the potting section core temperature approaches or exceeds Tg during normal operation, the compound will creep under the load of its own thermal expansion, potentially cracking the interface with embedded components or the enclosure.
This means the Tg requirement for a thermally conductive compound is determined by the thermal model output - specifically by the predicted core temperature of the potted section at maximum continuous load - not by the ambient temperature of the enclosure. In a dense power module where the potting layer reduces junction temperature but the core of the potted mass still reaches 110°C, a compound with Tg of 127°C (with an operating margin of ~17°C) is meaningful. A compound with Tg of 70°C would begin to lose dimensional stability under those conditions.
What a Proper Thermal Model Should Include for Potted Assemblies
A thermal model for a potted power assembly that excludes the potting compound thermal resistance will systematically underpredict junction temperature. The correct approach includes:
The junction-to-case thermal resistance of each dissipating component (from component datasheet).
The contact resistance between the component package and the surrounding potting compound (depends on wetting and void content at the interface).
The bulk thermal resistance of the potting layer from the component surface to the first cooling boundary (enclosure wall, heatsink, or PCB copper plane).
The contact or interface resistance between the potting and the cooling boundary.
The thermal resistance of the cooling boundary itself (enclosure wall thickness and material, heatsink efficiency).
In assemblies where the potting layer thermal resistance is the dominant term - identified by the fact that removing it from the model produces a junction temperature substantially below the measured value - the selection of potting compound thermal conductivity directly affects the thermal design. This is the condition where specifying 1.5 W/m·K versus 0.5 W/m·K produces a meaningful difference in system reliability.
When Thermally Conductive Potting Does Not Resolve the Problem
Specifying a 1.5 W/m·K potting compound will not solve a junction overtemperature problem when:
The component junction-to-case resistance is the dominant term. If the component itself is the thermal bottleneck, improving the potting compound's conductivity has marginal effect. The full thermal model must be analyzed to identify which resistance is dominant before changing materials.
The potting section is thin (below 5 mm). At low thickness, the absolute thermal resistance of the potting layer is small regardless of conductivity. Specifying 1.5 W/m·K to address a 5 mm potting layer adds process complexity without meaningful thermal benefit.
The cooling path between the potting outer surface and ambient is the limiting resistance. If natural convection from the enclosure surface is the thermal bottleneck, reducing the potting layer resistance moves the bottleneck one step outward - it does not reduce junction temperature proportionally.
The voids and filler distribution are not controlled. A thermally conductive compound with 10–15% void content may perform no better than a standard compound with zero voids, because the voids create local thermal resistances that exceed the bulk conductivity improvement.
Related Product for Thermal Management in Thick-Section Potting
E533/H533 is a heavily filled, two-component epoxy potting compound delivering 1.5 W/m·K thermal conductivity and Tg 127°C. It requires a two-stage heat cure (80°C × 2 hrs + 120°C × 4 hrs) to develop its rated properties. The base component (E533) has a viscosity of 500,000–1,500,000 cps at 25°C - mechanical pre-mixing and heated dispensing at 50°C (where mixed viscosity drops to 700–1,500 cps) are required for consistent property development and void-free fill.
UL 94 V-0 certification status under File E120665 (listed as E-53(Y)/H-53(Y)) should be confirmed with Fong Yong Chemical before specification, as follow-up testing status as of December 2025 requires verification. Engineers who require currently active UL certification should confirm reinstatement timeline before including E533/H533 in a UL-listed end product.
👉 🔗 E533/H533 Product Page - Technical Data, Thermal Conductivity, Application Notes
Key Engineering Questions
At what potting thickness does the thermal conductivity specification start to matter?
As a rough guideline, the potting layer thermal resistance becomes significant relative to other thermal resistances in the path when the potted section exceeds approximately 8–10 mm and power dissipation density exceeds 1 W/cm². Below these thresholds, the absolute resistance of the potting layer is typically not the dominant term, and increasing thermal conductivity from 0.5 to 1.5 W/m·K produces less than 5°C improvement in junction temperature. This should be confirmed by running the numbers in a full thermal model for the specific geometry before making a material change decision.
Can thermal conductivity be measured on production specimens to verify the compound is performing as specified?
Yes, but the measurement should be performed on cured specimens made at production batch size and degassing conditions, not on laboratory specimens prepared under ideal conditions. Thermal conductivity in highly filled systems is sensitive to void content and filler distribution. A production specimen with 5% void content and incomplete filler re-dispersion from inadequate pre-mixing may measure 0.8–1.0 W/m·K rather than 1.5 W/m·K. Periodic thermal conductivity measurement on production-representative specimens is the correct verification approach, not reliance on TDS values alone.
Does the Tg of the potting compound affect its thermal conductivity during operation?
Thermal conductivity in highly filled systems is less sensitive to Tg transition than mechanical properties. The primary concern above Tg is dimensional stability and creep - the compound softens, CTE increases by approximately 2–3×, and sustained load causes creep at the potting-component interface. Thermal conductivity does not drop dramatically at Tg for a heavily filled system because the filler particles (which carry most of the heat) remain in place. The Tg concern in a thermally loaded application is mechanical, not thermal conductivity-related.
Next Steps - Contact Fong Yong Chemical
