edfas.org 17 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 26 NO. 4 sample is scanned by XRM at low resolution to obtain an overview of the entire package and interconnect structure to check for defects or anomalies (Fig. 5). The process can be accelerated if needed information is already available from other fault isolation techniques or known data, allowing the low-resolution x-ray scan to be skipped. After the region of interest (ROI) is identified, the sample is moved into the LaserFIB and the topmost surface above the ROI is marked by the fs-laser, adding fiducials to easily reference the sub-surface feature. The sample is then rescanned using XRM at the desired resolution. This scan captures both ROI and the surface fiducials to localize the ROI with respect to the surface fiducials. Then the fs-laser performs a precise fine cut on the desired wire or interconnect to isolate features or parts of the circuitry, and the result is rechecked using XRM to determine if the precise cutting is sufficient and successful for further fault isolation. This approach makes it possible to precisely laser cut a 20-µm-wide wire that is 150 µm deep, without decapsulating or damaging any parts of the die and other interconnects or neighboring wires. Accuracy is further improved by an additional calibration step to determine the parameters for accurate positioning and laser milling depth. The entire process can be sped up for subsequent work, since the laser parameters can be replicated on additional wires or on other samples made of similar materials. Figure 6 shows a sequence of images that illustrate the parameter optimization strategy. The initial tests made three cuts across multiple wires using different laser Fig. 5 Sample preparation workflow using 3D x-ray to guide laser milling of interconnects for fault isolation. 3D XRM is especially effective to image the laser-marked fiducials, and correlative software on the LaserFIB allows for alignment of the XRM data to the SEM, enabling navigation and set-up of the laser cuts to the target area. Fig. 6 Establishing optimal laser processing parameters. a) Top surface (in green) with laser fiducials is overlayed with wires (in orange) and the position of laser milling test patterns 1, 2, and 3 with varying doses are shown. b) SEM image after fs-laser ablation. c) XRM plan-view virtual slice showing laser edits made to the wires and d) virtual cross-section view of one wire cut by the three test recipes. (a) (b) (d) (c)
RkJQdWJsaXNoZXIy MTYyMzk3NQ==