Martens Heat Resistance Testing Method for Underfill Adhesives

Martens heat resistance testing is a classical thermal-softening method used to evaluate how cured underfill adhesives behave under combined bending stress and rising temperature. This technical guide explains the testing principle, specimen preparation, apparatus, procedure, and data interpretation for underfill adhesives, with reference to recognized standards such as GB/T 1699 and the related ISO 75 and ASTM D648 heat deflection frameworks. It is intended for materials engineers, packaging engineers, and quality-control professionals involved in the qualification of underfill materials for advanced electronic assemblies.

Technical Overview

Underfill adhesives are low-viscosity, typically epoxy-based thermosetting materials that are dispensed into the gap between a silicon die, flip-chip, or ball grid array (BGA) component and its substrate. After curing, the underfill forms a rigid, highly silica-filled composite that mechanically couples the component to the board and redistributes thermomechanical stress away from the solder interconnects. Because assemblies are repeatedly exposed to elevated temperatures during solder rework, reflow processing, and long-term operation, the heat resistance of the cured underfill is a critical material property.

Martens heat resistance, usually reported as the Martens temperature, is a heat-resistance index determined by heating a cured specimen at a controlled rate while it is subjected to a constant bending stress. The temperature at which the specimen softens sufficiently to produce a defined deflection is recorded as the Martens temperature. The method follows the principle established in GB/T 1699, which was originally developed for hard rubber (ebonite) and is widely applied in industry to rigid thermosetting polymers, including filled epoxy systems such as cured underfills. Internationally, the closest equivalent concept is the heat deflection temperature (HDT) method standardized in ISO 75-1 and ISO 75-2, as well as ASTM D648.

• The test evaluates a cured, rigid polymer specimen, not the liquid adhesive.
• A constant nominal bending stress of approximately 5 MPa is applied through a lever-and-weight system.
• The oven temperature is raised at a uniform rate of about 50 °C per hour.
• The endpoint is the temperature at which the deflection indicator moves by 6 mm.
• The result is a practical softening index that correlates with, but is not identical to, the glass transition temperature measured by DSC or DMA.

For highly filled epoxy underfills, Martens temperatures typically fall in the approximate range of 90 °C to 150 °C, depending on resin chemistry, filler loading, and degree of cure. A higher Martens temperature generally indicates better dimensional stability at elevated temperature and greater resistance to softening during subsequent thermal processes.

Testing Samples

Cast and Cured Specimen Bars

Because underfill adhesives are supplied as liquids, test specimens must first be cast in molds and fully cured. Standard specimen bars are typically 120 mm in length, 15 mm in width, and 10 mm in thickness, in accordance with the dimensions specified in the Martens method. The liquid adhesive should be degassed before casting to remove entrapped air, and curing should follow the supplier-recommended schedule, for example 150 °C for one hour or an equivalent validated profile.

Machined Specimens from Cast Plaques

Alternatively, the adhesive may be cast into a flat plaque and cured, after which specimen bars are machined or cut to the required dimensions. Machined surfaces should be flat, smooth, and free of cracks, chips, and voids, since surface defects can act as stress concentrators and cause premature deflection or premature failure.

Customer-Submitted Cured Samples

Accredited third-party laboratories frequently accept customer-submitted cured bars or plaques for testing. Submitted samples must meet the dimensional requirements of the method, and the cure condition, batch number, and material identification should accompany the samples so that results can be correctly interpreted.

Sample Conditioning and Preparation

Prior to testing, specimens should be conditioned in a standard laboratory atmosphere of 23 ± 2 °C and 50 ± 5 % relative humidity, consistent with ISO 291 or the equivalent conditioning requirements of the applicable standard. At least three valid specimens should be prepared for each material condition, and all specimens should be visually inspected for voids, incomplete cure, or surface defects before mounting.

Testing Items

The core deliverable of the test is a single, well-defined thermal index, supplemented by supporting observations:

• Martens heat resistance temperature, expressed in degrees Celsius.
• Deflection-versus-temperature behavior during the test, where a continuous readout is available.
• Comparison of Martens temperature before and after thermal aging, moisture conditioning, or post-cure, as an optional supplementary study.
• Verification of specimen quality, including visual inspection for voids, cracks, and cure-related defects.

Results for individual specimens are reported together with the arithmetic mean, and the spread between specimens provides an indication of material homogeneity and preparation consistency.

Testing Methods

Apparatus

The test is performed on a Martens heat resistance apparatus, which consists of a heated oven with a controlled temperature ramp, a vertical specimen fixture, a lever loading mechanism with calibrated weights, and a deflection indicator with a scale readable to at least 0.5 mm. Temperature measurement is made with a calibrated thermometric device positioned near the specimen, with an accuracy of approximately ± 1 °C.

Test Conditions

• Nominal bending stress: approximately 5 MPa applied at the surface fibers of the specimen.
• Heating rate: 50 °C per hour, maintained uniformly after the test begins.
• Endpoint criterion: deflection of 6 mm indicated at the measuring scale.
• Starting temperature: standard laboratory ambient conditions, typically 23 ± 2 °C.
• Number of specimens: at least three valid replicates per condition.

