Thermal expansion characterization of gallium nitride (GaN) layers on alumina substrates is essential for predicting epitaxial stress, wafer bow, cracking risk, and long-term reliability in advanced optoelectronic and power semiconductor devices. This article describes the professional determination of the coefficient of thermal expansion (CTE) for GaN-on-alumina material systems, covering sample requirements, test items, and measurement methods such as high-temperature X-ray diffraction, dilatometry, thermomechanical analysis, and wafer curvature techniques. All measurements described can be performed by accredited third-party laboratories in accordance with ISO, IEC, and ASTM standards.
Technical Overview
Gallium nitride (GaN) device layers are widely grown by heteroepitaxy on alumina (sapphire, α-Al₂O₃) substrates for LEDs, laser diodes, RF amplifiers, and power transistors. Because GaN and alumina possess different coefficients of thermal expansion (CTE), significant biaxial stress develops in the GaN film during cooling from the growth temperature of approximately 1000–1100 °C. This stress causes wafer bow, crack initiation, dislocation generation, and eventual delamination, all of which directly affect yield and device lifetime. Accurate determination of the thermal expansion characteristics of both the GaN layer and the alumina substrate is therefore a fundamental task in epitaxial process development, packaging design, and reliability engineering.
Representative room-temperature values for the relevant materials include GaN with an in-plane (a-axis) CTE of approximately 5.6 × 10⁻⁶ K⁻¹ and a c-axis CTE of approximately 3.2 × 10⁻⁶ K⁻¹, and sapphire with an in-plane CTE of approximately 5.0–5.3 × 10⁻⁶ K⁻¹ and a c-axis CTE of approximately 6.0–6.7 × 10⁻⁶ K⁻¹, with all values increasing measurably at elevated temperature. Because both materials are hexagonal or trigonal and therefore thermally anisotropic, a complete characterization must distinguish in-plane and out-of-plane expansion behavior over the full operating temperature range.
- Quantification of CTE mismatch between the GaN film and the alumina substrate
- Temperature-dependent lattice parameter evolution of the epitaxial layer
- Thermal strain, thermal hysteresis, and residual stress analysis
- Anisotropic expansion behavior in the a-axis and c-axis directions
- Support for die-attach, laminate, and package-level CTE matching studies
Testing Samples
GaN Epitaxial Films on Sapphire Wafers
Standard samples include 2-inch, 4-inch, or 6-inch c-plane sapphire wafers carrying GaN epitaxial layers with typical thicknesses of 2–10 μm (including undoped buffers, n-type layers, and active heterostructures). Both as-grown wafers and processed device wafers can be evaluated. The wafer thickness, bow, and warp are documented before testing so that stress-induced geometry changes can be correlated with expansion data.
Alumina Ceramic Substrates
For thin-film and thick-film hybrid circuits, 96% or 99.6% alumina ceramic substrates in as-fired, polished, or metallized condition are tested. Ceramic samples are evaluated in the as-received state and, where relevant, after simulated process exposures such as firing, plating, or annealing, since processing history can influence apparent expansion behavior.
Device Wafers and Packaged Components
Diced die, singulated coupons, and packaged GaN-on-alumina components can be assessed for package-level CTE verification. For packaged parts, the measurement focuses on the effective expansion of the assembled stack, which supports die-attach material selection and solder joint reliability modeling.
Sample Preparation Requirements
- Dilatometry specimens: parallel-faced strips or rods, typically 3–25 mm in length depending on the instrument, with ground or lapped end faces and a length parallelism better than ±10 μm
- TMA specimens: flat coupons of 3–10 mm in size with parallel top and bottom surfaces
- HRXRD specimens: wafer pieces of at least 10 × 10 mm with the epi-surface free of particles and protective coatings removed where possible
- Curvature specimens: full wafers or quarter wafers with reflective front surface for laser scanning measurement
- All samples must be conditioned in a dry environment and weighed before high-temperature runs to exclude moisture effects
Testing Items
The scope of a complete thermal expansion characterization program for GaN-on-alumina material systems typically includes the following measurands, each reported as a function of temperature with a full uncertainty budget:
- Mean (secant) linear CTE over defined intervals, for example 25–200 °C, 25–300 °C, and 25–500 °C
- Instantaneous (tangent) CTE as a continuous function of temperature
- In-plane (a-axis) and out-of-plane (c-axis) CTE of GaN and sapphire, quantifying thermal anisotropy
- Lattice parameters a and c of the GaN film versus temperature, from which film CTE is derived
- Thermal strain–temperature curves and thermal hysteresis between heating and cooling cycles
- Biaxial film stress and wafer curvature versus temperature, calculated using the Stoney relationship
- Residual stress in the GaN layer after growth and after thermal cycling
- Effective CTE of multilayer stacks, metallizations, and packaged assemblies
- Dimensional stability and permanent length change (creep or relaxation indicators) after repeated cycling
Testing Methods
High-Temperature High-Resolution X-Ray Diffraction (HRXRD)
HRXRD with a temperature-controlled stage is the primary method for determining the CTE of the thin GaN film itself, which cannot be measured directly by bulk dilatometry. Symmetric (0002) and asymmetric (10-15) or (20-24) reflections are recorded at set points between room temperature and typically 600–800 °C under inert atmosphere. The refined lattice parameters a(T) and c(T) yield the film CTE values αa = (1/a)(da/dT) and αc = (1/c)(dc/dT). The method simultaneously provides information on strain relaxation and mosaicity changes with temperature.
