Seeing beneath the surface
Improving power device reliability with large-area EBIC.
By Greg Johnson, Carl Zeiss Microscopy and Andreas Rummel, Kleindiek Nanotechnik
The deep junction structure of IGBTs plays a vital role in their reliability.
Power semiconductor devices, such as insulated-gate bipolar transistors (IGBTs), are key to the electrification of the planet. They are used in electric and hybrid vehicles, industrial motor drives, renewable energy systems, uninterruptible power supplies in data centers, and more. IGBTs have complex vertical structures with multiple p/n junctions buried deep beneath the surface (Figure 1). These p/n junctions control the flow of electric current in the circuit. Understanding this subsurface electrical behavior is essential for improving device reliability and identifying failure mechanisms.
Figure 1: Depictions of an IGBT device: a) 3D cartoon showing structure, b) cross-sectional view labeled with implants, c) top-down SEM view.
Electron beam induced current (EBIC) is a powerful technique for probing these deeply buried junctions. When combined with advanced scanning electron microscopy (SEM), EBIC enables spatially resolved electrical characterization across large device areas. However, extracting meaningful data requires careful attention to signal interpretation and sample preparation.
This article discusses what EBIC is and why it is important for depth-profile analysis of p/n junctions. It will cover the commonly ignored signals that affect data interpretation. Additionally, it will explain why mechanical polishing provides a superior preparation method compared to traditional etching, enabling more failure analysis (FA) labs to have access to the technique.
EBIC-SEM integrated failure analysis workflow
EBIC is a technique performed inside an SEM that measures current generated when an electron beam interacts with a semiconductor. As the beam penetrates the material, it creates electron-hole pairs. In regions where electric fields exist—such as depletion zones at p/n junctions—these carriers are separated and collected, producing a measurable current.
Unlike purely imaging-based methods, EBIC provides functional electrical insight, making it particularly valuable for FA of power devices. EBIC is uniquely suited for mapping depletion regions, identifying leakage paths and defects, locating electrically active junctions and characterizing subsurface device behavior. For vertical devices like IGBTs, this is critical because the most important electrical activity occurs deep within the structure of the device, far below surface metallization and passivation layers.
While EBIC can be implemented on a variety of SEM platforms, ZEISS field emission (FE) SEMs such as the GeminiSEM series offer specific advantages that significantly enhance EBIC performance. These advantages include stable low-kV operation, signal stability and noise reduction, large-area imaging and an integrated electrical failure analysis workflow.
Stable low-kV operation: ZEISS FE-SEMs deliver stable imaging and beam control at very low landing energies (sub-1 kV). This is critical because lower beam energies reduce penetration depth, enabling surface-sensitive junction analysis. Higher energies increase penetration but can blur spatial resolution and complicate interpretation. The ability to precisely tune beam energy allows users to probe different depths within the device, effectively turning EBIC into a depth-profiling technique.
Signal stability and noise reduction: EBIC currents are often extremely small (nanoamp or lower). ZEISS FE-SEMs provide high beam stability, low noise imaging conditions and consistent signal generation across large scan areas. This stability is essential for distinguishing real EBIC signals from background noise.
Large area imaging: power devices often require analysis across millimeter-scale regions. ZEISS systems support wide field-of-view imaging and uniform signal response across large scan areas. This enables large-area EBIC mapping, which is essential for identifying spatially distributed defects or variations in junction behaviour.
An integrated electrical FA workflow: the ZEISS FE-SEMs integrate well with nanoprobing systems, such as the Kleindiek Nanotechnik PS8e prober shuttle. The integration enables one- or multi-probe EBIC measurements, electron beam absorbed current (EBAC) and resistance contrast imaging. This creates a complete electrical FA environment where structural and electrical information can be correlated directly within the same tool.
The challenge of EBIC measurement interpretation
Despite its strengths, EBIC is not a straightforward measurement. In the paper, “Low Impact Analysis of Junctions in Power Devices,” presented at the 51st International Symposium for Testing and Failure Analysis (ISTFA) conference in November 2025, it was noted that raw EBIC images can be misleading if interpreted naively1. A common assumption is
that bright regions in EBIC images correspond directly to meaningful electrical activity. However, this is often incorrect. Several additional signal sources contribute to the measured current—and failing to account for them can lead to data misinterpretation.
Figure 2: Depictions of power device EBIC measurements from previous work: a) 30 kV EBIC image, top-down, on as-received IGBT; b) top-down EBIC on lightly etched SiC MOSFET; c) low magnification image from same work.
