Thermal stability testing of underfill adhesives is essential for verifying the long-term reliability of flip-chip, BGA, and wafer-level packages exposed to sustained high temperatures and cyclic thermal stress. This article explains the experimental workflow for underfill thermal stability evaluation, covering sample preparation, key testing items such as glass transition temperature, coefficient of thermal expansion, decomposition temperature, and thermal aging behavior, along with the relevant ASTM, ISO, IPC, JEDEC, and UL-referenced test methods. Engineers and quality managers in semiconductor packaging will find practical guidance on acceptance criteria, instrumentation, and result interpretation.
Technical Overview
Underfill adhesives are epoxy-based or hybrid polymer materials dispensed between a silicon die and its substrate to redistribute thermomechanical stress generated by the coefficient of thermal expansion (CTE) mismatch between silicon (~2.6 ppm/°C), organic substrates (~15–17 ppm/°C), and copper interconnects. During solder joint formation and subsequent product operation, the underfill is routinely exposed to temperatures ranging from –55°C to 175°C or higher. If the material degrades, softens, or loses adhesion at these temperatures, solder joint fatigue, delamination, and premature device failure can result.
Thermal stability testing is therefore a core qualification activity for underfill materials. The experiment characterizes how the cured adhesive responds to heat through three complementary perspectives: thermophysical transitions (glass transition, softening), thermomechanical behavior (dimensional change, CTE drift), and thermochemical decomposition (weight loss, outgassing, chain scission). Combined with long-term thermal aging and temperature cycling, these measurements establish whether a material will survive the intended reflow profile, burn-in conditions, and operational lifetime of the electronic assembly.
Testing is typically performed by material suppliers, semiconductor packaging houses, or accredited third-party laboratories using standardized procedures defined by ASTM, ISO, IPC, JEDEC, and IEC. A well-designed test plan links material-level data to package-level reliability, enabling data-driven material selection and process window definition.
Testing Samples
Cured Bulk Material Specimens
Bulk underfill specimens are cast into plaques or coupons and cured according to the manufacturer's recommended schedule (for example, 150°C for 30 minutes to 3 hours, or snap-cure at 165°C). From the cured plaque, specimens are machined or die-cut into geometry suitable for each instrument: disks or squares of 3–5 mm for thermogravimetric analysis (TGA), 4–6 mm samples for differential scanning calorimetry (DSC), and 4–10 mm cubes or strips for thermomechanical analysis (TMA). Curing completeness is verified in advance by DSC residual exotherm measurement, since incompletely cured material will show artificially low thermal stability.
Capillary Underfill (CUF) Samples
Liquid capillary underfills are flowed under a populated flip-chip or BGA test vehicle and subsequently cured. Both the bulk material and the filled assembly are evaluated: the bulk material for intrinsic thermal properties, and the filled component for realistic behavior in the presence of solder joints, bumps, and substrate interfaces.
No-Flow Underfill (NFU) and Molded Underfill (MUF) Samples
No-flow underfills, which cure during the solder reflow step, are evaluated for thermal stability after exposure to one or more reflow excursions peaking at 240–260°C. Molded underfills are prepared as molded plaques or populated strip samples reflecting the compression molding process. These materials frequently face more severe thermal excursions than standard CUFs, so decomposition onset data are especially critical.
Aged Specimens
For thermal aging studies, replicate specimens are prepared in sufficient quantity to allow destructive testing at multiple aging intervals, such as 0, 250, 500, 1000, and 2000 hours, so that property drift can be plotted against exposure time.
Testing Items
A comprehensive underfill thermal stability evaluation includes the following measurable parameters:
- Glass transition temperature (Tg) by DSC and TMA, measured before and after thermal exposure to detect post-curing or degradation-induced shifts.
- Coefficient of thermal expansion (CTE), reported as alpha 1 (below Tg, typically 25–45 ppm/°C) and alpha 2 (above Tg, typically 80–120 ppm/°C), plus CTE after aging.
- Decomposition temperature (Td), defined by TGA as the temperature at 5% weight loss (T5%), frequently 300–360°C for qualified epoxy underfills.
- Thermal weight loss at fixed temperatures, such as percent mass loss after 24–168 hours at 150°C or 175°C.
- Residual cure and exotherm enthalpy by DSC, confirming full crosslinking before stability testing.
- Thermomechanical retention, including storage modulus retention by DMA and die shear strength retention after thermal aging.
- Delamination resistance, evaluated by scanning acoustic microscopy (C-SAM) of aged components.
- Temperature cycling survivability per JEDEC JESD22-A104, typically 500–1000 cycles from –55°C to 125°C or –65°C to 150°C.
- Thermal shock resistance per JEDEC JESD22-A106 for liquid-to-liquid excursion testing.
- Flammability classification per UL 94 where end-product safety requirements apply.
Testing Methods
Thermogravimetric Analysis (TGA)
TGA quantifies mass change as a function of temperature or time and is the primary method for decomposition characterization. A representative procedure aligned with ASTM E1131 and ISO 11358 uses 5–10 mg of cured underfill in an open alumina or platinum pan, heated from ambient to 600°C at 10°C/min under nitrogen purge (40–60 mL/min). Key outputs include the 5% and 10% weight-loss temperatures (T5%, T10%), the maximum-rate decomposition temperature from the derivative (DTG) curve, and residue at 600°C. An isothermal variant holds the sample at 150°C or 175°C for 2–4 hours to quantify volatile loss, unreacted diluent evaporation, and early-stage degradation, with weight loss below 1% generally considered acceptable.
