Delft University of Technology, imec Introduce 3D-COSTAR

Delft University of Technology (TU Delft) and nanoelectronics research center imec presented 3D-COSTAR, a new test flow cost modeling tool for 2.5/3D stacked integrated circuits (ICs). 3D-COSTAR aims to optimize the test flow of 3D stacked ICs (SICs), taking into account the yields and costs of design, manufacturing, packaging, test and logistics.

Due to its many high-precision steps, semiconductor manufacturing is defect-prone. Consequently, every IC needs to undergo electrical tests to weed out defective parts and guarantee outgoing product quality to the customer. For TSV-based 2.5D- and 3D-SICs that typically contain complex die designs in advanced technology nodes, testing is even more critical.

In addition, there are many possible test moment in their manufacturing flow: pre-bond (before stacking), mid-bond (on a partial stack), post-bond (on a completed stack), and final testing (on a packaged device). Although testing is expensive, filtering out the bad components in an early stage is critical to save costs later on in the production process.

“There is not a ‘one-size-fits-all’ test flow that covers all stacked-die products. The test flow needs to be optimized based on yield and cost parameters of an individual product and that is a complex optimization problem,” said Dr. Said Hamdioui, associate professor at TU Delft. “And different test flows, executed after manufacturing, may require different design-for-test features, which need to be incorporated in the various dies during their early design stages.”

“3D-COSTAR has proven to be a crucial tool to analyze the many complex trade-offs in 3D test flows, in terms of both cost and DPPM,” said Erik Jan Marinissen, principal scientist at imec. “Among others, we have used 3D-COSTAR to determine when pre-bond testing of the interposer in 2.5D-SICs pays off and what its maximum-allowed test cost can be. In some cases, the overall stack cost reduction amounts to 40%, showcasing that upstream testing can help avoid downstream costs. The tool also demonstrated under which circumstances mid-bond testing (on partially-completed stacks) can be avoided without compromising a high stack yield.”