edfas.org 15 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 27 NO. 4 long timescales, effectively “remembering” the irradiation history.[5] HOW TO AVOID SEM-INDUCED DEGRADATION This study not only investigates the degradation caused by SEM but also looks at ways to reduce or even prevent it. This is done by optimizing the SEM parameters, the tungsten probes, and the nanoprobe itself. These three factors are examined below. OPTIMIZATION OF SEM PARAMETERS The most direct means of reducing electron-induced degradation is through careful selection of beam parameters: • Lower landing energy: Reducing the accelerating voltage (e.g., to 100 to 200 eV) minimizes penetration depth and reduces charging in buried layers, while still providing surface-sensitive imaging.[6] • Reduced probe current: Lower beam currents directly decrease electron dose per unit time, though at the cost of reduced signal-to-noise ratio. • Fast scanning and reduced dwell times: Shorter pixel dwell times and pixel integration strategies can limit localized charge buildup and minimize exposure of any one region.[3] Together, these settings help ensure that the cumulative dose remains below the critical threshold for inducing dielectric charging or trap creation. As technology nodes continue to shrink, dose management will become increasingly critical, and emerging solutions such as beam deceleration techniques may play a role in maintaining device integrity during high-resolution elec- tron microscopy. OPTIMIZING TUNGSTEN PROBES Nanoprobing under low accelerating voltages in the SEM presents unique challenges that lab engineers regularly face during fault analysis of advanced semiconductor devices. Standard tungsten probes, while mechanically robust and conductive, can introduce unexpected imaging distortions when operated below ~200 eV. This is primarily due to the paramagnetic nature of tungsten, which interacts with the SEM’s electron beam and amplifies the effects of even weak magnetic fields—leading to beam deflection, image drift, or degraded resolution, resulting in a distorted image. To maintain stability and precision in such sensitive conditions, specially prepared probe needles are required. These must be mechanically sharp and clean, as well as have a special geometry to minimize the magnetic interaction with the electron beam. Off-the-shelf needles often fail to meet these criteria, especially for high-precision contact on sub-10 nm structures. The example image (Fig. 2) shows such special low-eV tungsten probes, which have a greatly reduced amount of tungsten near the electron beam and cause only minimal image distortion even at 100 eV. In this study, the probes were produced with the SMARPROBE etching station, which can be installed directly next to any SEM to produce fresh probes oxide-free and free of contaminants.[7] OPTIMIZING NANOPROBING While the aforementioned parameters, such as optimized imaging conditions and reduced electron beam energies, undoubtedly contribute to a reduced beaminduced degradation, particularly under low-energy conditions. Another effective strategy for reducing cumulative electron dose during nanoprobing lies in minimizing the total number of required images at the point of interest. Conventionally, a typical nanoprobing workflow involves continuous SEM imaging throughout the alignment and positioning of probes, often spanning several minutes and generating hundreds of SEM frames. But how to reduce the image count during probing? Fundamentally, after the relative spatial positions of both the sample and the individual probes have been determined—e.g., via three-point alignment routines or Fig. 1 The degradation dynamics of 22 nm FD-SOI technology were investigated by varying the electron dose incident on the buried oxide layer. The total electron dose depends on the beam current, beam energy, and imaging time (scan speed and total number of images).
RkJQdWJsaXNoZXIy MTYyMzk3NQ==