Nonlinear Viscoelastic Modulus Relaxation Modeling in Complex Multi-Compound Sensor Potting Architectures
Dual-compound sensor potting relaxes over time, shifting baseline bridge calibration unless modeled via nonlinear viscoelasticity and stabilized by post-cure baking.

Resin
A dual-stage fill using silicone gel under a rigid aromatic polyurethane shell puts an initial hydrostatic pressure of 4.2 megapascals on a 3 by 3 millimeter MEMS pressure sensor die at negative 40 degrees Celsius. Over 96 hours at 85 degrees Celsius, this contact stress falls to 1.1 megapascals, shifting transducer output by 14 millivolts across a 3.3-volt supply rail. The piezoresistive bridge reads this stress decay directly as baseline drift.
As potting compounds cool after their reaction exotherm, residual stresses set into the boundary between the soft inner gel and the hard cap. While the gel buffers wire bonds and the silicon diaphragm against physical impact, the outer shell protects the assembly against automotive fuels, salt spray, and high-pressure wash-down. Combining both materials leaves a dynamic interface where stress relaxes continuously over months in the field.
Mismatches in thermal contraction drive most early mechanical behavior in the package. Unfilled silicone gels have a coefficient of thermal expansion between 180 and 320 parts per million per Kelvin, compared to 14 to 17 parts per million per Kelvin for standard FR4 substrates and 2.6 parts per million per Kelvin for the micromachined silicon die. The outer polyurethane or epoxy shell contracts at 45 to 80 parts per million per Kelvin.
Cooling down from a 120-degree post-cure plateau to ambient room temperature forces the inner elastomer’s volumetric contraction against the rigid package walls and cap. That constraint creates triaxial tensile stress inside the soft core along with heavy shear loads along the housing sidewalls. Glass transition temperatures dictate how these materials respond: the inner silicone sits well above its glass transition temperature of negative 125 degrees Celsius, staying compliant, whereas the outer polyurethane crosses its glass transition between 45 and 65 degrees Celsius during standard thermal cycling.
A coefficient of thermal expansion mismatch of 200 parts per million per Kelvin across a four-millimeter fill cavity generates 1.8 megapascals of localized shear along the package perimeter during cold soak.
Under sustained strain, polymer chains rearrange to lower their internal energy. In dual-compound packages, this relaxation happens along several coupled paths at once. Secondary inter-chain bonds slip under load, while physical entanglements loosen over logarithmic time scales.
Although crosslink density sets the ultimate terminal modulus, short-term behavior depends on these transient entanglement networks. In two-part addition-cure silicones, unreacted oligomers and cyclic siloxanes account for two to six percent of the cured mass, functioning as internal plasticizers that speed up stress decay under heat. Meanwhile, in the polyurethane cap, hard segments from diisocyanate reactions isolate into glassy microdomains inside a soft polyol matrix.
As the outer shell’s modulus relaxes, the mechanical confinement on the inner gel shifts, driving internal forces back onto the sensor diaphragm.
- Interfacial delamination propagates along the boundary between the inner gel meniscus and the outer shell when shear stresses exceed the 0.4 megapascal adhesive threshold, converting triaxial tension into uncontrolled biaxial expansion.
- Piezoresistive bridge imbalance develops as asymmetric volume relaxation across the die surface warps the silicon substrate, introducing uncorrectable span errors into analog signal chains.
- Wire bond fatigue occurs when cyclic viscoelastic relaxation drives micro-motion across 25-micrometer gold bond wires, work-hardening the neck adjacent to the second bond stitch until fracture occurs.
- Capacitive gap modulation alters zero-pressure baseline capacitance in absolute pressure cells by flexing the package lid, which shifts the internal cavity volume against the relaxed potting boundary.
Tight control over meniscus height is critical for stabilizing the boundary layer. Small dispensing variations alter the ratio between the soft buffer layer and the rigid shell: a ten percent increase in silicone fill height lifts the package’s neutral mechanical axis upward by 140 micrometers, doubling the bending moment sent into the substrate during temperature swings. Dispensing needle tolerances govern volumetric consistency; a positive or negative five percent variance on a 12-microliter dispense leads to noticeable lot-to-lot spread in zero-offset drift.
If left uncalibrated, this ongoing stress relaxation eventually pushes sensor offsets past the limits defined by downstream control units, driving warranty claims.

