《Investigation on warpage characteristics for ultra- large MCM FCBGA package based on material-structure co-design and intelligent optimization》
Title “Investigation on Warpage Characteristics for Ultra-Large MCM FCBGA Package Based on Material-Structure Co-Design and Intelligent Optimization”
Author: Ping Wu, Student Member, IEEE; Zhen Liu; Mingming Yi; Yiou Qiu; Linzheng Fu; Wenhui Zhu, Senior Member, IEEE; and Liancheng Wang, Member, IEEE
Publication Date: 2025
DOI: 10.1109/TCPMT.2025.3602869
Journal: IEEE Transactions on Components, Packaging and Manufacturing Technology
Abstract: The growing demand for computational power driven by artificial intelligence (AI) and high-performance computing (HPC) has spurred significant advances in multi-chip integration and large-scale packaging approaches. As a result, multi-chip module (MCM) packaging has emerged as a prominent technological solution. However, the mismatch in thermal expansion coefficients (CTEs) between silicon chips and organic substrates can exacerbate warpage, leading to process and reliability issues such as solder bump bridging, non-wetting, and solder ball cracking. Consequently, warpage control has become a critical challenge for these types of products. This study proposes a warpage-control approach that integrates material-structure co-design with hybrid intelligent optimization. A simulation model of a 4-chip 102 mm × 102 mm flip-chip ball grid array (FCBGA) package, validated through shadow moiré experiments, achieves a prediction error of ≤7.4%. The warpage exhibits a non-uniform wave pattern influenced by chip layout; key influencing factors include chip spacing, substrate thermal expansion coefficient, ring thermal expansion coefficient, ring foot width, and ring foot thickness. A customized response surface methodology (RSM) accurately captures the nonlinear coupling effects, while a hybrid NSGA-II and PSO algorithm optimizes warpage control. Full-parameter optimization reduces warpage by 33.5% (from 138.5 μm to 92.1 μm) at 30 °C and by 88.3% (from 28.6 μm to 3.4 μm) at 245 °C. When optimizing only the ring structure, the warpage improvement is even more pronounced: at 30 °C, warpage is reduced by 50.9% (from 102.2 μm to 50.1 μm), and at 245 °C, it is reduced by 32.2% (from 166.4 μm to 113.0 μm). The proposed approach offers new insights and an effective strategy for warpage control in large-scale MCM packaging.
Conclusion: This study investigates and optimizes warpage in a 102 mm × 102 mm ultra-large four-chip flip-chip ball grid array (FCBGA) package through experimental testing and the finite element method. The accuracy and precision of the finite element analysis (FEA) model were verified using the shadow moiré technique. Based on this verification, the relationship between multi-chip layout and package warpage was systematically examined. In the warpage optimization design, first, a one-way simulation system was used to analyze the effects of package material properties and structural dimension parameters on warpage, identifying the key influencing factors. Subsequently, the response surface methodology (RSM) was employed to obtain interaction effects among design factors and develop a high-precision warpage prediction model. Finally, the non-dominated sorting genetic algorithm II (NSGA-II) and the particle swarm optimization algorithm (PSO) were utilized to achieve multi-objective optimization designs for warpage at 245 ℃ and 30 ℃. The main conclusions of this study are as follows:
1) The finite element analysis (FEA) model developed in this study exhibits high precision and accuracy, with a simulation-to-experiment error of approximately 7.4%.
2) The substrate exhibits a more complex wavy warpage than an “M” or “W” shape. There are distinct “valleys” and “peaks” at the centers of both the chip and the substrate, and asymmetric stresses are present at the four corners of the chip.
3) The effect of chip spacing on package warpage differs depending on whether or not a ring is present. Without a ring, the greater the chip spacing, the smaller the warpage, reaching its minimum at a spacing of 60 mm. With a ring, as the spacing increases, package warpage first decreases and then increases again, with the minimum warpage occurring at a spacing of 36–40 mm.
4) The unidirectional analysis results indicate that, taking a warp variation of ±20% as the threshold, five parameters—ring ring thickness, ring ring foot width, chip pitch, substrate coefficient of thermal expansion (CTE), and ring ring CTE—are key factors influencing package warping.
5) The customized response surface model (RSM) indicates that the coefficient of thermal expansion (CTE) of the ring and the CTE of the substrate have the greatest impact on warping, while previously there was a very strong interaction between the thickness of the ring and both the ring’s CTE and the substrate’s CTE.
6) Full-parameter optimization of encapsulation materials and dimensions has achieved optimal warp control: at 30℃, the warp amount was reduced by 33.5%, dropping to 138.5 μm; at 245℃, the warp amount was significantly reduced by 88.3%, down to just 28.6 μm. Optimization of key parameters related to the ring dimension resulted in a 50.9% reduction in warp at 30℃ (down to 102.2 μm) and a 32.2% reduction at 245℃ (down to 166.4 μm).
Image display:
About the Author:
The innovation of this study lies in the following:
(1) This paper proposes a warpage control method that combines material–structure co-design with hybrid intelligent optimization, providing a new approach for warpage suppression in large-scale multi-chip module (MCM) packaging.
(2) A simulation model of a 4-chip FCBGA package measuring 102 mm × 102 mm was constructed and validated using the shadow moiré experiment, with a prediction error of ≤7.4%.
(3) Accurately modeling nonlinear coupling effects based on a customized response surface methodology (RSM), and combining the NSGA-II and PSO hybrid algorithm to achieve full-parameter optimization, significantly reduces package warpage at 30 ℃ and 245 ℃.
Next Page