The coefficient of thermal expansion (CTE) of alumina (Al₂O₃) ceramic substrates is a decisive material parameter for thick-film hybrid circuits, thin-film electronics, power modules, and hermetic packages, where thermal mismatch directly controls solder joint fatigue, metallization adhesion, and assembly reliability. This experiment describes the measurement of the linear thermal expansion of 96% and 99.5% alumina substrates using push-rod dilatometry and thermomechanical analysis in accordance with ASTM E228, ASTM E831, ISO 17562, and GB/T 16535. The article covers test principle, specimen preparation, procedures, data reduction, typical CTE values, industrial applications, and frequently asked questions.
Technical Overview
Alumina (Al₂O₃) substrates are the most widely used ceramic substrates in the electronics industry, offering high electrical resistivity, moderate-to-high thermal conductivity (approximately 20–30 W/(m·K) depending on purity), excellent mechanical strength, chemical inertness, and cost-effective mass production. Alumina content in commercial substrates typically ranges from 92% to 99.9%, with 96% alumina dominating thick-film hybrid applications and 99.5%–99.6% alumina used for thin-film and precision RF circuits.
The coefficient of thermal expansion (CTE) describes the fractional change in length of the material per unit temperature change. For polycrystalline alumina substrates, the mean linear CTE is typically 6.3–7.5 × 10⁻⁶/°C (ppm/°C) between room temperature and 300 °C, increasing gradually to approximately 7.5–8.5 ppm/°C up to 800–1000 °C due to the intrinsic anharmonicity of the corundum lattice and, in lower-purity grades, the softening of the glassy grain-boundary phase.
Accurate CTE data for alumina substrates is essential because virtually every downstream assembly step involves a dissimilar material: silicon die (2.6 ppm/°C), copper metallization and direct-bonded copper (approximately 17 ppm/°C), Kovar lids (approximately 5–6 ppm/°C), and brazing or glass sealing compounds. CTE mismatch generates thermomechanical stress during reflow, power cycling, and environmental temperature swings, which can cause substrate cracking, conductor delamination, solder fatigue, and hermetic seal failure.
- Measurement principle: the specimen is heated at a controlled rate while its length change is continuously recorded; the mean CTE is calculated as α = ΔL / (L₀ × ΔT).
- Primary methods: horizontal or vertical push-rod dilatometry and thermomechanical analysis (TMA).
- Governing standards: ASTM E228, ASTM E831, ASTM E289, ISO 17562, GB/T 16535, and related fine-ceramic test standards.
Testing Samples
96% Alumina Thick-Film Substrates
Standard 96% alumina substrates (typical thickness 0.25–1.5 mm, commonly 0.635 mm and 1.0 mm) are supplied in as-fired or ground surface condition. The glassy silicate grain-boundary phase in this grade gives a slightly lower CTE than high-purity alumina but introduces a modest increase in expansion rate above 500–600 °C. Both white and black (color-added) 96% grades may be evaluated; the coloring oxides have negligible effect on CTE at typical loading levels.
99.5%–99.6% Thin-Film Grade Alumina Substrates
High-purity substrates for thin-film, microwave, and optical applications are supplied as-fired, lapped, or polished, with thicknesses down to 0.127 mm. The finer grain size and minimal glass content produce a slightly higher and more linear CTE (approximately 6.8–7.8 ppm/°C from 25 to 300 °C) and excellent reproducibility between thermal cycles.
Specimen Geometry and Preparation
For push-rod dilatometry per ASTM E228 and ISO 17562, specimens are cut into rectangular strips, typically 25 mm in length with a width of 4–6 mm at full substrate thickness. Proper preparation is critical to measurement quality:
- Both end faces are ground flat and parallel within ±0.01 mm using diamond tooling, since non-parallel ends introduce contact and seating errors.
- Specimens are cut in both in-plane directions (machine direction and transverse direction) for tape-cast substrates to quantify any anisotropy, which is normally below 3–5% for alumina.