Prony
Modelling time-dependent shear behavior usually relies on a generalized Maxwell model built from a series of decaying exponentials. Here, the relaxation shear modulus combines an equilibrium modulus with a sum of Maxwell elements, each assigned its own elastic stiffness and characteristic relaxation time. Bulk modulus relaxation uses a similar summation, though in filled encapsulation compounds it relaxes much slower than shear modulus, keeping the material nearly incompressible during fast mechanical transients.
Poisson’s ratio for the inner silicone holds between 0.495 and 0.499, whereas the outer polyurethane drops from 0.48 above its glass transition down to 0.38 once glassy. Capturing stress on a potted die over eight decades of physical time requires fitting eight to twelve Maxwell branches.
Temperature shifts the full relaxation spectrum horizontally on a logarithmic time scale. Using time-temperature superposition, dynamic mechanical thermal analysis data taken over discrete temperature steps can be joined into a single master curve. Above the glass transition, the Williams-Landel-Ferry equation sets the shift factor using two empirical constants and a chosen reference temperature.
Below glass transition, chain mobility departs from Williams-Landel-Ferry behavior and transitions to Arrhenius kinetics, with activation energies between 80 and 220 kilojoules per mole based on filler loading and crosslinking. As thermal cycling crosses the outer material’s glass transition, solver routines must handle the transition between these kinetic models smoothly to prevent artificial numerical spikes in finite element runs.
Master curves lose mathematical coherence when filler-matrix debonding occurs during large mechanical excursions.

What Drives Time Temperature Superposition Breakdown?
Filled encapsulation resins deviate from linear viscoelastic behavior once mechanical strain exceeds 0.5 percent. Particles of fumed silica, aluminum trihydrate, or carbon black form secondary networks within the polymer matrix that break apart under dynamic loading, causing the sharp drop in storage modulus known as the Payne effect. When multi-material packages experience fast temperature changes, differential expansion creates localized shear strains of 1.5 to 4.0 percent around package corners.
In this regime, the relaxation modulus becomes dependent on strain amplitude alongside time and temperature. Applying standard linear Maxwell models here underpredicts stress relaxation by 30 to 50 percent, skewing estimates of the net clamping load on the silicon transducer.
| Material Designation | Compound Chemistry | Initial Modulus (1s) | Relaxed Modulus (1000s) | Terminal Modulus (100000s) | Glass Transition |
|---|---|---|---|---|---|
| Gel-A | Two-part polydimethylsiloxane | 420 kPa | 180 kPa | 65 kPa | -120 °C |
| Gel-B | Fluoro-silicone dielectric gel | 850 kPa | 310 kPa | 110 kPa | -65 °C |
| Shell-P | Aromatic polyether polyurethane | 85 MPa | 32 MPa | 12 MPa | 48 °C |
| Shell-E | Filled cycloaliphatic epoxy | 3.8 GPa | 3.2 GPa | 2.9 GPa | 135 °C |
Running non-linear models in commercial finite element software usually requires custom user material subroutines. The Schapery formulation addresses non-linear viscoelasticity by applying stress-dependent scaling functions to transient compliance and reduced time. Alternatively, finite-strain formulations use a multiplicative split of the deformation gradient to separate elastic and viscous contributions.
Under high hydrostatic confinement, volumetric response couples into deviatoric behavior because compressive stress restricts chain movement, pushing relaxation times outward. A sensor die buried under two millimeters of rigid encapsulation can see hydrostatic pressures above five megapascals during thermal shock, noticeably shifting the material’s internal relaxation rate.
How fast physical aging shifts the relaxation spectrum toward longer timescales during long room-temperature storage remains a key uncertainty in predicting multi-year calibration stability.