Test Procedure

• Measure and record the width and thickness of each specimen at multiple points, and verify that dimensions are within tolerance.
• Mount the specimen vertically in the fixture so that it is rigidly clamped at the lower end, with the loading lever attached at the upper end.
• Select and position the weights so that the nominal bending stress on the specimen equals the specified value of approximately 5 MPa, based on the measured cross-section.
• Set the deflection indicator to zero, close the oven, and begin heating at the specified rate of 50 °C per hour.
• Observe the deflection indicator continuously and record the oven temperature at the moment the indicator shows the 6 mm endpoint deflection; this temperature is the Martens temperature for that specimen.
• Repeat the test for the remaining specimens and report the individual values, the average, and any observations such as cracking or non-uniform bending.

Validity and Quality Control of Results

Individual Martens temperatures from replicate specimens of a well-prepared material should normally agree within a few degrees Celsius. If the spread between specimens is unusually large, possible causes include voids, incomplete or non-uniform cure, moisture uptake, dimensional variation, or misalignment in the fixture. Suspect specimens should be examined and the test repeated. Results can also be cross-checked against glass transition temperature data obtained by DSC in accordance with ISO 11357 or GB/T 19466, or by DMA in accordance with ISO 6721, since the Martens temperature of a thermoset is closely related to the softening associated with the glass transition.

Relationship to HDT Methods

When results must be aligned with international material datasheets, the same cured specimens may also be evaluated by HDT testing under ISO 75-2 or ASTM D648 at applied fiber stresses of 1.80 MPa or 0.45 MPa, with the heating rate and deflection endpoint defined by those standards. The two methods share the same underlying principle of stress-assisted softening, but they differ in loading geometry, stress level, heating rate, and endpoint, so the resulting temperatures are related but not interchangeable.

Applications

Martens heat resistance data for underfill adhesives support a wide range of engineering and quality decisions in electronics manufacturing:

• Qualification of capillary underfill materials for flip-chip and BGA assemblies, where thermal softening of the underfill would compromise solder joint protection.
• Evaluation of corner-bond and edge-bond adhesives for chip-scale packages subjected to reflow and wave soldering temperatures.
• Verification of cure completeness, since an under-cured epoxy network exhibits a markedly lower Martens temperature than a fully cured one.
• Comparison of candidate materials from different suppliers under identical specimen preparation and test conditions.
• Assessment of thermal degradation after high-temperature aging or temperature-cycling preconditioning of assemblies.
• Support for automotive and industrial electronics programs, in which underfill performance is examined alongside reliability and qualification frameworks such as those of AEC-Q and IPC standards.
• Incoming inspection and lot-to-lot quality control of adhesive shipments in production environments.

In all of these contexts, the Martens temperature serves as a practical, reproducible index of how far the cured adhesive can be pushed in temperature before its load-bearing capability deteriorates.

FAQ

What is the difference between the Martens temperature and the glass transition temperature?

The Martens temperature is a mechanically driven softening index measured under a defined bending stress and heating rate, whereas the glass transition temperature is a physical property of the polymer network typically measured by DSC, TMA, or DMA. The two values are correlated for thermosets, but the Martens temperature is usually somewhat lower than the DSC midpoint glass transition temperature and reflects engineering behavior under load rather than a purely thermal transition.

What Martens temperature should a good underfill adhesive have?

Cured epoxy underfills commonly show Martens temperatures between roughly 90 °C and 150 °C. The required value depends on the application: assemblies exposed to lead-free reflow peaks of approximately 250 °C rely on the underfill remaining dimensionally stable well above its softening point, and higher values are generally preferred for demanding thermal environments. Material selection should always consider the Martens temperature together with the glass transition temperature, coefficient of thermal expansion, modulus, and adhesion data.

Why is it necessary to cast and cure specimens instead of testing the adhesive directly?

The Martens method evaluates the rigid, cured state of the material. Underfill adhesives are supplied as liquids, so representative bars must be cast, degassed, and cured under the recommended schedule before testing. The measured Martens temperature is therefore a property of the cured network, and it depends strongly on the degree of cure achieved during specimen preparation.

What factors most commonly cause variable results?

Frequent sources of variability include entrapped air bubbles or voids in cast bars, incomplete or non-uniform cure, moisture absorbed by hygroscopic epoxy specimens, dimensional deviations from the specified bar geometry, and fixture misalignment that alters the effective bending stress. Careful molding, validated curing, standard conditioning, and calibrated apparatus are essential for reproducible results.

Which standards govern this test?

The Martens heat resistance determination follows the principle established in GB/T 1699, which is applied in practice to rigid thermosetting materials such as cured underfill adhesives. Related international frameworks include ISO 75-1 and ISO 75-2 for heat deflection temperature, ASTM D648 for deflection temperature under flexural load, ISO 291 for conditioning atmospheres, and ISO 11357 or GB/T 19466 for supporting DSC measurements. Testing can be performed by accredited third-party laboratories or by in-house facilities with calibrated equipment.

How does the curing schedule affect the measured value?

The Martens temperature rises as the crosslinking reaction approaches completion. An under-cured network contains residual reactive groups and lower crosslink density, which lowers the softening point. If a material datasheet specifies a particular cure schedule, such as a defined time at 150 °C, specimens must be cured accordingly, and post-cure effects should be considered when comparing results from differently treated samples.