Figure 1. In wire-bonded assemblies, rigid epoxy potting applies cure shrinkage stress laterally across bond wire loops. The stress concentrates at the bond heel - the weakest cross-section in the wire - and the failure appears after multiple thermal cycles, not at initial test.
The specification called for a flame-retardant epoxy potting compound. A rigid, high-modulus system was selected - well-characterized, UL-listed, documented Tg and dielectric strength. The engineering team was confident in the material choice. Six months into production, wire bond failures begin appearing in returned units. Not all returned units - roughly 3% of shipments from a specific date range. Cross-section analysis shows bond wire fractures at the heel, with no evidence of over-current or mechanical shock. The metallurgy of the wire is normal. The die attach is intact. The bond pull strength on incoming material was within specification.
What the investigation does not find - because it is not in the failure analysis checklist - is that the fractures occurred at the heel of the bond loop because the rigid epoxy cured and contracted around the wire, pulling the loop laterally as it shrank, concentrating stress precisely at the heel where the wire cross-section transitions from the FAB bond to the wire body. The material was not the wrong manufacturer. It was the wrong modulus.
Most solder joint and wire bond fatigue failures in potted assemblies are generated by the encapsulant, not by the joint. The encapsulant applies the stress. Changing the joint geometry, alloy, or wire diameter does not address a stress source that is external to the joint.
What Rigid Epoxy Does During Cure
When a two-component epoxy system is mixed and dispensed into a cavity containing electronic components, the cross-linking reaction that produces the cured solid also produces volumetric shrinkage. For most rigid epoxy potting systems, linear shrinkage is in the range of 0.2–1.0%. In absolute terms, 0.5% linear shrinkage across a 30 mm cured section means 150 μm of contraction. This contraction is constrained by the embedded components and the cavity walls, which do not move. The resin cannot freely contract - it is held by its adhesion to all surfaces it wets. The result is a stress field distributed throughout the cured volume, with peak stress at the stiffest features: component leads, wire bonds, device corners, and the potting-to-housing interface.
In a rigid, high-modulus system (Shore D 75–95), the cured polymer cannot deform significantly to relieve this stress. The stress field that was set during cure remains in the cured part as a locked-in load. Every embedded component and every interface in contact with the potting is under a sustained, static stress from the cured encapsulant - before any service load, before any thermal cycling, before any vibration.
The magnitude of this stress depends on the shrinkage magnitude, the modulus of the cured epoxy, the modulus of the substrate and components, and the geometry. For typical rigid potting systems on circuit boards with through-hole and SMT components, cure shrinkage stress at solder joint interfaces can reach 5–15 MPa - well below the ultimate tensile strength of the joint, but sufficient to reduce its fatigue life when combined with service loads.
The Thermal Cycling Amplification
Cure shrinkage stress is a static load. Thermal cycling is a dynamic load. In service, every temperature excursion from the cure temperature generates additional stress at every interface where the CTE of the epoxy differs from the CTE of the adjacent material. The stress amplitude per cycle depends on the CTE mismatch, the temperature excursion magnitude, and the stiffness of the materials.
For a rigid epoxy (CTE ~50–70 ppm/°C below Tg) bonded to an FR-4 PCB (CTE ~14–18 ppm/°C in-plane, ~60–80 ppm/°C out-of-plane), a copper lead frame (CTE ~17 ppm/°C), a ceramic capacitor body (CTE ~7–10 ppm/°C), and an aluminum enclosure (CTE ~23 ppm/°C), the CTE mismatch at each interface generates shear stress during every temperature change. In a rigid encapsulant, this shear stress cannot be relieved by deformation of the encapsulant - it is transmitted to the weakest interface in the load path.
The weakest interface depends on the assembly geometry. In wire-bonded modules, it is typically the bond heel or the second bond (wedge bond). In fine-pitch SMT assemblies, it is the solder joint at the corner-most component positions, where the eccentricity from the neutral point is highest. In coil or transformer assemblies with mixed metal materials, it is the epoxy-to-housing interface where the CTE mismatch between the fill, the wire, and the enclosure produces the highest shear.
The combined effect of cure shrinkage static stress plus cyclic thermal stress determines the fatigue life of the joint. The cure shrinkage stress term raises the mean stress level. The thermal cycling term provides the cyclic amplitude. Both contribute to crack initiation; the rate of crack propagation depends on both terms.
Why the Failure Timeline Causes Misidentification
Stress-transfer failures driven by rigid epoxy potting do not appear immediately after cure. The crack initiation cycle count depends on the combined stress amplitude, which is a function of geometry and materials. In typical assemblies, failures appear after 100–500 thermal cycles in service, or after several months to a year of continuous vibration exposure. This timeline causes consistent misidentification:
At initial test - the assembly passes all electrical checks, hi-pot, and visual inspection. The cure shrinkage stress is present but below the crack initiation threshold. No failure is detected.