Swell
Atmospheric moisture enters exposed potting surfaces through both Fickian and non-Fickian diffusion. As water is absorbed, the polymer expands according to its coefficient of hygroscopic swelling ~ typically 0.15 to 0.35 percent strain per percent moisture weight gain in electronic-grade polyurethanes. A package exposed to 85 degrees Celsius and 85 percent relative humidity takes up 1.2 to 2.8 percent water by weight over 500 hours.
This volume growth opposes thermal contraction while shifting internal stress patterns. Inside the polymer, absorbed water breaks hydrogen bonds along urethane chains, acting as a plasticizer that lowers the glass transition temperature by 15 to 30 degrees Celsius.
This plasticization can speed up viscoelastic relaxation rates by up to two orders of magnitude. A polyurethane that takes 10,000 seconds to reach its relaxation half-life when dry will relax in under 200 seconds once moisture-saturated, altering how and when load reaches the internal die. During rapid drying, water leaves the surface much faster than it diffuses out from the center, producing steep concentration gradients.
The outer surface shrinks while the core remains swollen, generating tensile surface stresses that encourage cracking near the fill port. In silicone buffer gels, moisture moves very quickly ~ reaching saturation in hours ~ even though total equilibrium water solubility stays below 0.1 weight percent.
Moisture absorption depressing the glass transition temperature by twenty degrees Celsius accelerates relaxation rates tenfold under ambient conditions.
Hygroscopic swelling works directly against physical aging. Where aging gradually packs polymer chains closer together ~ reducing free volume and increasing stiffness ~ water absorption forces the matrix open. In automotive applications under mixed environmental conditions, these two processes compete continuously.
Modeling stress without hygro-mechanical coupling overpredicts internal residual stress by up to 40 percent after moisture preconditioning. That error often leads design teams to over-engineer package pedestals or specify excessively soft gels that erode under direct fluid exposure.
- Incoming compound verification requires Karl Fischer titration and dynamic vapor sorption analysis to establish baseline resin dryness before production line degassing.
- Degassing chamber cycles must maintain vacuum levels below two millibars for 45 minutes to pull dissolved moisture and trapped air out of high-viscosity polyols.
- Dispense cavity baking clears adsorbed surface moisture from FR4 substrates and premolded housing plastics at 105 degrees Celsius prior to primer application.
- Cure environment regulation restricts assembly area relative humidity to less than 35 percent to prevent premature isocyanate foaming and micro-void generation.
Moisture-induced swelling can offset overall thermal shrinkage across the package, yet differential drying still creates severe localized shear spikes along internal adhesive interfaces.

Offset
Decaying stress in the encapsulation material shows up directly as signal error at the transducer. Piezoresistive dies rely on Wheatstone bridges etched into the silicon surface along specific crystallographic axes to detect strain. Any residual compressive stress shifts the zero balance of the bridge.
As the surrounding resin relaxes over time, this compressive clamp decays, causing zero-pressure signal drift. On a 100-kilopascal full-scale absolute sensor, a five-kilopascal drop in potting stress causes a 50-millivolt drift on an unamplified bridge. The thermal hysteresis observed during temperature qualification is usually mechanical stress relaxation in action, rather than an electrical fault.
Digital interfaces handle part of this shift using register compensation and baseline tracking routines. An ASIC placed alongside the sensing element uses I2C or SPI to update calibration coefficients stored in non-volatile memory. Standard automotive sensors output readings over Single Edge Nibble Transmission or Local Interconnect Network protocols using polynomial correction tables to compensate for temperature.
However, those matrices assume the relationship between temperature and package stress remains fixed. As viscoelastic relaxation alters that relationship over time, factory calibration curves drift out of alignment, introducing errors during thermal swings.
A three-megapascal relaxation in package stress shifts uncompensated bridge zero outputs by more than four percent of full operating scale.

