No abstract

No abstract

As the IC product scribe line of logic 90nm (L90) technology shrinks from 80µm to 62µm, the wafer quality (W.Q.), will become weak and less distinguishable during the subsequent ASML scanner stepper's photo mask aligning. Many wafers having photo mask aligning errors will eventually lead to wafer scrapping. In order to improve the photo alignment signal (W.Q.) acquired from the relatively smaller 62um scribe-line's alignment mark while proceeding with the VIA layer photo aligning directly to its previous metal layer, it is found that removing the TiN hard mask (H.M.) just above the previous inter-metal dielectric (IMD) and alignment mark area can help the deep ultra-violet (DUV) 193nm wavelength ASML scanner stepper successfully acquires a better alignment signal and alignment accuracy (A.A.).However, due to copper (Cu) residues and CMP dishing after metal copper CMP, it has been found that both large area "half size open" and "full size open" approaches for TiN removing in the scribe-line alignment area can not be used. Hence, for safer photolithography aligning margin the "sizing + 0.25µm" mark on the scribe line's photo alignment area is suggested for better signal acquiring, whose experimental results in UMC shows that around 90% of the alignment signal (W.Q.) can be verified. The alignment accuracy (A.A.) can also be improved through using this technique and is accurate enough as compared to the conventional scanner alignment method used for above 0.13µm generation technology.

Strained silicon induced by the CMOS device process has been considered an important technology for improving the performance of MOSFETs by increasing local carrier mobility in the current channel. In order to evaluate the feasibility of using convergent beam electron diffraction (CBED) in lattice strain determination, high-order Laue zone (HOLZ) lines inside the center disc of a CBED pattern with specific zone axes were kinematically simulated. The intersecting HOLZ lines shift was plotted against the lattice parameter for the determination of uniaxial strain.

SRAM memory is an ideal vehicle for defect monitoring and yield improvement during process development because of its highly structured architecture. However, the success rate of defect detection, especially for soft single-column failures, is decreasing when traditional physical failure analysis (PFA) with only the bitmap is available for guidance. This is due to a variety of invisible or undetectable defects that cause leakage in the device. In order to understand the leakage behavior in advanced high voltage (HV) processes, a Conductive Atomic Force Microscope (C-AFM) [1-4] is introduced to perform junction-level fault isolation prior to attempting PFA. According to J. P. Morniroli [5], crystalline defects affect convergent-beam electron diffraction (CBED) and large angle convergent-beam electron diffraction (LACBED) patterns, so CBED and LACBED techniques were also applied to the specimens containing dislocations to allow further characterization of these defects. In this study quantified data extracted using the C-AFM is also used to establish a connection between the failure mechanism discovered and the soft single column failure mode.