Determination of the Planar Coefficient of Thermal Expansion (CTE) of Multilayer Packaging Substrate

The planar (in-plane) coefficient of thermal expansion (CTE) of multilayer packaging substrates is a critical thermo-mechanical parameter that directly governs package warpage, solder joint reliability, and long-term performance of advanced semiconductor assemblies. This article provides a comprehensive technical overview of in-plane CTE determination, covering test specimen preparation, measurement methods in accordance with recognized standards, typical parameter ranges, and practical engineering applications. It is intended for packaging engineers, substrate designers, materials specialists, and quality professionals involved in the qualification and incoming inspection of IC substrates.

Technical Overview

Multilayer packaging substrates are the structural and electrical foundation of modern semiconductor packages such as flip-chip ball grid arrays (FC-BGA), chip-scale packages (CSP), wafer-level packages, and system-in-package (SiP) assemblies. These substrates typically consist of a rigid or low-modulus core laminate, alternating build-up dielectric layers, patterned copper circuitry, plated through-holes and microvias, and solder mask layers. Because each of these material systems has a different thermal expansion behavior, the effective dimensional stability of the finished substrate in the X and Y (planar) directions becomes a decisive design parameter.

The planar coefficient of thermal expansion, commonly expressed in ppm/°C (10⁻⁶/°C), quantifies the fractional change in length of the substrate per degree of temperature change. It is normally evaluated as two distinct values: α₁, the average CTE below the glass transition temperature (Tg), and α₂, the average CTE above Tg. Since the silicon die expands at approximately 2.6 ppm/°C while the mating printed circuit board typically expands at 14–17 ppm/°C, the substrate must be engineered to bridge this mismatch. Excessive planar CTE mismatch between the die, substrate, and board drives solder joint fatigue during thermal cycling, package warpage, underfill delamination, and low-k dielectric cracking.

Reliable determination of the in-plane CTE therefore supports material selection, warpage simulation, process window definition, and reliability qualification. Key characteristics of the measurement include:

• Anisotropy between the X (warp) and Y (weft) directions caused by glass fabric reinforcement and circuit copper distribution.
• Distinct α₁ and α₂ regimes separated by the glass transition region of the organic dielectric materials.
• Dependence on copper coverage, residual stress from lamination, and moisture content of the laminate.
• Typical target values of approximately 8–17 ppm/°C for organic substrate cores below Tg, with low-CTE core designs achieving 6–10 ppm/°C for large-body, high-density packages.

Testing Samples

Bare Core Laminates and Copper-Clad Panels

Bare laminate sheets and copper-clad panels are tested during material qualification to characterize the intrinsic expansion behavior of the composite system. Copper foil may be fully etched away when the dielectric contribution alone is required, or retained when the effective laminate CTE is of interest. Specimens are cut in both the machine (X) and transverse (Y) directions to quantify fabric-weave-induced anisotropy.

Finished Multilayer Build-Up Substrates

Production substrates with completed build-up layers, patterned circuitry, and solder mask can be measured to determine the effective CTE of the finished product. The presence of dense copper structures, stiffener rings, and asymmetric layer stacks measurably shifts the effective expansion coefficient relative to the bare laminate, so finished-substrate data are essential for accurate warpage modeling of the assembled package.

Specimen Preparation and Conditioning

Proper specimen preparation is essential for repeatable results. Typical practice includes:

• Cutting strip specimens of suitable geometry along both X and Y axes, with clean, parallel edges and no edge chipping or delamination.
• Removing moisture by a pre-test bake (commonly 105–125 °C for 1–2 hours) followed by cooling in a desiccator, or conditioning at 23 ± 2 °C and 50 ± 5 % relative humidity where the as-received state must be represented.
• Testing a minimum of three to five specimens per direction to characterize variability.
• Marking specimen orientation and original gauge length, and verifying thickness uniformity prior to loading into the instrument.

Testing Items

A complete planar CTE characterization of multilayer packaging substrates typically includes the following measured and derived quantities:

• Average in-plane CTE below Tg (α₁) in the X direction.
• Average in-plane CTE below Tg (α₁) in the Y direction.
• Average in-plane CTE above Tg (α₂) in the X and Y directions, where applicable.
• Glass transition temperature (Tg) determined from the inflection in the dimension–temperature curve obtained by TMA.
• X/Y anisotropy ratio, used to evaluate fabric balance and layup symmetry.
• Full dimensional change versus temperature curves over the specified test range, reported as raw data for simulation input.

