Evaluation of Martens Heat Resistance Test Conditions for Underfill Adhesives

Underfill adhesives are highly filled epoxy encapsulants whose thermal endurance directly governs the long-term reliability of flip-chip, BGA, and chip-scale package assemblies. The Martens heat resistance test defines the temperature at which a cured specimen bar deflects by a specified amount under constant bending stress, so the evaluation of test conditions—cure schedule, specimen preparation, stress level, heating rate, and conditioning—is essential for reproducible and specification-grade results. This article presents a systematic framework for establishing, validating, and reporting Martens heat resistance test conditions for underfill adhesives with reference to GB, ISO, and ASTM standards.

Technical Overview

Underfill adhesives are one-component, latent-cure epoxy systems, typically loaded with 60–75 wt% fused silica filler, that are dispensed beneath silicon dies and area-array interconnects to redistribute thermomechanical stress away from solder joints. Because these materials operate in close proximity to heat-dissipating die, their heat resistance temperature is a primary acceptance parameter in package qualification and incoming material control.

The Martens heat resistance temperature is the temperature at which a rigid bar specimen, loaded in three-point bending to a nominal surface bending stress of 5 MPa, reaches a deflection reading of 6 mm at the magnifying indicator while the surrounding heat-transfer medium is raised at a uniform rate of 50 °C/h. The method originates from the classical Martens apparatus convention and is standardized for rigid thermosetting materials in GB/T 1699, while related deflection-under-load methods are covered by ISO 75-1, ISO 75-2, GB/T 1634, and ASTM D648, and glass transition measurements by ISO 11357-2 and ISO 11359-2.

Why Condition Evaluation Is Required

Underfill adhesives cure rapidly (snap-cure or conventional bake), contain high filler loadings, and are hygroscopic to a measurable degree. The measured Martens temperature can shift by 5–20 °C depending on degree of cure, moisture history, heating rate, and applied stress level. Consequently, a documented evaluation of test conditions is mandatory before the result can be used for datasheet declaration, supplier benchmarking, or contractual specification.

Governing References

• GB/T 1699 — Determination of Martens heat resistance temperature of rigid thermosetting materials.
• ISO 75-1 / ISO 75-2 / GB/T 1634 / ASTM D648 — Plastics: deflection temperature under load (for comparison and cross-referencing).
• ISO 11357-2 — Differential scanning calorimetry determination of glass transition temperature.
• ISO 11359-2 — Thermomechanical analysis determination of glass transition temperature.
• ISO 291 — Standard atmospheres for conditioning and testing.
• ISO/IEC 17025 — General requirements for the competence of testing laboratories.

Testing Samples

Cast Bulk Bars

The reference specimen is a cast bar of 120 mm × 15 mm × 10 mm prepared in clean, release-coated steel or silicone molds. The underfill resin should be warmed to 40–60 °C and vacuum-degassed for 5–10 minutes before casting to minimize entrapped air, and the mold should be tapped or vibrated to release bubbles at the walls. Casting must be followed immediately by curing to prevent filler settling, which creates a density gradient and asymmetrical mechanical properties through the thickness.

Machined Bars

Where casting of thin, highly viscous resins is impractical, bars may be machined from cured plaques or thick slabs, with all faces milled and edges lightly dressed to remove stress raisers. Machined specimens must be marked to preserve orientation relative to the casting direction so that the bending axis is consistent across the sample set.

Conditioning States

Specimens shall be assigned to defined conditioning states before testing: dry-as-cured (desiccator, 24 h), standard atmosphere (23 °C / 50% RH, 48 h, per ISO 291), and, where the application demands it, moisture-saturated (85 °C / 85% RH, 168 h) to bracket the plasticized worst case. A minimum of three valid specimens per condition is required, with five specimens recommended for arbitration or specification testing.

Testing Items

The evaluation program comprises the following measured and derived items:

• Martens heat resistance temperature (°C), reported as the arithmetic mean of valid specimens with the observed range.
• Cure-conversion sensitivity: Martens temperature as a function of cure schedule (supplier minimum cure, recommended cure plus post-cure, extended over-cure).
• Conditioning sensitivity: Martens temperature after dry, standard-atmosphere, and humidity-saturated conditioning.
• Heating-rate sensitivity: comparison of 50 °C/h Martens protocol against 120 °C/h deflection-under-load protocols.
• Stress-level sensitivity: classical 5 MPa Martens stress versus 1.8 MPa and 0.45 MPa stress levels used in ISO 75-type procedures.
• Cross-validation against DSC and TMA glass transition temperatures per ISO 11357-2 and ISO 11359-2.
• Precision metrics: within-laboratory repeatability and specimen-to-specimen scatter.

Testing Methods

Apparatus Configuration

A Martens heat resistance apparatus consists of a temperature-controlled bath (silicone oil or forced-air oven), a specimen support with a 100 mm span, a lever-loading system that applies the equivalent of a 50 N midpoint load (5 MPa nominal surface bending stress on a 15 mm × 10 mm section), and a magnifying deflection indicator reading to at least 0.1 mm. The bath temperature shall be measured with a calibrated sensor positioned adjacent to the specimen, and the lever geometry shall be verified with a certified mass before each campaign.

Baseline Test Conditions

The baseline condition set is: specimen 120 mm × 15 mm × 10 mm; support span 100 mm; nominal bending stress 5 MPa; starting bath temperature 25–30 °C, held until thermal equilibrium; heating rate 50 °C/h (tolerance ±5%); endpoint at a 6 mm deflection reading at the indicator. The recorded Martens temperature is the bath temperature at the moment the endpoint deflection is reached. Testing is discontinued and the specimen rejected if it cracks, slips on the supports, or shows visible voids at the failure surface.

