February_EDFA_Digital

edfas.org 41 ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 24 NO . 1 advanced algorithms to ‘fill in’ the pixel information gaps in a sparse data field. High resolution imaging techniques often result in materials damage from the application of excessive dose (electrons or ions). Reducing the applied dose considerably and obtaining an image almost as good as the benchmark has been demonstrated with this methodology on SEM/STEM systems. Whether it can be used effectively on already very data-limited deep hole FIB images will be a worthy challenge. Valerie Brogden fromCAMCOR at University of Oregon gave a follow-up talk to last year’s introduction to the multi-ion species Plasma FIB. This year she presentedpre- liminary TEM results characterizing amorphous damage at multiple kV for Xe, Ar, N, and O compared to what is expected fromGa into silicon. Argon and nitrogen yielded similar amorphous damage depth to Ga, but the argon results were overall better in appearance. Xenon depth damage was about half that of Ga, as expected. The wild card was oxygen. Damage depth didn’t always track kV, and patchy spots suggest possible chemical interaction in addition to the expected lattice disruption. Results were compared and contrasted to those obtained fromThermo Fisher studies. Lolita Rotkina from TRUMPF Photonics could not be with us in person, but she sent a video. As Valerie was familiar with her work, she gave summary results. Lolita is working with GaAs semiconductor material, and a Ga FIB will yield unacceptable milling results due to interac- tion and the formation of Ga droplets. Can a plasma FIB with multi-ion species do a better job? Xenon interaction with the substrate gave a similar droplet result. Argonwas somewhat better, but still resulted in droplets. Nitrogen milling under a variety of conditions gave a consistently smooth and droplet-free surface. SteveHerschbein, consultant, gave the final talk on the subject of lab design for high performance imaging and milling tools. His single slide listeda number of factors that should be considered, including temperature, humidity, airflow, electrical power conditioning, electromagnetic fields mitigation, floor motion, and operator placement. Seeminglysmall thingscanseverely limit tool functionality, and exceeding site survey specifications is a must. Plenty of tool owner horror stories ensued; it was a fun topic! User initiated topics and general Q&A included large area planar delayering techniques (several were men- tioned and discussed). The group then went back to a discussion on Rick’s “Future of FIB Edit” topic and talked more about FIB tools in the wafer fab. Many thanks to Dustin Davis, Apex Semiconductors, for being a scribe. “HIGH RESOLUTION IMAGING TECHNIQUES OFTEN RESULT IN MATERIALS DAMAGE FROM THE APPLICATION OF EXCESSIVE DOSE (ELECTRONS OR IONS).” semiconductor technology and the current capability of the FA community. “FIB” is a widely diverse topic, with a number of primary applications and tool configurations. A printed list of 40 FIB topics as discussion suggestions assembled from past inputs was provided along with post-it notes for attendees to vote or suggest new items. While voting took place, seven preparedmicro-talks were presented as conversation starters. Cheryl Hartfield from Zeiss and Scott Fuller from Tescan presented two different views of the logical combination of a laser for fast and high volume material ablation as a precursive step to fine FIB cross-sectioning or machining for other analysis tasks. The Tescan system uses navigation hooks to a commercial stand-alone laser system in conjunction with their plasma FIB, while the Zeiss uses a laser built into their LMIS galliumcrossbeam. Discussions followed with the two panelists about tool utilization, speed and workflow optimization, damage heating zone of femtosecond vs picosecond lasers, sub- sequent analysis by 3D x-ray, prep of difficult shapes like atom probe samples, sample/chamber cleanliness, the ability to use different gases with the laser, and the capa- bility of multi-axis samplemovement in conjunction with patterning during ablation. Rick Livengood from Intel took the group through the evolution of FIB chip edit, the pinch points when working with current module technology, and the challenges to be faced as next-gen system-in-package products come online. Embedded silicon bridgeswithin packagematerial (EMIB) and power to gate-all-around transistors delivered directly by way of an underside bonded wafer (PowerVia) will leave very limited opportunity for access to enable FIB circuit modification. Is the end near for this treasured prototyping method? Can we remove the underside power grid and restore? Revisit wafer level editing prior to bonding? Edward Principe from Synchrotron Research intro- duced us to an image processing technique that uses