“…The two models also define the strategic differences taken to identify a root cause explanation for dielectric breakdown: for the 1/E-model, a fluence model leading to impact ionization damage [359]; 0 ULK LK LK Gate oxide SiCOH, 90 nm SiO 2 _PECVD [Noguchi, 2001] SiOF [Ogawa, 2003] SiCOH [Ogawa, 2003] SiCOH [Aubel, 2011;unpublished] pMSQ [Ogawa, 2003] SiCOH SiCN pSiCOH, 50 nm pSiCOH, 90 nm pSiCOH, MOS Figure 11.14 A dielectric breakdown comparison is shown for different dielectrics. There are also a couple of interesting observations that demonstrate the importance of process/integration quality: (1) for similar LKs, a more recent process shows superior TDDB performance compared to earlier results; (2) in comparatively integrated structures, both LK with SiC x N y capping dielectric and SiC x N y -only structures show very similar TDDB, indicating that capping dielectric breakdown is also important; (3) TDDB in a MOS structure with ULK only shows better nominal TDDB (it needs an area-scaling correction, however) than integrated ULK and a tighter space integrated-ULK shows a worse performance than a wider space ULK (see color Plate 13) for the E-model, a field-driven model, where dipole-field coupling reduces the activation energy for thermally induced bond breakage [34]. ; however, a clear trend to lower breakdown performance with a decreasing k value is evident.…”