Determination of the Softening Temperature of Semi-Insulating Silicon Carbide Substrates

The softening temperature of semi-insulating silicon carbide (SiC) substrates is a critical thermal-stability parameter that governs dimensional integrity during high-temperature epitaxial growth and device processing. This article explains how the softening and high-temperature deformation behavior of semi-insulating 4H- and 6H-SiC substrates is measured using thermomechanical analysis and refractoriness-under-load methods in accordance with standards such as ASTM E831, ISO 11359-2, and ISO 1893. Wafer manufacturers, epitaxy houses, and RF device developers can use these data to qualify material lots, define safe process windows, and diagnose thermally induced wafer deformation.

Technical Overview

Semi-insulating silicon carbide substrates are high-resistivity single-crystal wafers, most commonly of the 4H-SiC or 6H-SiC polytype, with bulk resistivity typically in the range of 10^5 to 10^9 Ω·cm or higher. Semi-insulating behavior is achieved either through deep-level compensation (for example, vanadium doping) or through high-purity crystal growth that minimizes residual nitrogen and boron impurities. These substrates form the foundation of GaN-on-SiC epitaxy for RF amplifiers, radar modules, and satellite communication front-ends, where the substrate must combine electrical isolation with outstanding high-temperature mechanical stability.

Unlike metals or glasses, silicon carbide does not exhibit a true melting point at atmospheric pressure; the covalently bonded crystal sublimes and decomposes at approximately 2700 °C. In engineering practice, the softening temperature of a semi-insulating SiC substrate is therefore defined as the temperature at which the material begins to show measurable plastic or visco-plastic deformation under a specified mechanical load, that is, the onset of load-bearing capacity loss. This behavior is evaluated through thermomechanical analysis (TMA) and refractoriness-under-load (RUL) type testing, which track dimensional change as temperature rises under a defined stress.

The parameter matters because every downstream process step, including GaN or AlN epitaxy at 1000–1650 °C, activation annealing, and high-temperature metallization, imposes thermal and mechanical loads on the substrate. If deformation begins too early, wafers warp or bow and slip dislocations can propagate into the epitaxial layer, degrading yield and long-term device reliability. A quantitative softening temperature determination therefore provides the scientific basis for setting safe process windows and for qualifying incoming substrate lots.

Testing Samples

Semi-Insulating 4H-SiC Substrates

Standard 4H polytype semi-insulating wafers from 50.8 mm (2 inch) to 150 mm (6 inch) diameter, 350–1000 μm thick, in as-sliced, lapped, or double-side polished condition. Both vanadium-doped (V:SiC) and high-purity uncompensated grades can be evaluated.

Semi-Insulating 6H-SiC Substrates

6H polytype semi-insulating substrates used in legacy RF and optoelectronic applications are tested with identical procedures, enabling direct comparison between polytypes and doping routes.

Prepared Test Coupons

Because whole wafers are rarely consumed, testing is usually performed on coupons diced or cored from representative wafers: small rectangular pieces for TMA (typically 3–10 mm across) and cylindrical or prismatic compression specimens for RUL and creep testing. All load-bearing surfaces are ground flat and parallel to tight tolerances to avoid stress concentrations, and samples are cleaned and pre-conditioned to remove moisture and machining residues.

Reference and Control Materials

Calibration verification coupons such as sapphire or alumina reference standards are run alongside samples, and conductive n-type SiC comparison pieces can be included where a reference contrast is required.

Testing Items

The standard test portfolio for semi-insulating SiC substrates includes:

• Softening point (onset temperature of probe penetration or departure from linear expansion) by thermomechanical analysis
• Characteristic temperatures at 0.5%, 1.0%, and 5.0% deformation under compressive load (refractoriness-under-load values)
• Complete load–deformation and deformation–temperature curves with softening onset determined by the tangent method
• High-temperature flexural strength (hot modulus of rupture) at selected temperatures as supporting evidence
• Compressive creep strain and steady-state creep rate at constant temperature and stress
• Coefficient of thermal expansion up to the test temperature, obtained simultaneously by dilatometry
• Post-test inspection covering dimensional change, warpage, surface oxidation check, and microscopic confirmation of the deformation mode

Testing Methods

Thermomechanical Analysis (TMA / Static Dilatometry)

In accordance with ASTM E831 and ISO 11359-2, a polished coupon is placed in a TMA or vertical dilatometer fitted with alumina or silicon carbide probes and fixtures. A small static load is applied and the specimen is heated at a controlled rate, commonly 5 °C/min, in high-purity argon or vacuum to prevent oxidation of SiC above approximately 800 °C. The softening temperature is identified as the point where the dimensional-change curve departs from the linear thermal-expansion trend, with the tangent intersection defining the onset of softening. Instrument calibration against certified reference materials such as sapphire is verified before each campaign.

