Tomohisa Konno

Stress and Slurry Chemistry Effects on CMP Damage of Ultra-Low-k Dielectrics

The yield and reliability of the next generation Cu/low-k interconnects depends critically on the control of damage in the form of crack growth in ultra-low-k (ULK) dielectrics. The ULK dielectrics are mechanically fragile and susceptible to environmentally accelerated cracking in reactive aqueous environments. Nevertheless, during chemical mechanical planarization (CMP) and post-CMP cleaning these extremely brittle thin-film structures are subjected to mechanical loads in the presence of harsh aqueous solutions. We demonstrate that both process stress and chemistry are crucial for the rate of damage evolution during CMP. Small changes in CMP slurry chemistry and surfactant additions can have a dramatic effect on damage processes and associated CMP yield. These are crucial aspects for the reliable integration of ultra-low-k materials at next technology nodes.

Quantitative Roadmap for Optimizing CMP of Ultra-Low-k Dielectrics

Optimization of chemical mechanical planarization (CMP) is crucial for reliable integration of ultra-low-k (ULK) materials for the next technology nodes. We present a quantitative road map for optimized CMP of ULK in which critical factors such as defect evolution, CMP damage, diffusion, effective k increase, and CMP removal rates are for the first time all correlated. Additions of common nonionic surfactants are shown to have dramatic effects on defect evolution. In addition, the surfactant melts and their aqueous solutions can readily diffuse in even strongly hydrophobic soluion ca redilydifusein venstroglyhydophbic nanoporous ULK films affecting k values. Finally, the same solution chemistries and surfactants are shown to have an important effect on CMP removal rates. Measured defect growth rates, solution diffusion coefficients, and removal rates varied markedly depending on molecular weight, hydrophilic-lipophilic balance (HLB), and molecular structure of the surfactants. A roadmap is provided in which all of these variables are quantitatively correlated.