May_EDFA_Digital

edfas.org ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 24 NO . 2 20 polished using an autopolisher with a sequence of coarse to fine Al 2 O 3 and diamond abrasives; the final polish was done with a 0.05 µm Al 2 O 3 disk as the gate contact G5 came into view. [11,12] For optimum SCM results, the root- mean-square (RMS) roughness (R q ) of the polished surface should be at 5 Å or lesswhen the (device) region of interest is exposed. Therefore, all subsequent “final” polishing was performed by hand using an in-house-built fixture (Fig. 3) and a stationary (i.e., non-spinning) polishing pad with 0.04 µm SiO 2 slurry. After reaching the G5 metal contact (Fig. 2 near the middle), an iterative approach was used to expose the FinFET device of interest. With the sample mounted on the tripod, the cross-section face was made to rest flat (parallel) on a smooth surface (e.g., glass slide, silicon wafer, etc.) by adjusting the micrometers. The tripod was then placed on to a polishing pad with 0.04 µm SiO 2 slurry. Using moderate downward hand pressure, the tripod was moved on the pad in a figure-eight pattern for a couple of times. The sample was then cleaned and imaged using atomic force microscopy (AFM) to determine proximity to the region of interest (i.e., are the P4 fins visible?), and surface roughness measured in the bulk silicon region. This iterative approach was labor- intensive and time-consuming. It was repeated until the FinFET devices of interest were visible. A fillet of silver paint was applied around the perimeter of the sample to help provide a good electrical contact to the sample via the silicon sub- strate and the deposited Cr layer. As a final preparationprior toSCMandSCS analyses, the freshly polished, cross-section surface underwent an oxidation process by baking the sample on a hot plate at 250ºC for 30 minutes while exposing the region of interest to ultraviolet light (λ=365 nm). [13] The second set of samples were pre- pared by inverted TEMsample preparation in a FIB-SEM followed by low-energy Ar ion milling. To target a specific fin, the sample was ionmilled to removematerial and the target site was reached. The milling angle, voltages (acceleration and gun voltage), mill rate, and time were optimized to minimize damage and differential milling Fig. 5 SCM image of sectionA-A’ showing the presence of a dopant-relateddefect on the rightmost fin of P4. All of the other fins associated with N4 and P4 appear normal. Fig. 4 (a) Sample affixed to a TEM grid. (b) Cross-sectional SEM image of 14 nm SRAM cells showing the waterfall effect. (c) SEM showing Pt wire for electrical contact. (d) TEM image of the SRAM cell FinFETs. Fig. 3 In-house-built polishing tripod with a cross-section sample affixed using “black wax.” (a) (b) (c) (d)