May_EDFA_Digital

edfas.org ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 23 NO . 2 34 acknowledged that “FIB” is a widely diverse topic with a number of primary applications and tool configurations. We listed the 15 FIB topics suggested by registrants, and the sevenmicro-talks prepared by our panel as conversa- tion starters. Moderator Valerie Brogden kicked off the talks with a discussion of the experimental results obtained on the Hydramulti-species Plasma FIB. Experiments were run to compare yields with various ions (O + , N + , Ar + , Xe + vs Ga + ) at different accelerating voltages. Surface smoothness, the possibility of some level of chemical etching/interaction, shaping as a result of sputter yield, and the surprising findingof changes in surfacehardness promptedanumber of questions and discussion. Fundamentals like pattern- ing choices may need to be adjusted for specific interplay between a selected primary ion and the target material. Sina Shahbazmohamadi from the University of Connecticut REFINE labs presented work on a unique combination tool platform that combines a highly flex- ible fs-laser systemwith anOrionmicroscope. Correlative methods allow precise laser patterning overlay for selective clearing of materials, example being dielectric between structures under investigation. Three ion beam choices (He + , Ne + & Ga + now, more possible) allow for milling, polishing, andultrahigh-resolution imaging. Other microscopy methods are possible after a inter-chamber transport system is fully realized. SteveRandolph fromThermoFisher added to the laser- FIBdiscussionwith a presentationon their combinationof a fs-laser with a PFIB (Xe + & e - ) in a “3-beam” arrangement. White light and IR camera can also be added plus gases for additional materials patterning and imaging choices. A shutter system protects sensitive components from ablatedmaterial. The inclusion of the laser to the already high volume of material removal by PFIB allows for very rapid prep on larger structures. Depending on the mate- rial, the laser ablationmight yield a lowsurface roughness suitable for immediate analysis. At worst a relatively fast final polish will be required. Jozef Vincenc Obona of TESCAN finished the laser and PFIB discussion by presenting their vision of an optimized workflow for best lab tool utilization, especially when packaging volumes are involved. Several possible multi- tool workflows were described. Work involving stacked chips, system-level investigations, even ball bonded die can greatly benefit from the rapid removal capability of a laser-first approach where the laser is not necessarily integrated. Final slice/view and sectioning face polish is available post laser in the PFIB. Rocking stage and other techniques optimizes final quality. Oneunrelatedquestion about PFIB use involved planar delayering for reverse engineering (covered last year). Valery Ray fromPBS&T introduced the audience to an alternative to a factory service contract or single purchase OEM support for out-of-warranty tools. He discussed the availability of third party or custom made, along with recovered/recycled/rebuilt components that could be obtained tokeepolder tools functional. Togetherwithdata from TechInsights and SEMION, he closed his talk with a few slides that showed that FIB and SEM columns could be rebuilt and requalified in the field. Common “column killers” were discussed in his Q&A. Roy Goldman of Amazon’s Annapurna Labs walked us through the backside chip edit process, and the particular challenges that the 7 nmprocess presents for prototyping design modifications. While tungsten contacts could be selectively etched with XeF 2 , no such solutions currently exist for cobalt interconnects, not to mention line width/ density challenges. Is it time to look beyondGa + He + or Ne + FIBs for these critical cuts/connects? Discussion ensued around sub-pAbeamcurrents, lowkV andother tricks that might extendGa into some 7 nmsituations, alongwith the unresolved challenges of alternate ion tool editing. The final talk was by moderator Steve Herschbein. He covered the difficulty of continuing to offer certain highly specialized laboratory services in a cost-constrained environment, and the very survival of some labs in a time of rapid consolidation. Foundries don’t do FIB chip edit, fabless design houses are going lab-less, and many companies are in search of outsourcing. While big critical customers can still demand these services, smaller cus- tomers lack the clout. Froman employee standpoint, your survival oftendepends onhowwell you communicatewith uppermanagement about your customer’s needs, and the value-add your work brings. The general Q&A section of the User Group included discussions about what type of non-traditional samples can safely be placed in a high vac tool, handling oversized assemblies that will not go through a load-lock, FIB chem- istry for new semiconductor materials, junction defect localization, andhow to get that failuremode photograph. Special thanks to our scribe, David Pan. "FOUNDRIES DON’T DO FIB CHIP EDIT, FABLESS DESIGN HOUSES ARE GOING LAB-LESS, AND MANY COMPANIES ARE IN SEARCH OF OUTSOURCING."