Differential Scanning Calorimetry (DSC)
DSC measures Tg and residual cure in accordance with ASTM E1356 and ISO 11357. A 5–15 mg specimen is heated at 10°C/min under nitrogen, typically from –50°C or 25°C to 250–280°C. The midpoint of the step transition in heat flow defines Tg. Absence of a residual exotherm confirms complete cure; a residual enthalpy exceeding roughly 2% of the total cure enthalm indicates under-cure that would compromise subsequent stability data. A second heating scan on the same specimen reveals whether the first scan drove additional crosslinking (Tg increase) or degradation (Tg decrease).
Thermomechanical Analysis (TMA)
TMA determines CTE and dimensional stability per ASTM E831 and IPC-TM-650 method 2.4.24. A specimen of 2–4 mm thickness is heated at 5°C/min with a low-force quartz probe, and the slope change in the expansion curve identifies Tg and defines alpha 1 and alpha 2. Because solder joints operate below the underfill Tg under normal service conditions, a Tg comfortably above the maximum operating temperature (a margin of at least 25–50°C is common) and a low alpha 1 are primary selection criteria.
Long-Term Thermal Aging
Bulk and assembled specimens are placed in a forced-convection aging oven at a temperature representative of worst-case use or burn-in conditions, commonly 150°C, 175°C, or 200°C. At scheduled intervals, specimens are withdrawn and retested for Tg, CTE, weight loss, die shear strength (per ASTM D2197 or die-shear-specific procedures), and interfacial integrity by C-SAM. A material is generally judged thermally stable when Tg drift remains within approximately ±10°C, weight loss stays below 2–3%, and shear strength retention exceeds 70–80% after the qualification duration.
Temperature Cycling and Thermal Shock
Package-level stability is validated by temperature cycling per JEDEC JESD22-A104 (air-to-air, condition G or H, 500–1000 cycles) and thermal shock per JEDEC JESD22-A106 (liquid-to-liquid). After cycling, samples undergo electrical test, dye-and-pry or cross-sectional analysis, and C-SAM inspection to detect underfill cracking, delamination, or solder fatigue. Preconditioning per J-STD-020 or J-STD-020-referenced moisture sensitivity levels may precede cycling to simulate realistic assembly history.
Data Interpretation
Results are consolidated into a stability profile: TGA establishes the absolute upper thermal limit, DSC and TMA define the safe operating window below Tg, and aging plus cycling data demonstrate retention of that window over time. Any downward Tg trend, rising alpha 2, or accelerating weight loss signals oxidation-driven or hydrolysis-driven degradation that shortens service life and warrants reformulation or process adjustment.
Applications
Thermal stability test data for underfill adhesives directly support design and qualification decisions across the electronics industry:
- Flip-chip and chip-scale package assembly, where the underfill must survive multiple reflow excursions above 240°C without decomposition or voiding.
- Wafer-level chip-scale packaging (WLCSP) and panel-level packaging, where thin underfill layers must maintain CTE control across large areas.
- Automotive electronics compliant with AEC-Q100 Grade 1 (–40°C to 125°C ambient) and Grade 0 requirements, which demand demonstrated stability through extended temperature cycling and high-temperature operating life.
- High-performance computing and 2.5D/3D advanced packaging, where high-power die create sustained junction temperatures of 125°C or more.
- Memory module reinforcement, board-level reliability enhancement, and ruggedized industrial or aerospace assemblies requiring long service life at elevated temperature.
- Supplier incoming inspection and material change qualification, where thermal stability data serve as lot acceptance and comparability evidence.
FAQ
What temperature stability should a qualified underfill adhesive have?
A typical qualified epoxy underfill exhibits Tg of 110–160°C, T5% decomposition above 320°C, weight loss below 1% after prolonged 150–175°C exposure, and survival of 1000 temperature cycles from –55°C to 125°C without delamination. Requirements are adjusted upward for automotive, high-power, or reflow-intensive applications.
Why is Tg measured by both DSC and TMA?
DSC detects Tg from the heat capacity step and is the reference method for the bulk material, while TMA detects Tg from the expansion-rate change, which reflects the property most relevant to stress in a solder joint. The two values can differ by 5–15°C, so both are reported for a complete picture.
How is T5% decomposition temperature determined?
A cured specimen of 5–10 mg is heated at 10°C/min in nitrogen to 600°C in a TGA. T5% is the temperature at which cumulative mass loss reaches 5% of the starting mass, corrected for any low-temperature moisture loss. Values below roughly 300°C usually indicate volatile content, incomplete cure, or degradation-prone chemistry.
How long does a thermal stability test program take?
Instrumental characterization (TGA, DSC, TMA) requires only a few days including cure verification. Long-term aging to 1000 hours plus 500–1000 temperature cycles typically extends the program to six to ten weeks. Accelerated aging at higher temperatures can shorten the schedule when an Arrhenius-based lifetime extrapolation is acceptable.
Which standards govern underfill thermal stability testing?
Commonly referenced standards include ASTM E1131 and ISO 11358 for TGA, ASTM E1356 and ISO 11357 for DSC, ASTM E831 and IPC-TM-650 for TMA and CTE, IPC-SM-817 for underfill qualification, JEDEC JESD22-A104 and JESD22-A106 for cycling and shock, JEDEC J-STD-020 for preconditioning, and UL 94 for flammability classification. Accredited third-party laboratories routinely execute these procedures and issue standards-based reports suitable for customer qualification submissions.
Can thermal stability testing predict field lifetime?
Yes, within limits. By combining aging kinetics at several temperatures with an Arrhenius model, and by correlating cycling data with Coffin–Manson fatigue relationships, engineers can extrapolate a use-temperature lifetime estimate. The prediction should always be validated against package-level reliability results, since interfaces, geometry, and process defects influence real-world durability beyond intrinsic material stability.