Underfill adhesives safeguard flip-chip, BGA, and CSP assemblies against thermomechanical fatigue, and their Martens (Martin) heat resistance temperature is a key indicator of stiffness retention at elevated temperature. This article explains the Martens heat resistance test method for cured underfill materials and provides practical guidance on data processing, statistical evaluation, and acceptance criteria. Engineers involved in semiconductor packaging, material qualification, and reliability assessment can use this framework to convert raw deflection-temperature records into defensible material judgments.
Technical Overview
Underfill adhesives are typically one-component or two-component epoxy systems, frequently loaded with 50–75 wt% fused or crystalline silica, that are dispensed beneath flip-chip dies and ball-grid-array packages and cured to form a rigid, low-expansion support matrix. These materials must survive solder reflow peaks of 245–260 °C during assembly and maintain load-bearing rigidity during continuous operation at 100–150 °C in high-power and automotive-grade electronics.
The Martens heat resistance test, described in standards such as GB/T 1699 and historically in DIN 53462, quantifies the temperature at which a rigid polymer specimen under a defined bending stress softens to a specified deflection. Unlike the Vicat softening point covered by ISO 306 or the heat deflection temperature covered by ISO 75-2 and ASTM D648, the Martens method applies a comparatively high bending stress of approximately 5 MPa and uses a 6 mm indicator deflection criterion, making it a conservative measure of hot rigidity. For underfill adhesives, the resulting Martens temperature reflects the material's ability to restrain solder joint displacement and die warpage under combined thermal and mechanical loading, complementing Tg data obtained by DSC or DMA.
Testing Samples
Standard Bar Specimens
Cured underfill resin is cast or compression-molded into flat bars of 120 mm × 15 mm × 10 mm, matching the geometry required by GB/T 1699 and comparable Martens procedures. Because the effective bending stress scales with the section modulus of the specimen, thickness must be controlled within ±0.2 mm and measured at the mid-span region to 0.02 mm. Specimens must be free of voids, bubbles, delamination, and sink marks, since internal defects near the outer fibers will depress the measured heat resistance temperature.
Curing and Conditioning
Specimens should be cured using the schedule recommended by the adhesive supplier, for example 150 °C for 1–2 hours for snap-cure capillary underfills or 165 °C for 30–60 minutes for conventional systems, followed by any specified post-cure. After cure, specimens are conditioned at 23 ± 2 °C and 50 ± 5% relative humidity for 24 to 48 hours in accordance with ISO 291 before testing. Deviations from the nominal cure schedule must be recorded, because under-cure systematically lowers the Martens temperature.
Sampling Plan
A minimum of three, preferably five, replicate specimens should be tested for each material lot or process condition. Batch numbers, dispensing or casting date, storage history, and shelf-life status must be documented so that any anomalous result can be traced to its origin.
Testing Items
The core deliverable of the test program is the Martens heat resistance temperature, supported by auxiliary measurements that strengthen the overall evaluation:
- Martens (Martin) heat resistance temperature at 5 MPa bending stress and 6 mm deflection criterion
- Full deflection-temperature curve, including softening onset identification
- Glass transition temperature by DSC per ISO 11357-2, for cross-reference
- Storage modulus profile by DMA per ISO 6721, for stiffness correlation
- Heat deflection temperature per ISO 75-2 or ASTM D648, for comparative benchmarking
- Optional Martens temperature after moisture preconditioning or thermal aging, for environmental sensitivity assessment
Testing Methods
Apparatus and Loading Configuration
The Martens apparatus comprises a vertical specimen holder, a lever-and-weight loading system, and a dial indicator or optical scale at the lever end, all housed within a stirred oil bath or forced-convection air oven. Weights are selected so that the specimen outer fiber is subjected to a nominal bending stress of 5 MPa. The lever arm amplifies the true specimen deflection, and the lever ratio must be verified during apparatus calibration.
Test Procedure
The specimen is measured, mounted vertically, loaded, and the indicator is zeroed at the starting temperature, typically 25–30 °C or at least 50 °C below the expected Martens temperature. The bath is then heated at a controlled rate of 50 ± 3 °C/h. Deflection readings are recorded at regular intervals, typically every 5 minutes or every 2–3 °C, with denser sampling as the deflection rate accelerates. The test terminates when the indicator deflection exceeds 6 mm or when a pre-set safety limit is reached.
Raw Data Records
Each run produces a paired dataset of bath temperature and indicator deflection, together with specimen geometry, applied stress, heating medium, thermocouple position, and any procedural deviations. Complete raw records are essential for traceable data processing and evaluation.
