edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 28 NO.4 22 shown in Fig. 1, and additional details on the instrumentation can be found in prior work.[10-12] Nanoscale resolution is achieved via a combination of high geometric magnification and a nanoscale x-ray focal spot size. Geometric magnification is the ratio of the source-to-detector distance to the source-to-IC distance, with higher magnification corresponding to larger magnified feature sizes at the detector plane. In this setup, the metal target is held only 10 µm away from the first layer of circuitry, resulting in geometric magnification of approximately 25,000. At this magnification, a 100 nm feature in the circuit would be projected to 2.5 mm at the detector plane. With HPCD pixel sizes of 75 x 75 µm, which projects to 3 x 3 nm in the sample plane, the spatial resolution is not limited by detector pixel size. In a highmagnification cone beam x-ray CT system such as this and in the absence of other experimental artifacts, the imaging resolution becomes limited by the x-ray focal spot size.[14] Here, the electron beam parameters can be easily adjusted to achieve a certain spot size, allowing imaging conditions to be chosen based on the device being measured. Electron beams are routinely focused to sub-10 nm sizes, resulting in nanoscale x-ray focal spot sizes in the metal target and theoretically enabling imaging of fea- tures much smaller than what is demonstrated in the current measurement. However, reducing the spot size also limits the electron beam current, which is directly proportional to the generated x-ray flux. This balance between spatial resolution and imaging speed is a fundamental trade-off when performing x-ray CT measurements. EXPERIMENTAL CONSIDERATIONS In practice, factors other than the x-ray focal spot size, geometric magnification, and detector pixel size contribute to the final achieved spatial resolution. Two important considerations, positional drift and angle-to-angle alignments, are discussed below. POSITIONAL DRIFT During data collection, a grid of dwell points is acquired at each projection angle. In the measurement shown here, the electron beam was held for 60 seconds at each dwell position. If the beam or sample positioning stage drifts during that period, the final image quality will be degraded by an amount proportional to the magnitude of the drift. If the drift is significant, such as equal to or above the desired minimum resolved feature size, relevant device wiring and defects will not be resolved. Fig. 1 An overview of the x-ray CT instrumentation, show- ing the focused electron beam, metal target and IC, generated and transmitted x-rays, and the x-ray detector. The IC sample is moved relative to the electron beam during data acquisition. This figure is reproduced with permission from Ref 13. Fig. 2 The positional drift per 60 s dwell in X (left) and Y (right) during the data acquisition from one projection angle. With the exception of a small number of outliers with high drift, the majority of dwells move less than 25 nm in any direction during the data acquisition.
RkJQdWJsaXNoZXIy MTYyMzk3NQ==