Condition Evaluation Matrix

The evaluation matrix is executed as a controlled comparison in which one parameter is varied at a time:

• Cure schedule: Condition A (supplier minimum cure, e.g., 150 °C/30 min or 165 °C/10 min), Condition B (recommended cure plus post-cure, e.g., 150 °C/1 h plus 150 °C/2 h), Condition C (deliberate over-cure, e.g., 175 °C/2 h). A well-designed material shows a 5–15 °C increase from A to B and a plateau from B to C; Condition B is then fixed as the reference cure.
• Conditioning: dry, 23 °C/50% RH, and humidity-saturated states. Humidity-saturated specimens typically read 5–20 °C lower due to epoxy plasticization; the declared specification must state which conditioning applies.
• Heating rate: 50 °C/h versus 120 °C/h. Faster heating generally shifts the apparent temperature upward by several degrees due to thermal lag; the 50 °C/h rate shall be retained for Martens comparability.
• Stress level: 5 MPa versus 1.8 MPa. Values obtained at lower stress are systematically higher and shall never be mixed or averaged with classical Martens results.
• Specimen provenance: cast versus machined bars, assessed for systematic offset due to surface effects or filler orientation.

Precision and Acceptance Criteria

Within-laboratory repeatability is considered acceptable when the specimen-to-specimen range does not exceed 3 °C for a single operator and apparatus. Ranges between 3 °C and 5 °C trigger a root-cause review of cure uniformity, bath agitation, and lever friction; ranges above 5 °C invalidate the set. Outliers may be discarded only for documented physical causes (voids, slippage, misalignment) and never on statistical grounds alone.

Correlation with Thermal Analysis

For fully post-cured, well-filled epoxy underfills, the Martens temperature typically falls within roughly 10–25 °C below the DSC glass transition midpoint, because the 5 MPa bending stress causes deflection to accumulate before the full transition is traversed. This relationship is formulation-dependent and shall be verified empirically for each material family using ISO 11357-2 (DSC) and ISO 11359-2 (TMA) rather than assumed; a consistent offset confirms cure completeness, while a drifting offset usually indicates incomplete conversion or moisture uptake.

Test Report Content

The final report shall state: material identification and lot; specimen type (cast or machined) and dimensions; cure and post-cure schedule; conditioning state; apparatus and calibration status; stress level and heating rate; individual and mean Martens temperatures with range; number of specimens tested and any rejections with cause; and the governing standard (GB/T 1699 for the Martens protocol, with ISO 75 or ASTM D648 cited where comparison data are included).

Applications

Evaluation of Martens heat resistance test conditions supports the following engineering activities:

• Qualification of capillary underfills for flip-chip assemblies, where the heat resistance temperature must exceed the package maximum continuous operating temperature with a documented safety margin.
• Corner-bond and edge-bond adhesive selection for BGA and chip-scale package board-level assembly.
• Automotive electronics programs, where underfills in engine-adjacent modules face ambient temperatures of 125–150 °C and require demonstrated thermal endurance.
• Stacked-die, package-on-package, and wafer-level packaging platforms in which molding and rework thermal exposures stress the cured underfill layer.
• Incoming inspection and supplier datasheet verification, using fixed reference conditions so that lot-to-lot comparisons are statistically meaningful.
• Failure analysis, where a depressed Martens temperature relative to the established baseline provides evidence of under-cure, moisture exposure, or formulation deviation in field returns.

Once the condition evaluation is complete, the fixed reference condition set—cure schedule, conditioning, stress, and heating rate—becomes the contractual basis for routine quality control testing by accredited third-party laboratories operating under ISO/IEC 17025.

FAQ

Q: What does Martens heat resistance measure, and how does it differ from ISO 75 heat deflection temperature?

A: The Martens method loads a bar to a 5 MPa surface bending stress, heats at 50 °C/h, and reports the temperature at a 6 mm indicator deflection. ISO 75-type procedures use lower stress levels (typically 1.8 MPa or 0.45 MPa), different specimen orientations, and a 120 °C/h rate. The two numbers correlate but are not interchangeable, and any comparison must state both condition sets.

Q: Why does an underfill adhesive require a dedicated condition evaluation rather than direct testing?

A: Underfills are snap-cure, highly filled systems whose measured temperature depends strongly on cure conversion, moisture history, and thermal protocol. Without a documented condition evaluation, differences of 5–20 °C caused by the test setup can be misattributed to the material and corrupt acceptance decisions.

Q: How many specimens should be tested?

A: A minimum of three valid specimens per condition is required, with five recommended for specification or arbitration testing. The report shall include individual values, the mean, and the range.

Q: What Martens temperature values are typical for underfill adhesives?

A: Properly cured, silica-filled epoxy underfills generally exhibit Martens temperatures in the approximate range of 110–170 °C, depending on resin chemistry, filler loading, and cure schedule. The acceptance limit should be set relative to the application's maximum continuous operating temperature with an engineering margin defined by the package designer.

Q: Can the test be performed directly on an assembled package?

A: No. The Martens method requires standardized bulk bars. Package-level thermal endurance is assessed separately through device-level qualification testing; the bulk-bar Martens temperature characterizes the cured adhesive itself.

Q: Where should these tests be performed?

A: Testing can be conducted in-house with calibrated Martens apparatus or outsourced to ISO/IEC 17025-accredited third-party laboratories. In either case, the laboratory must document the evaluated condition set—cure, conditioning, stress, and heating rate—alongside every reported result.