Micro-/Nano-Technology for Future - Physics-based modeling of wafer-to-wafer bonding

Vertical stacking of integrated circuits (IC), known as the 3-D integration, has emerged as a breakthrough solution to overcome the limitations of traditional continuous scaling of individual components. 3-D integration enables superior performances due to shorter interconnections, reduced system sizes and improved system heterogeneity. Circuit layers can be fabricated separately, enabling the combination of incompatible manufacturing processes into a single 3D IC (e.g., memory and logic). Among several 3-D integration schemes, wafer-to-wafer hybrid bonding is distinguished as the key technology for achieving high-density interconnections, required for fine-partitioning 3-D system-on-chip applications. In this bonding technology, both wafers are finished with a dielectric layer with embedded Cu pads. The wafer pair is then accurately aligned with a small vertical distance between them, in the order of a few tens of nanometres. The top wafer is pushed through a localized area at its centre to establish the initial contact between wafers. The bonding is then propagated in a wave-like pattern due to the interaction forces between opposing wafers.

A major challenge associated with this process is acquiring a sufficiently low alignment error between the bonded wafers, for which the current and future industry demands are extremely stringent. It is not a trivial task to meet these demands since the final alignment is affected by various parameters, some of which can be listed as: wafer properties (shape, residual stress, mechanical properties), the dielectric material choice, extrusion/recession of Cu pads, the chuck design holding the wafers, the bonding recipe, initial distance between wafers, the magnitude of the point contact force, adhesion forces between dielectrics, air viscosity and gravitational effects. Considering the presence of numerous parameters and their potential interactions, optimization attempts that only focus on experimental approaches will have limited capabilities due to their slow-paced and expensive nature. Therefore, it is necessary to develop supplementary methodologies based on simulation techniques.

The purpose of this PhD topic is to deepen the understanding on both the physics of bonding propagation and the impact of mechanical boundary conditions using modelling and simulation techniques. To achieve this, a physics-based mechanical modelling environment (based on the finite-element method) will be developed to study the bonding phenomena. Modelling activities will first be initiated using simplified 2-D models, which will later be extended to full 3-D wafer bonding simulations. Where possible, bond wave metrology data along with experimental information will be provided to calibrate the simulations. The learnings obtained through simulations will be utilized to guide the industry and academia in terms of wafer preparation and bonding configurations.