- Coupons are ultrasonically cleaned in solvent or deionized water and dried at 110–150 °C for 1–2 hours to remove cutting residues and moisture.
- An optional stabilization anneal (for example 30 minutes at 800 °C) may be applied to relieve residual machining stresses before measurement.
- A minimum of three specimens per material condition is recommended so that mean values and standard deviations can be reported.
Testing Items
The experiment quantifies the thermal expansion behavior of alumina substrates through the following reported items:
- Mean (secant) linear CTE over specified temperature intervals, most commonly 25–100 °C, 25–300 °C, 25–500 °C, 25–800 °C, and 25–1000 °C.
- Instantaneous (tangent) CTE as a continuous function of temperature, derived from the first derivative of the expansion curve.
- Complete fractional length-change curve (ΔL/L₀ versus temperature) across the full test range.
- Repeatability assessment across two consecutive heating cycles to verify structural and dimensional stability.
- In-plane anisotropy ratio between the length and width directions of the substrate.
- Statistical summary: individual values, mean, and standard deviation for the specimen set.
Testing Methods
Push-Rod Dilatometry (ASTM E228 / ISO 17562 / GB/T 16535)
A push-rod dilatometer applies a small, constant contact force (typically 0.2–1 N) through an alumina or fused-silica rod onto the specimen end face, transmitting length change to a high-resolution displacement sensor such as an LVDT with resolution of 0.05–0.1 µm. The specimen rests in a homogeneous-temperature zone of a resistance furnace capable of operating from room temperature to 1000–1600 °C. Testing is performed at a constant heating rate of 3–5 °C/min under a purge of dry nitrogen, argon, or helium to stabilize heat transfer and prevent sensor contamination. The specimen length change is recorded continuously against calibrated specimen temperature, and the measured expansion is corrected for the expansion of the reference dilatometer components.
Thermomechanical Analysis (ASTM E831)
For thin substrate coupons, limited material, or sub-ambient testing, a TMA in expansion mode uses a flat-tipped quartz probe with a very low contact force (approximately 0.02–0.2 N) on specimens 2–10 mm in length. Typical TMA range is −120 °C to 600 °C at 5 °C/min, making the method well suited to evaluating alumina substrates over solder-reflow, automotive (−55 °C to 125 °C), and junction-temperature regimes.
Calibration and Reference Materials
Instrument accuracy is established in three steps traceable to national metrology institutes: zero-point correction using a dummy specimen of the reference material; expansion calibration using a certified single-crystal sapphire or fused-silica reference standard; and temperature calibration using the melting points of pure metals or verified thermocouple standards. A system verification check is performed before each test series, and calibration validity is confirmed by measuring the reference standard as an unknown specimen.
Test Procedure and Data Reduction
Each specimen is loaded with its prepared end faces perpendicular to the measurement axis, the purge gas flow is stabilized, and the furnace is heated at the selected rate to the maximum test temperature, followed by controlled cooling. A first conditioning cycle may be run without data acquisition, with reported results taken from the subsequent cycle. The mean CTE between temperatures T₁ and T₂ is calculated as α(T₁–T₂) = (L₂ − L₁) / (L₀ × (T₂ − T₁)), where L₀ is the initial measured length at the reference temperature. The test report includes specimen identification, purity grade, L₀, temperature program, atmosphere, heating rate, calibration references, tabulated CTE values for each interval, and the complete expansion curve.
Typical measurement capability for alumina substrates is a repeatability of ±0.05–0.1 ppm/°C or ±2–4% relative, whichever is greater. Where the highest precision is required for very low expansion values, interferometric methods per ASTM E289 offer nanometer-level displacement resolution.
Applications
CTE measurement of alumina substrates underpins design and qualification across the electronics and advanced packaging industries:
- Thick-film hybrid integrated circuits, resistor networks, and automotive control modules, where paste systems and substrate must remain compatible through multiple firing and reflow cycles.