At early field use - the assembly functions normally. The accumulated thermal cycles have not reached the crack initiation threshold. No failure is detected.
At 3–12 months in service - failures begin appearing. The investigation focuses on the failed component or joint, not on the encapsulant. Wire bond pull strength on returned units may meet incoming specification because the wires that did not fail are intact - the statistical population of failed wires is already in the failed units.
During failure analysis - cross-sections show cracking at the bond heel or solder joint interface. The investigation attributes this to metallurgical fatigue, which is technically accurate - fatigue crack propagation was the final failure mode - but omits the root cause: elevated stress amplitude from the rigid encapsulant.
The correct root cause identification requires comparing the failure rate and crack location pattern with what would be expected from the calculated stress field in the potted geometry. Cracks that initiate at predictable high-stress locations (bond heels in wire-bonded modules, corner components in SMT arrays, lead exits in potted coils) distributed uniformly across the population - rather than randomly at random locations - are consistent with a systematic stress source in the encapsulant.
What a Low-Modulus Encapsulant Does Differently
A semi-flexible epoxy with Shore A 80–90 and elongation of approximately 140% responds to cure shrinkage and thermal cycling stress by deforming, rather than transferring stress to embedded components. The modulus of a Shore A 80 material is approximately two orders of magnitude lower than Shore D 80 - the same way that a rubber band and a steel rod respond differently to the same applied force. The rubber band deforms. The steel rod transmits the force.
When a low-modulus encapsulant cures and shrinks, it cannot generate high stress at embedded interfaces because its stiffness is insufficient to sustain a large stress field. The shrinkage occurs, but the resin deforms to accommodate it rather than transmitting the contraction load to the adjacent components. The residual stress state in the cured part is substantially lower than in a rigid system with the same shrinkage percentage.
During thermal cycling, the low-modulus system deforms to accommodate the differential CTE movement between the epoxy and the embedded materials. The shear stress at the interface is reduced because the encapsulant moves with the substrate rather than resisting it. The CTE mismatch still exists - the materials have not changed - but the stress that results from the mismatch is absorbed by the encapsulant's deformation rather than being transferred to the joint.
This is the engineering basis for specifying a semi-flexible system. It is not that the semi-flexible system makes the assembly stronger. It is that the semi-flexible system removes the encapsulant as a stress source, allowing the assembly to operate under its designed load conditions without the additional imposed load from the potting compound.
Figure 2. A rigid epoxy cannot deform to accommodate cure shrinkage - the stress is transmitted to the weakest interface in the load path. A semi-flexible system with ~140% elongation deforms instead, removing the encapsulant as a stress source without changing the joint geometry.
The Trade-Offs of Low Modulus: What Semi-Flexible Cannot Do
The properties that make a semi-flexible system effective for stress relief are the same properties that make it unsuitable for applications requiring mechanical rigidity, structural support, or aggressive thermal performance:
Dimensional stability under sustained mechanical load. Shore A 80–90 will creep under sustained compressive or shear load. If the potted assembly is mechanically constrained by a press-fit pin, a hold-down bracket that exerts sustained force, or a connector that transmits insertion force to the potted area, the semi-flexible matrix will deform over time. A rigid epoxy is required for load-bearing applications.
Thermal conductivity. Semi-flexible systems have thermal conductivity in the same range as standard rigid potting compounds - typically 0.5–0.7 W/m·K. If the design requires the potting layer to conduct heat from a power-dissipating component to a cooling surface, a semi-flexible system at this conductivity level will not provide meaningful thermal improvement. A thermally conductive rigid system (1.0–1.5 W/m·K) is needed.
Thick-section behavior. The elongation property that makes a semi-flexible system useful for stress relief is accompanied by higher exothermic heat generation per unit volume at the center of a thick pour, because the higher catalyst level needed for room-temperature cure produces a faster reaction. Large volume pours in deep sections may generate enough exothermic heat to cause local overtemperature. Section thickness and pour volume should be validated before production.
Creep at the upper service temperature. A Shore A 80–90 system operating near its upper service temperature limit (100°C for typical semi-flexible systems) will exhibit higher creep rates than a rigid system at the same temperature. Applications where dimensional precision under thermal load is required should use a rigid, high-Tg system.
Application Conditions Where Encapsulant Modulus Is the Governing Selection Criterion
The following assembly conditions indicate that the stress-transfer mechanism is the governing failure risk, and that encapsulant modulus - rather than dielectric strength, thermal conductivity, or Tg - should drive material selection:
Wire-bonded modules (gold or copper wire, ball or wedge bonds) enclosed in a rigid potting compound, operating under thermal cycling or vibration.
Fine-pitch SMT assemblies (0.5 mm pitch or finer) with multiple component types having different CTEs - ceramic passives, polymer packages, and metal-body inductors in the same potted area.
PCBs with thin, unsupported sections or flexible substrates enclosed in rigid potting - the rigidity differential between the substrate and the potting generates high interfacial stress during cure.
