Effect of Heterogeneity of Laminated Structure in FR-4 on Thermal Fatigue Life of Interstitial Via Hole in Printed Wiring Board (First Report, Investigation of Variation Factors of Life Using Thermal Cycling Test and Finite Element Analysis)

Kunihiro Takenaka, Qiang YuJournal of The Japan Institute of Electronics Packaging 2013 · DOI

TL;DR

This paper investigates why identical-looking Interstitial Via Holes (IVHs) in printed wiring boards fail at different rates during thermal cycling. By moving beyond the standard assumption that the FR-4 substrate is a uniform material and instead modeling the detailed "weave" of the glass cloth, the researchers proved that the local microscopic arrangement of glass fibers and resin is a primary driver of fatigue life variation.

What Problem Does This Paper Address?

As electronics shrink, the reliability of the tiny vertical connections between board layers—known as Interstitial Via Holes (IVHs)—becomes critical. Engineers typically use Finite Element Analysis (FEA) to predict how long these copper connections will last under thermal stress.

However, there is a persistent gap: experimental tests show a wide "scatter" in failure times that standard simulations can't explain. The culprit is the FR-4 substrate itself. While usually treated as a homogeneous block in simulations, FR-4 is actually a complex composite of woven glass fabric and epoxy resin. This paper addresses the need for a practical way to account for this microscopic heterogeneity to better understand and predict via failure.

Methods

The authors combined physical experimentation with high-fidelity simulation:

  • Thermal Cycling Tests: Samples were subjected to temperatures ranging from -60°C to 120°C for up to 2,700 cycles. Electrical resistance was monitored to detect failures (defined as a 10% increase in resistance).
  • Detailed FEA Modeling: Unlike standard models, the researchers used ANSYS to create a 3D model that explicitly included the plain-weave structure of the glass cloth (specifically the "7628" type). They modeled individual fiber bundles as anisotropic materials embedded in resin.
  • Cross-Sectional Analysis: After testing, the boards were sliced and polished. The researchers measured the exact "shift" between the two layers of glass cloth and the precise position of the IVH relative to the weave to feed these real-world coordinates back into their simulation.
  • Strain Extraction: They focused on "Average Equivalent Plastic Strain Range" ($\Delta\epsilon_p$) at the via barrel, using the Coffin-Manson law to relate this strain to fatigue life.

Datasets

This is an experimental mechanics paper rather than a machine learning study, so it does not use public digital datasets. Instead, the authors generated their own experimental data:

  • Physical Samples: 8 PWB samples, each containing 6 IVHs connected in series (totaling 48 vias).
  • Material Characterization: They measured the dimensions of glass fiber bundles (width, thickness, pitch, and wavelength) across 24 different cross-sections to establish the average geometry for their "heterogeneous" model.

Key Insights and Ideas

  • The "Hidden" Geometry Matters: The researchers found that even if two vias have the same copper plating thickness, their lifespans can differ significantly based on where they sit relative to the glass weave.
  • Lamination Shift: FR-4 is made of multiple layers of glass cloth. The relative horizontal shift between these layers changes how the resin is "constrained." If the via is in a region where the resin can expand freely, it puts more stress on the copper.
  • Drilling Position: The position of the hole relative to the "knots" or "gaps" in the glass weave significantly changes the local mechanical environment.
  • Validation through Crack Patterns: The detailed FEA model successfully predicted not just when a via would fail, but where the crack would start. The simulation showed high strain areas that perfectly matched the curved crack paths observed in the physical cross-sections.

Key Results

  • Plating vs. Structure: While copper plating thickness is a major factor (thicker is better), it only explained part of the lifespan variation.
  • Failure Order Prediction: The detailed FEA model correctly predicted the order of failure for different samples (Sample A failing before Sample B, etc.) by accounting for the specific glass cloth arrangement.
  • Quantifiable Impact: The study showed that a 9% difference in local strain—caused purely by the via's position relative to the glass weave—could result in a 15% difference in predicted thermal fatigue life.
  • Correlation: A strong correlation was found between the "Average Equivalent Plastic Strain" calculated by the detailed model and the actual cycles to failure observed in the lab.

Limitations and Future Work

  • Scope of Materials: The study focused specifically on FR-4 with 7628-type glass cloth. Other weave patterns (like "spread" glass) might behave differently.
  • Manual Extraction: The process of cross-sectioning and manually measuring the weave shift to build the FEA model is labor-intensive.
  • Future Steps: This is "Report 1." The authors suggest that future work will focus on developing more automated or statistical methods to account for these variations during the design phase, rather than just post-failure analysis. They also aim to explore how to control these "variation factors" during the manufacturing process to improve overall board reliability.

Summary updated

Abstract

PWBにおけるIVHの熱疲労寿命に及ぼすFR-4の積層構造の不均質性の影響を検討するために,熱疲労寿命に影響する要因を検討する実用的な手法を示し,その手法に基づいて冷熱サイクル試験と有限要素法解析を用いて熱疲労寿命の変動要因を検討した。解析モデルにPWBの平織りのガラスクロスを詳細にモデル化したものを使用することで,解析結果が試験結果のき裂発生状況や故障順序と一致することを確認した。その結果,積層されたガラスクロスの積層状態の違いと,ある積層状態のガラスクロスに対するIVHの形成位置の違いが,IVHの熱疲労寿命に影響していることを示唆した。これらの違いはIVHの熱疲労寿命の変動要因であり,その要因が組み合わさってIVHの熱疲労寿命に影響すると考えられる。
BibTeX
@article{takenaka2013effect,
  author = {Kunihiro Takenaka and Qiang Yu},
  title = {{Effect of Heterogeneity of Laminated Structure in FR-4 on Thermal Fatigue Life of Interstitial Via Hole in Printed Wiring Board (First Report, Investigation of Variation Factors of Life Using Thermal Cycling Test and Finite Element Analysis)}},
  journal = {Journal of The Japan Institute of Electronics Packaging},
  year = {2013},
  doi = {10.5104/jiep.16.106},
}
Figure