Is Solder Fatigue Decoupled from Fill Relaxation?
Solder joints under surface-mount sensors interact directly with the surrounding potting material. Lead-free SAC305 solder creeps under thermal cycling through grain boundary sliding and dislocation movement. The polyurethane potting compound relaxes at its own rate.
At 125 degrees Celsius, the polyurethane softens quickly, shifting mechanical load onto the solder joints. When temperatures drop to negative 40 degrees Celsius, the potting compound passes below its glass transition into a rigid state that grips the component. That rigid clamping locks the solder joints under heavy tensile strain, accelerating fatigue cracking by up to 250 percent compared to unpotted boards.
| Architecture Description | Inner Fill Thickness | Outer Cap Material | Full-Scale Zero Drift | Solder Crack Incidence |
|---|---|---|---|---|
| Single rigid epoxy overmold | None (Direct fill) | Filled Epoxy | +6.2% | 14% of joints |
| Silicone gel buffer with epoxy cap | 500 µm Silicone | Filled Epoxy | +1.8% | 2% of joints |
| Silicone gel with polyurethane cap | 800 µm Silicone | Polyurethane | +0.9% | 0% of joints |
| Dual-gel floating core | 1200 µm Soft Gel | Tough Silicone | +0.4% | 0% of joints |
Firmware baseline tracking helps mitigate slow viscoelastic drift in the field. For barometric and ambient pressure sensors, host controllers sample data during known idle periods and gradually adjust offset registers over I2C to offset mechanical relaxation. This digital correction works as long as stress decay is much slower than real changes in atmospheric pressure.
In hydraulic or brake line monitoring, continuous static pressure prevents routine zero-point recalibration. Those applications demand intrinsic package stability, requiring engineers to replace rigid caps with compliant floating-core designs or isolated stainless-steel diaphragm assemblies.
While compliant inner buffers protect electrical calibration, thin rigid outer caps tend to drive long-term baseline drift.

Bake
Post-cure heat treatment speeds up polymer chain stabilization before sensors enter factory calibration. Holding modules at 100 degrees Celsius for 24 hours drives off the steep initial stress relaxation, moving the material into its slower, logarithmic decay regime. In typical two-part polyurethanes, this bake increases crosslink density by raising isocyanate conversion from 92 percent to over 98.5 percent.
Higher conversion pushes the glass transition temperature safely above expected operating ranges. Stabilizing the resin thermally compresses what would otherwise be weeks of room-temperature settling into a predictable process, keeping newly calibrated units from drifting on warehouse shelves.
Post-cure schedules are tied directly to oven capacity and throughput. Large batch convection ovens process thousands of modules at once on aluminum trays. Ramp rates are kept under two degrees Celsius per minute to prevent thermal shock and differential expansion between the shell and substrate.
Dwell times at peak temperature run from 4 to 48 hours based on component mass and resin chemistry. Cooling must stay below one degree Celsius per minute to avoid locking thermal stresses into the matrix; pulling parts from the oven too quickly quenches the outer skin under high tension, causing early adhesive failure during thermal shock testing.
Manufacturing costs depend heavily on whether sensors are potted on-line or purchased as pre-sealed modules. Bare surface-mount pressure sensors cost between $1.80 and $3.20 per unit at annual volumes of 50,000 pieces. Setting up an automated in-house potting cell requires over $250,000 in capital equipment for meter-mix dispense units, vacuum chambers, and curing ovens.
Maintenance, purging, scrap, and floor space add about $0.85 per part. By comparison, purchasing fully encapsulated, pre-calibrated standalone modules with built-in connectors and automotive seals costs $7.50 to $14.00 per unit. The financial break-even point where in-house potting and post-cure processing becomes cheaper than buying pre-packaged modules is around 85,000 units per year.
Overall yield relies on maintaining strict cleanliness and process controls during dispense and cure. Volatiles outgassing from under-cured FR4 solder masks poison platinum-catalyzed silicones, leaving an uncrosslinked, tacky layer at the die interface. That uncured layer flows unpredictably over time, degrading signal baseline stability.
Quality teams screen board lots using contact angle measurements and thermal desorption gas chromatography before releasing substrates to dispensing lines. In-line audits cover dispense head calibration, mixing ratios checked via Shore hardness test coupons, and continuous datalogger tracking of oven profiles. When dispensing accuracy and curing kinetics are tightly managed, the final assemblies maintain stable output over years of demanding automotive and industrial service.


