November_EDFA_Digital

edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 19 NO. 4 42 (Fig. 8) depicts the dose matrix in subregions within the 800 µm × 800 µmwindow. The exposed regions vary from the contact level through to M3. CONCLUSIONS This article demonstrated aworkflow for deprocessing ICs from the backside using a combination of automated adaptive backside ultrathinning and large-area plasma FIB delayering. Advantages to this approach include a reduction in manual planarization and depackaging. Automated ultrathinning also achieves a higher degree of precision and repeatability. The plasma FIB delayering process following ultrathinning initiates from within 1 to 2µmof theactivedevice structure, permittinghigh-quality delayering and imaging over large areas beginning at the highest-density device structures. Using a backside approach, it is also possible to preservemaximumdevice functionality for probing and powering the device within the plasma FIB-SEM. Recently the entire process for plasma FIB delayering and SEM imaging has been fully automated via Python scripting. This additional automation permits higher pre- cision in the process and allows unattended operations to perform any desired number of delayering/imaging cycles. The delayering time and imaging parameters may be defined by the user. This automation has been coupled with a commercial computational visualization engine by Object Research Systems. The computation visualization engine is programmatically controlled via Python, and capabilities related to the delayering application include the ability to perform image analysis, stitching, and visu- alization in a near-real-time environment. Image datamay be collected by the computational visualization engine as it becomes available. Instrument control commands can also be fed back into the FIB-SEM during operation, all within the same Python program environment. The x-ray data, thickness data from the ultrathinning process, and delayering image data may be processed and displayed within an integrated volume representing the data cube. Collectively, the interfacing of a programmatic compu- tational visualization engine with an FIB-SEM platform creates a new and significant opportunity for computa- tional guided microscopy and user-defined automation. Fig. 7 (a) Gate structures are highlighted at 5 kV following removal of the contact layer. (b) The 30 kV image from the sequence in the delayering process reveals the M1 and M2 layers. The field of view is 30 µm in both images. (a) (b) (a) (b) Fig. 8 (a) 5 kV and (b) 30 kV image pair taken from the same region of interest and at the same cycle in the delayering sequence. The area shown is part of the dosematrixwhere regionswere exposed to different plasma FIB delayering times. The field of view is 107 µm.