Predicting the Thermomechanical and Adhesive Properties of a Layered Polyimide Packaging Material Using Molecular Simulation

The increasing demand for sophisticated electronic devices has placed challenging requirements on integrated circuits and the advanced packaging materials that encapsulate and protect them. To meet these engineering challenges, computational materials science offers powerful predictive tools, particularly molecular dynamics (MD) and quantum mechanical (QM) simulations. These methods are routinely employed to accurately compute the properties of materials directly from fundamental principles of chemistry and physics. MD and QM have been well-validated for polymer materials. They have proven utility for the design of new materials, formulation design, and characterization of phenomena that can be challenging and expensive to assess experimentally. Using these tools, we developed a molecular-scale model of a photo-imagable dielectric (PID) polyimide. This model predicts how the polyimide’s thermomechanical properties evolve during fabrication and characterizes its performance under reliability testing conditions, such as thermal cycling and mechanical stress. Additionally, it simulates the interaction of the polyimide with metallized surfaces, as seen in applications like redistribution layers (RDLs).

Over the course of fabrication, a packaging polymer may transform through chemical processes like crosslinking and physical processes like solvent evaporation. The study involved the creation of detailed molecular models that accurately represented the material starting from its precursor state, progressing through an intermediate crosslinked state, and finally reaching its fully cured state. This step-wise approach is critical for capturing how properties mature throughout the manufacturing process. Using dedicated workflows within Schrödinger’s Materials Science Suite, the material was assessed for a range of properties. The Elastic Constants workflow was used to determine the bulk, shear, and Young’s moduli, which define the material’s mechanical stiffness. The Thermophysical Properties workflow was used to compute coefficients of thermal expansion (CTEs), which govern its dimensional stability under temperature changes, and its glass transition temperature, a key indicator of its thermal stability. Finally, the Amorphous Dielectric Properties workflow was used to compute the dielectric constant and dielectric loss, essential for electrical insulation and signal integrity. In the final product state, the dependence of these properties on temperature was assessed, revealing their stability outside of typical environmental conditions. Since experimental reliability testing is often time-consuming and costly, this computational approach offers an efficient alternative.

The interface between packaging polymers and inorganic substrates, such as metallization layers, is critical to device performance. This study developed interfacial models to evaluate the adhesion energy of polymers with Cu and CuO surfaces, which are typical metallization materials. By applying strain to these interfaces, we quantified their strength and identified atomic-level failure modes. This insight is vital for preventing delamination and maintaining the structural integrity and electrical performance of the packaging system.

This work showcases a comprehensive computational modeling approach to characterize a polyimide packaging material from its molecular origins to its final application. By simulating the material’s evolution through fabrication and its critical interactions with metallic interfaces, the model successfully predicts a full suite of thermomechanical, electrical, and adhesive properties. The strong correlation of these predictions with available experimental data serves to validate the model’s accuracy and utility. This predictive framework demonstrates a powerful and efficient methodology for designing and validating new materials, significantly accelerating the development cycle for advanced electronic packaging.