EDX Analysis and SEM Imaging in PCB Failure Analysis

Failures rarely announce their own cause. A pad that lifts after reflow, a joint that cracks in the field, a residue that corrodes a trace months after shipment — each of these leaves chemical fingerprints behind. When a board fails, the fastest route from symptom to root cause usually runs through a microscope and an X-ray detector. This article explains how EDX analysis and SEM imaging combine to turn a failed assembly into hard, defensible evidence that a process engineer can act on.

Why Elemental Evidence Beats Guesswork

Eyeballing a failed board produces hypotheses, not answers. A discoloured pad could be flux residue, intermetallic growth, or a plating defect, and those three possibilities demand completely different corrective actions. Visual inspection alone cannot separate them, and sectioning without a plan often destroys the very surface that carried the clue. Elemental data settles the argument quickly because it identifies what is genuinely present rather than what the inspector suspects. That is the core reason laboratories investigate solder defects with instruments rather than eyes alone.

It matters commercially as well as technically. When a customer returns a field failure, the supplier usually absorbs the cost unless the mechanism can be demonstrated. A documented failure analysis with spectra and images assigned to specific locations gives both parties a shared factual basis. It shifts the conversation away from blame and toward process correction, which is the only outcome that actually prevents the next recurrence on the line.

How the SEM Builds the Image

A scanning electron microscope focuses a beam of electrons onto the sample and collects the signals that come back. Secondary electrons reveal topography with resolution far beyond optical microscopy, which is why fine cracks, whiskers and etch pits become visible at all. Backscattered electrons are sensitive to atomic number, so heavier elements appear brighter and a plated layer separates cleanly from the surrounding laminate in the image.

Depth of field is the other advantage. A polished cross section of a plated through hole can be held sharply in focus across a large height range, letting an operator trace copper thickness from the surface down to the barrel wall in a single frame. That combination of resolution and depth is what makes the microscope the natural platform on which to mount an elemental detector.

What EDX Analysis Actually Measures

Energy dispersive X-ray spectroscopy detects the characteristic X-rays emitted when beam electrons knock inner-shell electrons out of atoms in the sample. Each element returns a fingerprint of energies, so the detector reports which elements are present and roughly how much of each. The technique is fast, needs little or no coating on conductive samples, and works on features only a few micrometres across.

The word roughly deserves emphasis. EDX analysis is semi-quantitative at best for light elements, and the interaction volume is larger than the beam spot, so a thin surface layer is always measured together with the material beneath it. Treat the numbers as ratios that support a comparison between locations, not as certified assay values from an accredited laboratory.

Sample Preparation Without Destroying the Evidence

Every cut removes information. The goal of preparation is to reach the feature of interest while preserving enough context to interpret it. For a suspect joint, that usually means a targeted cross section through the joint rather than a full board grind. For a residue, it means lifting or replicating the deposit onto a clean substrate so the surrounding laminate does not dominate the spectrum.

Cleanliness of the preparation tools is critical. Carbon tape, aluminium stubs and diamond blades all contribute their own peaks. Silicon from polishing media and chlorine from tap water appear regularly in novice spectra and are misread as real board contamination. Keeping a blank reference sample and a log of every material that touched the specimen lets an analyst subtract the preparation signature with confidence.

Elemental Mapping Across a Board

A single spot spectrum answers what is here, while an elemental map answers where it is. In mapping mode the beam rasterises across a chosen area and the detector records an X-ray count for each element at every pixel. The result is a set of coloured overlays that show, for example, tin concentrated on the joint, copper along the trace, and chlorine clustered at a corrosion site.

Maps are persuasive in reports because they are visual, but they need care. Long acquisition times improve statistics and also increase the risk of beam damage to polymers and thin organic films. A practical compromise is to map at moderate resolution first, identify the regions worth quantifying, then return to those points for longer spot acquisitions.

