Microsection Preparation: Cutting, Mounting and Measuring Plated Barrels

A microsection is the only check that shows what is inside a plated hole, and almost every error in the answer comes from the way the sample was made rather than from the way it was measured. Cutting, mounting and polishing each introduce their own artefacts, and a section that has been prepared badly looks exactly like a board that was plated badly.

What the Section Is Asked to Show

The usual questions are the copper thickness on the barrel wall, the presence and size of voids in the plating, the fill of a soldered barrel, the thickness of the intermetallic layer and the condition of the laminate around the hole. Each of those is measured at a defined magnification and quoted against a specification figure.

Because the measurement is destructive and slow, the section is used where a non-destructive check cannot answer the question at all. Our hole copper notes describe which of these figures can be estimated from a coupon or an X-ray measurement first, and which need the section.

Choosing the Section Plane

The plane has to pass through the axis of the barrel, because a section taken off axis shows a wall that appears thicker than it is and a void that appears smaller. For a grid of holes, the sample is cut so that the plane runs through the centres of a column of barrels, which lets several holes be measured in one mount.

The position is marked before cutting from an X-ray image or from the drill file, and the deviation from the intended plane is kept inside about 25 micrometres. Beyond that, the apparent wall thickness changes by a measurable amount, and the error is larger than the tolerance on the plating itself on a small hole.

Cutting Without Damaging the Barrel

Cutting is done on a precision saw with a diamond or abrasive wheel and with continuous coolant, at a feed rate slow enough that the sample stays cool. Heat is the enemy here, because the resin softens and the copper smears across the hole, producing a wall that looks continuous and a void that has disappeared.

The cut is also made at a distance from the region of interest and finished with a second, finer cut or with grinding, so that the wheel never passes directly through the barrel being measured. A sample that has been cut through the hole can still be mounted, but the last few micrometres of material are gone.

Mounting and Edge Support

The cut sample is mounted in a cold setting resin and cured under vacuum, so that the resin fills the holes and supports the copper at the edge of the mount. Resin that does not penetrate a barrel leaves an unsupported wall that rounds over during polishing, which is then measured as a thin or a broken plating.

Cure time matters as much as vacuum, because a resin that is still soft allows the sample to move in the mount. Cold mounting resins are usually left for at least thirty minutes at room temperature, and the mount is labelled at this stage with the sample identity, since it cannot be marked afterwards.

Grinding and Polishing Sequence

The sequence runs from coarse to fine silicon carbide paper, commonly 320, 600, 800 and 1200 grit, followed by diamond suspensions at 3, 1 and 0.25 micrometres on a rotating platen. Each step removes the damage introduced by the previous one, and skipping a step leaves scratches that hide the feature being measured.

Water lubrication is used for the papers and an alcohol based lubricant for the diamond stages, because water attacks some laminates and swells the resin at the interface. The sample is washed and dried between steps, and the polishing direction is rotated ninety degrees at each stage so that the remaining scratches show which step was last.

Artefacts and How They Are Recognised

The common artefacts are copper smearing across a void, pull-out of a plated wall, rounding of the mount edge and relief polishing where the soft resin wears faster than the copper. Smearing and pull-out both make a defect disappear, while relief polishing makes a good wall look stepped and can be mistaken for wicking.

Recognising them is a matter of looking at the whole field rather than at the feature: a smear usually has a direction, a pull-out leaves a jagged edge, and relief appears on both sides of the hole. Our solder joint acceptance criteria describe the same discipline for solder features, where the mount is prepared in the same way.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/Best-PCB-Manufacturers-for-Complex-Boards-1536×1024-1.jpeg" alt="Polished cross section of plated through holes at high magnification” />

Etching and Intermetallic Measurement

A light etch with an ammonium hydroxide and hydrogen peroxide mixture, applied for five to ten seconds, brings out the grain structure of the copper and the boundary between the plating and the laminate. Etching too long removes the very interface that is being measured, and the section then cannot be used for thickness.

Intermetallic layers at a solder interface are measured at higher magnification, typically 500 to 1000 times, and a layer between one and four micrometres is normal for a lead free joint. A layer that is much thicker indicates an extended time at temperature rather than a plating problem.

Measurement, Calibration and Acceptance

Thickness is measured with a calibrated image analysis system or a graduated eyepiece, and the calibration is verified with a stage micrometer at the start of the session. Copper on a barrel wall is commonly specified as a minimum average of 25 micrometres with no single reading below 20 micrometres for the higher classes.

Each figure is measured at three or more points around the barrel because plating is never uniform, and the minimum is reported alongside the average. Our plating thickness notes set out the same reporting rule for surface finishes, where the averaging is done over a defined area.

Documentation and Sampling

The record for a section carries the sample identity, the position of the plane, the magnification, the calibration check and the measured values with the photographs. Without the magnification, a photograph of a void cannot be interpreted, because the same image could be a 25 micrometre void or a 250 micrometre one.

Sampling follows the process rather than the panel: a section per plating line per shift during qualification, and a section per lot afterwards or whenever a change is made to the chemistry or the equipment. The frequency is written into the control plan so that it does not depend on someone remembering.

The usual questions are the copper thickness on the plated barrel wall, the size and distribution of voiding in the plating, the fill of a soldered barrel, the thickness of the intermetallic layer and the condition of the laminate around the hole. Each of those is measured at a defined magnification and quoted against a specification figure.

The cross section has to pass through the axis of the barrel, because a plane taken off axis shows a wall that appears thicker than it is and a void that appears smaller. For a grid of holes, the sample is cut so that the plane runs through the centres of a column of barrels, which lets several holes be measured in one mount.

FAQ

Can a microsection be taken from a production board? It can, and it is the most direct evidence available. The cost is the board itself, which is why production sections are usually taken from a coupon or from a deliberately scrapped panel rather than from a board that has already passed assembly.

Why does a void sometimes disappear between two sections? Because a void has a three dimensional shape and the plane may pass through a different part of it. Two sections from the same barrel can show different void areas, which is why the reporting quotes the largest void seen rather than an average over one plane.

How long does a good preparation take? Cutting and mounting together take about an hour with cure, and grinding and polishing another thirty minutes. Shortening the cure or skipping a polishing step saves minutes and produces a sample whose measurements cannot be trusted on a 25 micrometre feature.