Ferrite core assemblies (transformers, inductors, common-mode chokes) where the ferrite body CTE (~10 ppm/°C) is substantially different from the surrounding epoxy CTE (~50–70 ppm/°C).
Assemblies in continuous vibration environments (automotive, industrial motor drives, outdoor fixtures) where the cumulative cyclic load is the dominant failure driver.
Any assembly where prior failure history shows cracking, intermittent opens, or delamination that correlates with thermal cycle count rather than with a specific overstress event.
Modulus Selection as a Design Decision, Not a Default
The standard selection process for epoxy potting compounds in most B2B procurement workflows begins with flame rating (UL 94 V-0), moves to dielectric strength, and then evaluates cure schedule and Tg. Modulus and elongation are often listed last in the TDS and are rarely weighted heavily in the initial selection. This ordering reflects the sequence of compliance requirements - flame rating is legally mandated, dielectric strength is measurable, modulus is not in most equipment standards.
The consequence is that assemblies with mechanically sensitive structures are routinely potted with rigid, high-modulus compounds because there was no selection gate that asked the modulus question. The specification passes compliance review. The failure appears in the field. The investigation does not return to the selection process.
The correct approach is to add mechanical stress analysis to the early design stage - before potting compound selection is made. The question "what stresses does this encapsulant apply to the assembly during cure and service?" must be answered before specifying a material, not after the first field returns.
This requires knowing the approximate shrinkage of the candidate compound, the modulus of the cured system, the CTE of the substrate and components, and the geometry of the potted section. None of these require finite element analysis - a first-order estimate from material properties and geometry is sufficient to determine whether stress transfer is likely to be a governing failure mechanism before material selection is finalized.
Related Product for Stress-Sensitive Assembly Potting
E759/H759 is a two-component, semi-flexible epoxy potting compound with Shore A 80–90 and approximately 140% elongation at break. It is UL 94 V-0 certified under UL File E120665 at minimum thickness 1.58–1.74 mm. Service temperature range is –30°C to +100°C. Mix ratio is 100:30 by weight; pot life is approximately 60 minutes for a 60 g mass at 25°C. Cure is by room temperature (7 days at 25°C) or heat-accelerated (50–60°C × 2 hrs + 80°C × 2 hrs).
It is appropriate when the dominant risk is mechanical stress transfer - wire bond fatigue, solder joint cracking, CTE-mismatch delamination, or vibration-induced fracture. It is not appropriate for load-bearing structural potting, high-heat-flux thermal management, or assemblies requiring Shore D rigidity for dimensional tolerance. Selection should be validated on representative specimens under the actual thermal cycle profile of the application.
→ 🔗E759/H759 Product Page - Technical Data, UL Certification, Application Notes
Key Engineering Questions
How do I estimate whether stress transfer is occurring in my current assembly?
A first-order estimate can be made from the shrinkage of the potting compound (from the TDS, typically listed as % linear shrinkage), the modulus of the cured system (correlated with Shore D - Shore D 80 corresponds to roughly 1,500–2,500 MPa tensile modulus), and the geometry of the potted section. Stress at a rigid embedded interface is approximately E × ε, where E is the epoxy modulus and ε is the constrained shrinkage strain. If the resulting value is a significant fraction of the solder joint or wire bond fatigue limit, stress transfer is likely occurring. This is a rough estimate - geometry and load path details affect the actual stress significantly - but it identifies whether the mechanism warrants detailed analysis or experimental validation before finalizing material selection.
If the assembly currently uses a rigid epoxy and has a field failure history consistent with stress transfer, what is the correct evaluation sequence for a semi-flexible alternative?
Start by confirming the failure mechanism through cross-section analysis of returned units - crack initiation location, crack propagation path, and correlation with thermal cycle count. Then produce test specimens of the actual assembly with the semi-flexible candidate at the same geometry and cure schedule, and run accelerated thermal cycling to a cycle count that covers the same failure range observed in the field (typically 2–5× the cycle count where field failures first appeared). Compare failure rate and crack initiation location between the rigid and semi-flexible specimens. This process takes 4–8 weeks depending on thermal cycling equipment availability, but it is the only reliable basis for a material change decision. Datasheet comparison alone does not predict in-service behavior for this failure mechanism.
Does a lower-modulus system provide less environmental protection than a rigid one?
A semi-flexible system at Shore A 80–90 retains the environmental protection function - it seals the assembly against moisture ingress, provides electrical insulation, and meets UL 94 V-0 flame performance. What it does not provide is mechanical rigidity - it will deform under sustained compressive load. For environmental protection in non-load-bearing applications, Shore A 80–90 is adequate. The comparison that matters is whether the reduction in modulus from Shore D to Shore A is relevant for the specific mechanical loading the assembly will see in service, not whether the semi-flexible system provides "less protection" in an abstract sense.
Next Steps - Contact Fong Yong Chemical
