Cross-Section SEM
Cross-section SEM examines an exposed face through a structure, revealing geometry that a top-down view cannot directly show. A prepared section can reveal layer thicknesses, sidewall shape, interfaces, and internal voids at the selected location. The electron microscope images the newly exposed surface; it does not simply look through an intact opaque wafer.
The result is a two-dimensional section through a three-dimensional object. Its meaning depends on where the cut passes, how the face was prepared, and how the specimen is oriented for imaging. A sharp image is useful evidence, but it is not automatically an undistorted or representative section.
The exposed stack connects shape to process history. A cross section can help distinguish a tapered opening from a vertical one, reveal incomplete fill, or locate a discontinuity between layers. Brightness alone does not uniquely identify composition: detector choice, topography, charging, and material response all influence contrast. Where chemical identification matters, use appropriate complementary analysis and account for its spatial resolution.
Preparation determines what can be observed. Cleaving can expose a section efficiently when the specimen and target geometry permit it. Mechanical preparation and broad ion-beam polishing can serve other scales and materials. Focused ion beam (FIB) milling provides site-specific access, often guided by surface landmarks or coordinates from earlier inspection.
A dual-beam FIB-SEM combines ion-beam preparation with electron-beam imaging. Gallium is one ion-source option, not the definition of FIB: plasma sources and other species are also used. Material-removal rate, targeting, surface damage, and image quality depend on the instrument and preparation conditions. A protective cap and suitable finishing steps can help preserve the region of interest, but the preparation must be qualified for the specimen.
Cut position changes the apparent dimension. Consider a cylindrical hole. A section through its center shows its diameter, while a parallel section away from the center shows a shorter chord. An apparently narrow feature can therefore result from an off-center cut rather than a narrower manufactured opening. Real structures add taper, asymmetry, and depth-dependent geometry to this targeting problem.
The cut direction also matters for long lines and trenches: an oblique section can exaggerate a lateral width. Separately, imaging a face at a tilt can foreshorten an apparent dimension. Measurement should use the known specimen and viewing geometry, calibrated image scale, and the intended definition of thickness or width. A scale bar alone does not correct an incorrectly selected section plane.
Preparation can create convincing false structure. Uneven milling can produce curtaining, visible as streaks or relief along the exposed face. Sputtered material can redeposit on surfaces. Ion interaction can modify a near-surface region, and mechanical preparation can smear or damage some materials. These effects may obscure a real defect or resemble one.
Suspicious contrast should be evaluated against preparation direction, detector response, and supporting evidence. Another section, a changed imaging condition, or a complementary method can help test an interpretation. A dark patch is not automatically a void, and a bright line is not automatically a distinct material. Likewise, absence of a defect in one section does not prove that it is absent elsewhere in the volume.
A cross section is not a three-dimensional reconstruction. Serial sectioning removes material in increments and images successive faces; aligned images can support a reconstructed volume. Slice spacing, drift correction, segmentation, and preparation artifacts affect that reconstruction. One well-chosen section may answer a local question, while connectivity or a distributed defect can require volumetric evidence.
SEM sections and TEM lamellae are different specimens. SEM commonly images the exposed face of a comparatively bulk specimen. Transmission electron microscopy requires a suitably electron-transparent specimen, often prepared as a thin lamella. FIB can assist both workflows, but preparing a cross-sectional SEM face does not automatically make a TEM-ready sample.
Destructive does not mean excluded from manufacturing control. Cleaving or milling changes the sampled region. Nevertheless, specialized wafer-level FIB-SEM systems can support in-line metrology and process monitoring where destructive sampling is authorized. That is different from nondestructively measuring every product feature and returning all sampled material unchanged. Site choice and material disposition must reflect what was removed or altered.
Do not substitute a headline instrument resolution for the uncertainty of a measured feature. Surface quality, contrast, calibration, cut placement, viewing geometry, and edge definition all contribute. The report should identify the site, orientation, preparation method, relevant imaging conditions, and how the measured boundary was selected.
The key idea: cross-section SEM reveals a chosen interior plane, not an unquestionable ground truth of the whole device. Its strongest results combine accurate targeting, suitable preparation, controlled imaging geometry, and evidence that separates real structure from artifacts.
Sources: Thermo Fisher Scientific cross-sectioning, FIB-SEM sample-preparation, and semiconductor metrology documentation; ZEISS FIB-SEM characterization and tomography documentation. Diagrams are explanatory schematics, not measured micrographs or equipment specifications.