Push-Rod Dilatometry (ASTM E228)
A horizontal or vertical push-rod dilatometer measures the bulk expansion of the alumina substrate and of sapphire reference pieces over a broad range, typically from room temperature to 1000 °C. The sample is heated at a controlled rate of 3–5 °C min⁻¹ in a flowing inert gas, and the length change ΔL/L₀ is recorded continuously. Calibration is verified against certified reference materials such as fused silica and single-crystal sapphire. Typical repeatability for a well-prepared specimen is better than ±0.1 × 10⁻⁶ K⁻¹.
Thermomechanical Analysis (ASTM E831 / ISO 11359-2)
TMA with a flat-ended quartz or alumina probe is used for smaller coupons and for substrate thicknesses below the practical limit of dilatometry. Tests are commonly run from −50 °C to 300 °C at 5 °C min⁻¹, matching the operating window of most electronic assemblies. TMA is also suitable for detecting softening or relaxation events in attached polymer, solder, or bonding layers within a multilayer GaN-on-alumina assembly.
Laser Interferometry (ASTM E289)
Fizeau or Michelson interferometric dilatometry
provides the highest precision, with displacement resolution in the nanometer range and CTE resolution approaching 10⁻⁸ K⁻¹. It is applied when small differential expansions between film and substrate must be resolved or when low-expansion comparative studies between candidate substrates (sapphire, AlN, SiC) are required.
Wafer Curvature and Stress Measurement
Laser-scanning curvature measurement combined with the Stoney equation maps the stress–temperature trajectory of the GaN film on the alumina substrate. By cycling the wafer between room temperature and the process or operating extremes (for example −55 °C to +300 °C), engineers can identify stress reversals, plastic relaxation thresholds, and hysteresis that signal reliability risk. This measurement directly links the fundamental CTE mismatch to the macroscopic bow behavior observed in production.
Temperature-Dependent Raman Spectroscopy
The position of the GaN E₂(high) phonon mode, with a biaxial stress coefficient of approximately 2.9 cm⁻¹ GPa⁻¹, is tracked with a heated microscope stage. Raman mapping provides spatially resolved stress data that complements the curvature method and reveals stress non-uniformity near wafer edges, pattern boundaries, and known crack initiation sites.
Applications
Measured thermal expansion data for GaN-on-alumina systems directly support engineering decisions across the entire product lifecycle:
- LED and laser diode epitaxy: optimization of buffer layers, growth temperature ramps, and cooldown profiles to keep tensile stress below the cracking threshold
- GaN power HEMTs and RF MMICs on sapphire: wafer bow control for lithography depth of focus and for backside processing
- Packaging and thermal management: selection of die-attach solders, sintered silver, and substrates whose effective CTE lies between that of the GaN-on-alumina die and the heat spreader, minimizing joint shear strain
- Reliability qualification: definition of test conditions for thermal cycling in accordance with IEC 60068-2-14 and JEDEC JESD22-A104, and inputs for solder fatigue and delamination lifetime models
- Failure analysis: correlation of measured CTE deviation, stress relaxation, or hysteresis with observed cracking, peeling, or light-output degradation
- Material comparison studies: benchmarking sapphire against alternative substrates such as silicon, silicon carbide, and aluminum nitride
FAQ
Why is the CTE mismatch between GaN and alumina so critical?
When the GaN film and sapphire substrate expand and contract by different amounts, the film experiences biaxial stress on every thermal excursion. During cooldown from growth, the differential contraction typically leaves the GaN layer in tension, which can nucleate cracks; during device operation, cyclic mismatch stress drives fatigue of die-attach joints and interfaces. Quantifying the mismatch allows designers to keep these stresses within safe limits.
What temperature range can be covered?
Standard evaluations cover −55 °C to +300 °C, which brackets most electronic operating and qualification conditions. HRXRD film measurements typically extend to 600–800 °C, and bulk dilatometry of the alumina substrate can extend to 1000 °C or higher when process-relevant data are required.
How is the CTE of a thin GaN film measured if it is only a few micrometers thick?
Thin-film CTE cannot be obtained by contact dilatometry. Instead, high-temperature HRXRD tracks the lattice parameters of the film as a function of temperature, and the film CTE is derived from the relative lattice change. Raman spectroscopy and wafer curvature analysis then convert the expansion data into film stress values.
What accuracy and repeatability can be expected?
For well-prepared specimens, push-rod dilatometry achieves repeatability of about ±0.1 × 10⁻⁶ K⁻¹ or 2–5% of the measured CTE, TMA achieves approximately ±0.05–0.1 × 10⁻⁶ K⁻¹, and laser interferometry offers the highest precision. Reported results always include calibration records against certified reference materials, heating rate, atmosphere, and an uncertainty statement.
Which standards govern these measurements?
Bulk expansion testing follows ASTM E228 (dilatometry), ASTM E831 and ISO 11359-2 (TMA), and ASTM E289 (interferometry). Device-level verification is supported by IEC 60068-2 environmental test standards and JEDEC JESD22 qualification methods. Film-level XRD and Raman methods follow documented laboratory procedures validated through interlaboratory comparison.
How should samples be submitted?
Clients should provide wafers or coupons with documented orientation (c-plane specification), layer structure and thickness, growth or processing history, and the target temperature program. Recommended practice is to submit at least three replicate specimens per condition so that statistical scatter, hysteresis, and cycle-to-cycle repeatability can be reported with confidence.