Testing Methods

Thermomechanical Analysis (TMA)

Thermomechanical analysis is the most widely used method for in-plane CTE determination on substrate materials. Strip specimens are clamped in a film/extension fixture or supported on a flat quartz platform with a low-force probe, and length change is tracked while the furnace applies a controlled temperature ramp. Typical test conditions are a heating rate of 3–5 °C/min under nitrogen purge (20–50 mL/min), over a range such as 25 °C to 150 °C for α₁ or extended to 250–260 °C when α₂ and Tg are also required. A two-run (heat–cool–reheat) program is recommended, with the first heating cycle used to relieve residual thermal and moisture history and the second cycle used for CTE calculation. Applicable standards include ASTM E831, ISO 11359-2, and the relevant IPC-TM-650 test methods for in-plane expansion of printed board materials.

Push-Rod Dilatometry

Push-rod dilatometry in accordance with ASTM E228 employs longer specimens and a precision dilatometer to measure average linear expansion with high accuracy. This approach is well suited to bare core laminates and reference material verification, and it provides an independent cross-check of TMA results, particularly for low-expansion cores where small dimensional changes must be resolved.

Non-Contact Optical Methods

Optical techniques such as digital image correlation, laser displacement sensing, and interferometric dilatometry measure planar expansion without contact forces on the specimen. These methods are valuable for very thin build-up films, large panels where full-field strain mapping is needed, and studies that couple CTE behavior with real-time warpage observation. Calibration against certified reference materials such as fused silica or copper standards is performed before each test campaign.

Calculation and Data Evaluation

The average CTE over a temperature interval is calculated from the slope of the dimensional change curve using the relationship α = (1/L₀) × (ΔL/ΔT), where L₀ is the original gauge length at the reference temperature, ΔL is the length change, and ΔT is the temperature interval. Linear regression is performed over clearly defined segments below and above the transition region, and results are reported with the test direction, temperature interval, heating rate, specimen count, mean value, and standard deviation. Measurement uncertainty of approximately ±0.5–1 ppm/°C is typical for well-controlled TMA testing.

Applications

Planar CTE data for multilayer packaging substrates are used throughout the product lifecycle:

• Material qualification and lot acceptance of core laminates, prepregs, and build-up films against specification limits referenced in material standards such as the IPC-4101 slash-sheet system, the GB/T 472 series for copper-clad laminates, and JIS C 6481.
• Warpage prediction and finite-element simulation of die attach, molding, reflow, and operation, where accurate α₁ and α₂ inputs are prerequisites for reliable models.
• Solder joint reliability assessment, since cyclic thermal expansion mismatch directly controls inelastic strain accumulation in first- and second-level interconnects during temperature cycling.
• Stack-up and design optimization, including selection of low-CTE cores, balanced glass styles, and symmetric copper distribution to minimize X/Y anisotropy and residual-stress-induced distortion.
• Failure analysis support, where measured CTE deviations from baseline values help identify material substitution, incomplete cure, or process-induced degradation.

FAQ

Why is the in-plane CTE different in the X and Y directions?

Glass fabric reinforcement, the woven structure of the reinforcement, and non-uniform copper distribution cause directional differences. The warp and weft directions of the fabric exhibit different stiffness and expansion behavior, and unbalanced circuit copper further modifies the effective CTE, so both directions must be measured independently.

What is the difference between planar (X/Y) and out-of-plane (Z) CTE?

Planar CTE is constrained by the glass fabric and copper layers and is typically 8–17 ppm/°C below Tg for organic substrates. The Z-axis CTE is resin-dominated and much larger, commonly 40–70 ppm/°C below Tg and several hundred ppm/°C above Tg, which is why plated via integrity is assessed using Z-axis expansion data while warpage and solder reliability rely on planar values.

What temperature range should be used for testing?

The range must reflect the intended application. For α₁ determination, a range such as 30 °C to 150 °C or 50 °C to 150 °C is common. When Tg and α₂ are also required, testing is extended toward 250–260 °C, remaining below the decomposition temperature of the organic materials.

Why is a double-heating procedure recommended?

The first heating cycle removes residual stresses from lamination and processing and drives off absorbed moisture, which otherwise distorts the expansion curve. The second heating cycle provides a thermally conditioned, reproducible measurement that reflects the intrinsic material behavior.

What acceptance criteria are typically applied?

Acceptance limits are product-specific and are usually agreed between the substrate supplier and the packaging customer. Common qualification targets for FC-BGA-class substrates specify α₁ in the range of approximately 10–17 ppm/°C, with tighter windows of 6–10 ppm/°C for large-body or low-warpage designs, verified by an accredited third-party laboratory or the supplier's certified laboratory using standardized test methods.