Refractoriness Under Load (RUL)

Following the principles of ISO 1893 and the equivalent GB/T 5989 methodology, a cylindrical or prismatic specimen is loaded in compression at constant stress (typically 0.05–0.2 MPa) and heated at a defined rate to a maximum of 1600–1800 °C. Specimen compression is recorded continuously, and the temperatures at 0.5%, 1%, and 5% deformation are reported. For single-crystal semi-insulating SiC, which is exceptionally creep-resistant, this method establishes a conservative lower bound of the usable thermal envelope.

High-Temperature Mechanical Verification

Complementary hot modulus-of-rupture testing based on ASTM C1211 and EN 820-1 concepts, together with short-term compression creep tests at fixed temperatures, confirms that no significant stress-assisted deformation occurs within the customer's process window, typically up to 1650 °C for GaN-on-SiC epitaxy.

Atmosphere, Equipment, and Data Quality

All elevated-temperature tests are conducted in inert (argon) or vacuum atmospheres using graphite-heater or tungsten-heater furnaces, with temperatures monitored by calibrated Type C tungsten–rhenium thermocouples or calibrated pyrometry. Every report includes raw deformation–temperature curves, calibration records, the softening-onset determination method, and measurement uncertainty. Testing is performed in accredited third-party laboratories operating under ISO/IEC 17025-aligned quality systems.

Applications

• Qualification of semi-insulating SiC substrate lots for GaN-on-SiC MOCVD and HVPE epitaxial growth processes running at 1000–1650 °C
• Definition of safe thermal budgets for activation annealing, ohmic contact alloying, and other high-temperature device fabrication steps
• Comparison of substrate grades: vanadium-doped versus high-purity semi-insulating material, 4H versus 6H polytypes, and different crystal growth batches
• Root-cause investigation of wafer warp, bow, slip-line formation, and epitaxial defect generation traced to thermal deformation
• Reliability support for RF front-end modules in radar, electronic warfare, satellite communication, and 5G infrastructure applications
• Research and development of next-generation crystal growth and compensation strategies, and generation of material property data for thermal and mechanical simulation models

FAQ

Does silicon carbide actually melt or soften like glass?

No. Silicon carbide is a covalently bonded ceramic that sublimes and decomposes near 2700 °C rather than melting. The reported softening temperature is an engineering parameter: the temperature at which measurable deformation begins under a defined load, determined by thermomechanical analysis or refractoriness-under-load methods.

What softening temperature values are typical for semi-insulating SiC substrates?

Dense single-crystal SiC is exceptionally creep-resistant. Under conventional RUL loads of 0.05–0.2 MPa, measurable deformation typically does not begin until well above 1600 °C, and significant softening is generally observed only in the 1700–2000 °C range, far above epitaxial growth temperatures. Reported values depend strongly on applied stress, heating rate, atmosphere, and crystal orientation, so direct comparison requires identical test conditions.

What sample size and condition are needed?

TMA coupons of roughly 3–10 mm are sufficient, while RUL and creep tests use small cylinders or prisms diced from a representative wafer. Load-bearing surfaces must be flat, parallel, and clean. If the original wafer must be preserved, testing can be performed on witness coupons taken from the same boule or growth run.

Why is an inert atmosphere required?

Silicon carbide begins to oxidize in air at approximately 800–1000 °C, forming silica surface scales that alter surface mechanics and distort deformation measurements. Testing in high-purity argon or vacuum isolates the intrinsic softening behavior of the crystal.

Which standards govern the testing?

Primary references include ASTM E831 and ISO 11359-2 for thermomechanical analysis, ISO 1893 and GB/T 5989 for refractoriness under load, and ASTM C1211 and EN 820-1 concepts for high-temperature strength verification, with laboratory quality managed under ISO/IEC 17025 systems.

How are the results reported and used?

Each test yields a deformation–temperature curve with the softening onset temperature and characteristic deformation temperatures, together with calibration data, measurement uncertainty, and post-test inspection findings. Customers use these values to set epitaxy and annealing process windows, compare substrate suppliers, and release incoming material lots with documented thermal stability.