- Thin-film and RF/microwave circuits on 99.6% alumina, where multilayer metal stacks are dimensionally stable only if substrate expansion is precisely known.
- Direct-bonded-copper (DBC) and power module substrates for IGBT and SiC applications, where copper–alumina interface stress during the 1065 °C eutectic bonding process and subsequent power cycling is governed by CTE mismatch.
- High-temperature co-fired ceramic (HTCC) multilayer packages, where alumina CTE must match tungsten and molybdenum conductor metallization during co-firing and hermetic sealing.
- Hermetic packages with Kovar or glass-to-ceramic seals, requiring expansion matching to prevent leak-path formation.
- LED ceramic boards, sensor carriers, and thermal-management assemblies in aerospace, rail, and downhole electronics, where wide temperature excursions make validated CTE data mandatory for lifetime modeling.
- Incoming material inspection and supplier qualification of substrate lots at accredited third-party laboratories using standardized dilatometric methods.
FAQ
What is the typical CTE value of an alumina substrate?
A 96% alumina substrate typically exhibits a mean CTE of 6.3–7.2 ppm/°C from 25 to 300 °C, rising to approximately 7.5–8.0 ppm/°C from 25 to 800 °C. High-purity 99.5%–99.6% grades typically measure 6.8–7.8 ppm/°C from 25 to 300 °C and approximately 8.0–8.5 ppm/°C from 25 to 1000 °C.
Why does the CTE of alumina increase with temperature?
The increase is primarily due to the intrinsic anharmonicity of atomic vibrations in the corundum crystal lattice, which raises expansion rate continuously with temperature. In 96% grades, the additional contribution of the glassy silicate grain-boundary phase becomes noticeable above roughly 500–600 °C.
Which test method and standard should be selected?
Push-rod dilatometry per ASTM E228, ISO 17562, or GB/T 16535 is the preferred method for the 25–1000 °C range typical of substrate qualification and brazing simulation. TMA per ASTM E831 is better suited to thin coupons, small samples, or sub-ambient-to-moderate temperature ranges. Interferometry per ASTM E289 is reserved for ultra-low-expansion verification requiring the highest displacement resolution.
How should the temperature interval for the reported CTE be chosen?
The interval should reproduce the relevant service or process condition: 25–260 °C approximates solder reflow, −55 °C to 125 °C covers automotive qualification, and 25–800 °C or higher represents brazing and high-temperature co-firing. Reporting several standard intervals on a single expansion curve allows all design cases to be covered in one experiment.
What accuracy can be expected from the experiment?
With calibrated instruments and properly prepared specimens, typical reproducibility is ±0.05–0.1 ppm/°C or ±2–4% relative, whichever is greater. End-face parallelism, sensor zero drift, furnace temperature gradients, and reference-material corrections are the dominant error sources and are controlled through the standardized procedures described above.
Can metallized or as-fired production substrates be tested directly?
Yes, but thin printed metallization (less than approximately 20 µm) has a negligible composite effect on the measured CTE of a 0.6–1.0 mm substrate. For heavily metallized or direct-bonded-copper products, the bare substrate should be tested if the material property itself is required, since thick copper layers shift the apparent expansion of the composite assembly toward that of copper.
Does specimen orientation relative to the tape-cast direction matter?
Polycrystalline alumina substrates are close to isotropic in-plane, with directional differences normally below 3–5%. Measuring one specimen along each in-plane direction is nevertheless good practice during material qualification, and any significant anisotropy should trigger a review of lot processing history.
How many specimens and cycles are required for a valid result?
A minimum of three specimens per material condition is recommended, each measured over at least one reported heating cycle, with a second cycle used to confirm repeatability. Results are reported as the mean with standard deviation, together with the full expansion curve and all test conditions, in line with the documentation requirements of ASTM E228 and ISO 17562.