Capability and Uncertainty Analysis of Desktop SEM-Based Surface Roughness Measurement Using Backscattered Electron Split-Detector Imaging

Desktop scanning electron microscopes (SEMs) are increasingly adopted in industrial environments due to their accessibility and rapid imaging capability; however, their potential for quantitative surface metrology remains insufficiently validated. This study investigates the capability of a desktop SEM for surface roughness measurement using backscattered electron (BSE) split-detector imaging, with emphasis on measurement accuracy, repeatability, and uncertainty. A structured experimental framework was implemented using certified roughness artefacts covering a wide range of surface textures. A physics-informed signal processing pipeline was developed to convert BSE intensity variations into surface gradients, followed by numerical reconstruction and extraction of standard roughness parameters (Ra, Rq). Measurement uncertainty was evaluated in accordance with the ISO Guide to the Expression of Uncertainty in Measurement (GUM), considering contributions from detector response, surface alignment, electron interaction effects, and signal processing. Results show that the desktop SEM provides reliable and consistent measurements within a defined operational range. In the mid-range roughness regime (Ra ≈ 0.5–2 µm), measurement deviations were below 6% relative to certified reference values, with good repeatability across repeated measurements. At lower roughness levels, increased variability was observed due to signal noise and interaction volume effects, while higher roughness surfaces exhibited deviations associated with non-linear signal response and shadowing effects. Uncertainty analysis identified calibration artefacts, detector asymmetry, and surface alignment as dominant contributors to the overall measurement uncertainty. The expanded uncertainty (k = 2) remained within acceptable limits for industrial screening applications within the identified operational range. The findings demonstrate that desktop SEM systems can be extended beyond qualitative imaging to provide quantitative, uncertainty-aware surface roughness measurements. While not a replacement for established metrology techniques, the approach offers a practical and flexible solution for preliminary inspection and process monitoring in modern manufacturing environments.