Every failure analyst knows the moment: the curve tracer says the device is dead — a shorted junction, a leaky diode, a solar cell whose fill factor fell off a cliff — and the SEM image says everything looks fine. Electrically broken, visually perfect. The defect exists, it’s just invisible, because what’s broken isn’t the shape of the material. It’s the electrical behavior inside it.
EBIC — electron-beam-induced current — is the SEM technique built for exactly this gap. Instead of collecting electrons or X-rays coming out of the sample, EBIC measures a current flowing through it, and turns that current into an image of where the electronics actually work. Our EBIC technology page covers the configuration details; this post is about what the technique tells you and when to reach for it.
The physics, in one paragraph
When the beam enters a semiconductor, each primary electron generates thousands of electron-hole pairs — creating one pair costs roughly three times the bandgap energy, so a 20 keV electron landing in silicon produces on the order of five thousand of them. Ordinarily they just recombine. But if they’re generated near a p-n junction or a Schottky contact, the built-in electric field sweeps electrons one way and holes the other before recombination can happen — and that charge separation is a real, measurable current in an external circuit. A picoampere beam becomes nanoamperes of collected signal: the device itself amplifies by a factor of a few thousand. Scan the beam, record the current at every pixel, and you have an image drawn entirely in “how well does this spot collect charge?”
How to read an EBIC image
Bright means collecting. Regions near a healthy junction light up — in fact, the bright band is the electrically active region, which makes EBIC the most direct way there is to see where a junction actually sits, how uniform it is, and how far the depletion region extends. Since beam energy sets penetration depth, stepping through kV profiles the junction in depth as well.
Dark means something is eating carriers. Dislocations, stacking faults, grain boundaries, metallic precipitates, and process-induced damage all act as recombination centers — electron-hole pairs generated nearby die before the field can separate them. They appear as dark spots, lines, and boundaries on a bright background: an electrical-defect map at SEM resolution, showing flaws that cast no shadow in a secondary-electron image.
Dead means disconnected. Whole regions that should collect but don’t — a delaminated contact, a cracked finger on a solar cell, a shunted zone — show up as extinguished territory, which localizes the failure to a place you can then cross-section.
What people actually use it for
- Solar cells: the classic application. Grain-boundary recombination in multicrystalline silicon, shunts, dead fingers, and process-damage halos are all textbook EBIC contrast — and cell-level efficiency questions become spatial maps you can act on.
- Power devices: junction termination, edge structures, and localized breakdown sites in silicon, SiC, and GaN devices — materials where a single killer defect controls yield.
- Failure localization: EBIC narrows “the diode leaks” to “the defect is here, 40 µm in from the pad,” which is what makes the subsequent cross-section worth cutting. It slots naturally into the workflow we described in desktop SEM electronics failure analysis.
- Junction metrology: delineating implant and diffusion profiles in cross-sectioned devices — seeing the junction rather than inferring it.
- Materials research: line scans across a junction yield minority-carrier diffusion length, the single number that summarizes electronic material quality.
For a sense of how this pairs with elemental work on the same platform, our semiconductor process-control post is the companion read: EDS and EBSD tell you what the material is and how it’s built; EBIC tells you whether it works.
What EBIC needs from you
Honesty section, as always. EBIC is not point-and-shoot imaging:
- It needs a junction. The technique images charge collection, so the sample must contain a p-n junction or Schottky barrier. On bare material with no built-in field there is nothing to collect — though evaporating a simple Schottky contact onto a wafer is a standard and quick way to make one for materials evaluation.
- It needs electrical contact. Two connections to the device, made inside the chamber — via a purpose-built EBIC sample holder, probe needles, or bonded wires. Sample prep becomes sample wiring, and a low-noise current amplifier handles the signal. The EBIC detector option for the SNE-Alpha packages this.
- Beam conditions matter. kV sets generation depth (too high and you generate carriers far below a shallow junction; too low and you never reach it), and beam current wants to stay modest — carrier generation is the signal, but heavy injection washes out contrast and can bias the measurement.
- Interpretation is comparative. Bright/dark contrast is unambiguous; turning it into quantitative recombination parameters takes controlled line scans and a model. Most industrial questions — where is the junction, where are the killers, is this region dead — don’t need the model.
Why this works on a benchtop
EBIC has no exotic vacuum or source requirements — it needs a stable, small probe, a sample stage you can wire through, and quiet electronics. The SNE-Alpha carries a factory EBIC port alongside its EDS, EBSD, Raman, and CL ports, with the 1–30 kV range covering shallow and deep junction work. That puts an electrical-defect microscope in the same room as the process it debugs — the difference between “send it out and wait” and “answer before lunch,” which for yield engineering is the whole game. More on the application space on our semiconductor applications page.
Frequently asked questions
Do I need to bias the device?
No — classic EBIC uses the junction’s built-in field alone, which is what makes it so clean for mapping. Applying bias is an option for studying breakdown sites and field-dependent behavior.
Can I run EBIC on a finished, packaged part?
The beam must reach the active region, so packaged parts need decapsulation, and thick passivation may call for higher kV or careful thinning. Cross-sections of failed devices are often the practical route.
How is EBIC different from a photoluminescence or thermal map?
Same family of question, different probe and scale: EBIC uses the electron beam, so its spatial resolution is far finer, and it works on structures light can’t reach. Many labs use PL or thermography to find the neighborhood and EBIC to find the house.
Have a device that measures broken but looks perfect?
That’s an EBIC question. Tell us what the part is and what the electrical symptom looks like, and we’ll talk through whether EBIC on a benchtop SEM can localize it — and image a sample if you send one. Get in touch.