Reading Spectra: Peaks, Overlaps and Artifacts

Spectra are not self-explanatory. Peak overlap is common: sulphur and molybdenum, or antimony and tin, can sit close enough to confuse an untrained reader. Sum peaks appear when two X-ray quanta reach the detector almost simultaneously, and escape peaks sit a fixed distance below a major line. An operator who has never seen these before will confidently report elements that are not present in the sample.

Accelerating voltage also changes the picture. At low voltage the interaction volume shrinks and surface sensitivity improves, which is useful for thin films. At high voltage more of the bulk is excited, which suits thick plated layers. Choosing a voltage that matches the question is one of the few genuinely skilled decisions in routine elemental work, and it is worth documenting the choice for whoever repeats the test later.

Contamination and Corrosion Casework

Most field returns that reach an elemental laboratory fall into two families: contamination and corrosion. Contamination shows up as unexpected elements — chlorine from handling, bromine from flame retardants, sodium from fingerprints, or silicon from release films. Corrosion shows up as element pairs that have moved together: copper with oxygen, silver with sulphur, aluminium with chlorine.

The pattern matters more than any single reading. Finding chlorine on a corroded trace is only half the story; the useful question is whether the chlorine is localised at the failure or spread across the assembly. A map that shows a tight cluster points to a specific handling or cleaning event, while a uniform distribution points to a material or process source upstream.

Correlating Elemental Data with Cross Sections

Elemental data becomes far more powerful when it is tied to physical structure. That is why experienced labs combine an image of a polished cross section with spectra taken from defined points along it. A bright band between copper and solder can be measured and identified as a copper-tin intermetallic rather than guessed at from contrast alone. Practical plating thickness checks follow the same logic.

The approach resolves plating disputes efficiently. Measuring copper thickness in the barrel and comparing it with the drawing is much stronger when the measurement is anchored to an image that shows the barrel geometry. Combined with a review of fabrication records, this kind of evidence usually closes a supplier argument in a single meeting instead of three rounds of email.

Reporting Results Engineers Can Act On

A report that contains only spectra is a curiosity; a report that connects a spectrum to a process step is a tool. The best failure analysis documents state the observed feature, the identified elements, the most probable mechanism, and the process variables that could produce it. Each conclusion should point at a specific image and a specific acquisition point so a reviewer can retrace the reasoning. Where the failure sits near a plated hole, the geometry of the barrel belongs in the same figure as the spectrum.

Keep the language calibrated. Because the method is semi-quantitative, conclusions should be phrased as supported or not supported by the data rather than as absolute proof. Where the evidence is ambiguous, say so and propose the next test — a cross section, an ion chromatography check, or a controlled build — instead of forcing a verdict the data cannot carry. Reviewers and customers both respond better to a bounded claim that survives scrutiny than to an overreach that collapses on inspection.

FAQ

Can EDX analysis detect organic residues? Not directly. The detector measures X-rays from elements, and organic molecules are mostly carbon, oxygen and hydrogen, which are poorly resolved or effectively invisible. In practice EDX analysis identifies inorganic contamination such as chlorine, sulphur, sodium or bromine, while organic residues are better characterised by FTIR or ion chromatography. The two families of technique are often used together when a flux or cleaning problem is suspected.

How small a feature can be analysed? Modern detectors handle features down to about one micrometre, and the interaction volume beneath the surface is typically one to three micrometres across depending on accelerating voltage and material density. Anything smaller than that will be diluted by surrounding material, which is why very thin surface films are best examined at low voltage or with a dedicated thin-window detector.

Is the analysis destructive? The technique itself is generally non-destructive on conductive, vacuum-stable samples, though the electron beam can darken or crack polymers and can deposit carbon under prolonged exposure. The sample preparation needed to expose the feature of interest — cross sectioning, cutting or residue lifting — is usually the destructive step, not the measurement. For quality disputes, that distinction is worth stating explicitly in the report.