Data Processing and Evaluation
Curve Construction and Interpolation
Indicator deflection is plotted against temperature for each specimen. The curve typically shows an initially flat glassy region followed by a steep rise near the glass transition. The Martens temperature is determined by linear interpolation between the two data points bracketing the 6 mm deflection; polynomial smoothing may be applied, but extrapolation beyond the measured range is not acceptable. Where reporting requires actual specimen deflection rather than lever-end reading, the value is divided by the calibrated lever ratio.
Statistical Treatment
Replicate results are reduced to the arithmetic mean, standard deviation, and coefficient of variation, with individual values reported alongside the summary statistics. Outliers are screened using the Grubbs test at 95% confidence or the practices of ASTM E178, and any rejected value must be investigated for root causes such as internal voids, load misalignment, or incomplete cure. Well-prepared filled underfill specimens should achieve a coefficient of variation below 3% within a single laboratory.
Corrections and Quality Checks
Before results are released, the following checks should be completed:
- Verify thermometer or thermocouple calibration and immersion depth correction against traceable references
- Confirm bath temperature uniformity within 1–2 °C across all specimen positions
- Confirm the actual heating rate, since rates above specification artificially inflate the Martens temperature
- Recalculate the effective bending stress from measured specimen width and thickness to confirm the 5 MPa nominal value
Interpretation and Acceptance Criteria
The Martens temperature of a filled underfill generally lies slightly below the DSC Tg midpoint and typically below the heat deflection temperature measured at 1.80 MPa, reflecting the higher applied bending stress of the Martens configuration. For acceptance, the mean Martens temperature should exceed the maximum continuous operating temperature of the assembly by a safety margin of 10–25 °C, and lot-to-lot consistency should fall within the specification band agreed with the supplier. The final report should include specimen geometry and cure schedule, conditioning state, stress level, heating rate and medium, individual and mean Martens temperatures with standard deviation, the deflection-temperature plot, and any deviations from the standard procedure.
Applications
Martens heat resistance data for underfill adhesives support a wide range of engineering decisions across electronics packaging:
- Qualification of capillary, no-flow, and molded underfill materials for flip-chip and wafer-level packaging
- Corner reinforcement and edge-bond adhesives for large BGA and CSP assemblies
- Comparative evaluation of formulation changes, such as filler loading, hardener chemistry, or catalyst level, on high-temperature rigidity
- Assessment of reflow-crack resistance and die-warpage control in stacked-die and fan-out packages
- Material screening for high-power RF modules, automotive engine-control units, and 5G infrastructure hardware
- Incoming quality control and supplier lot-consistency monitoring in production environments
FAQ
What does the Martens heat resistance temperature represent?
It is the temperature at which a cured underfill bar, held under a 5 MPa bending stress and heated at 50 °C/h, deflects by the 6 mm indicator criterion. It represents the practical upper limit of load-bearing stiffness under combined thermal and mechanical stress.
Which standards govern the test?
The Martens method is specified in GB/T 1699 and historically in DIN 53462. Related heat-distortion methods include ISO 75-2 and ASTM D648 for HDT, and ISO 306 and ASTM D1525 for Vicat softening, each of which applies different stress and deflection criteria.
Why are 5 MPa and 6 mm used as the criteria?
These parameters originate from the classical Martens apparatus geometry and provide a deliberately conservative assessment, applying higher bending stress than the common 1.80 MPa HDT condition so that the reported temperature reflects worst-case mechanical loading.
How many specimens are required and how are outliers handled?
At least three, preferably five, replicate specimens should be tested. Statistical outliers are identified using the Grubbs test or ASTM E178 practice, documented, and investigated for root cause before the final mean value is accepted.
How does the Martens temperature relate to DSC Tg and HDT?
The Martens temperature typically falls slightly below the DSC Tg midpoint and below HDT measured at 1.80 MPa because of the higher applied stress. DMA storage modulus data provide the clearest mechanistic link, showing the modulus collapse that drives the deflection acceleration.
What Martens temperature should an underfill adhesive have?
It should exceed the maximum continuous operating temperature of the assembly by at least 10–25 °C. Commercial filled capillary underfills commonly deliver Martens temperatures in the range of 110–190 °C depending on formulation and cure schedule.
Can the test be run in air instead of oil?
Yes, a forced-convection air oven variant exists, but heat transfer is less uniform than in a stirred oil bath and results may shift slightly. The heating medium must always be stated in the report, and comparisons should only be made between results obtained with the same medium and heating rate.