| AVS 63rd International Symposium & Exhibition | |
| Advanced Ion Microscopy Focus Topic | Thursday Sessions |
| Session HI+MI+NS-ThA |
| Session: | Ion Beam Based Imaging and Nanofabrication |
| Presenter: | Yuval Greenzweig, Intel Corporation |
| Authors: | Y. Greenzweig, Intel Corporation Y. Drezner, Intel Corporation A. Raveh, Intel Corporation |
| Correspondent: | Click to Email |
The challenges of Circuit Edit (CE) using focused ion beam (FIB) are driven by the perpetual down-scaling of minimum features per VLSI process technology generation. The recent emergence of FIBs with much reduced probe sizes relative to Ga LMIS based tools, may provide a long-needed revitalization of FIB nanomachining capabilities such as FIB image resolution and machining acuity, necessary for nanomachining tasks such as CE. However, other requirements must go along, driving preferences of ion species, ion energies, and requirements for system performance in several areas.
Among the challenging requirements of CE is the task of milling in a controlled and planar fashion through layers of parallel metal lines with intervening dielectric, and end-pointing on a metal layer of choice. The end-pointing is based on the real-time secondary electron (SE) image during ion milling, and the requirement is leaving most of the target metal intact. If linear dimensions of features, such as minimum metal widths, reduce by a factor α from one VLSI generation to the next, then maintaining quality realtime milling images, i.e., sufficient resolution and signal to noise, requires milling vertically through a layer proportional to α-2 relative to the previous generation - same number of ions, but in a smaller pixel. On the other hand, the vertical thickness of the metals has also decreased by a, causing the etching to scale as α-3 relative to the thinner new metal thicknesses. Previous VLSI process generation scaling factors have been approximately α = 0.7, so that the severity of this challenge has been getting worse by ~3X for several generations and is now at the feasibility limit. To improve on this, SE emission and collection efficiency must improve, and in particular SE collection efficiency of normally emitted SEs, which are the bearers of the information from the bottom of these milling boxes. The figure of merit representing this challenge is the SE yield times the SE detector collection efficiency, divided by the sputter yield (or etch rate), this provides opportunity for GFIS sources.
Other challenges of the CE application which providing preferences of ion species and ion energy will be discussed.