One of the most overlooked contributors to EBIC measurements is secondary electron emission (SEE). In a typical EBIC process, a probe is placed on the device, the sample is grounded, and the SEM beam generates both electron-hole pairs and secondary electrons. These emitted secondary electrons can be picked up by the probe but are not related to junction behavior. This is especially the case for metallic areas within the sample. For example, strong signals observed at the edge of the chip or metal-covered regions often originate from SEE rather than true EBIC effects. These regions may appear bright but contain no useful electrical information.
Another frequently ignored factor is that electric fields at p/n junctions are vector quantities. Depending on the probe placement, carriers may be driven toward or away from the probe. This results in positive or negative current contributions. Thus, EBIC signals are not simply “strong vs. weak”—they can invert depending on geometry. Without being aware of this, interpretation becomes ambiguous.
EBIC signals strongly depend on electron beam landing energy. Low kV results in shallow penetration and limited interaction with deeper junctions. High kV results in deeper penetration and broader interaction volume. As the beam’s energy increases, the apparent width of junction-related signals increases; signal peaks may shift or merge, and contrast can decrease if the beam penetrates beyond the depletion region. Ignoring this dependence can lead to incorrect conclusions about junction size or location.
A key insight from the ISTFA 2025 paper is that maximum signal intensity does not directly correspond to junction position. Instead, useful information emerges when signals are analyzed relative to a baseline. Subtracting background contributions reveals true junction-related features. This approach helps to transform the data into meaningful electrical maps.
By systematically varying beam energy and analyzing EBIC profiles, a consistent pattern emerges. At low energies, only shallow portions of depletion regions are detected, and at higher energies, deeper portions become visible. This signal broadening reflects increased electron penetration. Combining electron transport simulations enables estimation on junction depth, mapping of depletion region geometry and quantitative correlation between beam energy and electrical response. This transforms EBIC from a qualitative imaging method into a quantitative characterization tool.
Figure 3: View of the chip under dimple polishing: a) view of the dimple polisher, b) samples after three minutes, c) after an additional one minute, d) completed work.
Sample preparation is critical for EBIC
Sample preparation is one of the most critical factors in EBIC analysis. Acid etching has traditionally been used to expose subsurface features, but it introduces several problems. Non-uniform material removal results in inconsistent surfaces, and buried regions remain blocked by metal or dielectric layers. It also results in surface roughness variability, which complicates interpretation. Etching produces surfaces where EBIC signals are incomplete, spatially inconsistent and difficult to interpret at low magnification (Figure 2).
Carefully controlled mechanical polishing, especially dimpling techniques, offers significant advantages (Figure 3). It creates a gradual depth profile across the sample exposing surface structures at the edges and deep junctions near the center, enabling continuous observation across multiple device layers in a single scan.
Unlike etching, polishing produces smooth, controlled surfaces and reproducible depth gradients—essential to correlating EBIC signals with physical structures. Large portions of the device remain intact, which is critical for large-area mapping. With polishing, there is a more gradual exposure of junctions, allowing for clear correlation between structure and EBIC response and systematic analysis of depth-dependent behavior—resulting in better interpretation of the signals. Finally, mechanical polishing offers environmental and safety benefits since there is no need for hazardous chemicals or complex waste management. This makes EBIC more accessible to a broader range of FA labs.
Enabling large-area, depth-resolved EBIC
In summary, with controlled mechanical polishing, stable low-kV SEM operation and careful EBIC signal interpretation, it becomes possible to perform large-area, depth-resolved electrical characterization of power devices (Figure 4).
This approach provides:
- Wide-field mapping of junction behavior
- Identification of electrically active regions
- Quantitative estimation of junction depth
- Improved understanding of failure mechanisms
Figure 4: Low magnification EBIC analysis showing signals from the entire width of the chip. The analysis is taken from the location on one of the metallized emitter lines, as shown in the lower centre.
Conclusion
EBIC is a uniquely powerful technique for probing the internal electrical behavior of semiconductor devices, particularly for complex vertical structures like IGBTs. However, its full potential is only realized when both instrumentation and methodology are carefully optimized.
ZEISS FE-SEM platforms provide a strong foundation for EBIC through stable low-kV operation, high signal fidelity and seamless integration with nanoprobing systems. These capabilities enable precise control over beam interaction and support large-area analysis.
Equally important is a correct understanding of EBIC signal formation. Contributions from secondary electron emission, field directionality and beam energy effects are often overlooked but critically shape the measured signal. Proper interpretation requires moving beyond simple image contrast and adopting a more rigorous, physics-based analysis.
Sample preparation plays a critical role. Mechanical polishing offers significant advantages over acid etching by providing uniform, large-area access to subsurface structures while improving reproducibility and interpretability.
These advances transform EBIC into a quantitative technique for evaluating p/n junctions across large areas, supporting improved reliability and deeper insight into advanced power semiconductor devices.
