Microsection Analysis of a PCB: What a Cross Section Proves

When a board fails and the cause is not on the surface, the cross section is what settles the argument. A microsection takes several hours to prepare and destroys the sample, which is why it is often deferred, and why the deferred question is usually the one that mattered.

What a Microsection Actually Shows

A cross section is a polished slice through a region of interest, examined under a microscope at magnifications from fifty to a thousand times. It reveals layer count and sequence, dielectric thickness at the point of the slice, copper thickness on the surface and in the holes, hole wall plating quality, and the interface between copper and resin.

It does not reveal the whole board. The slice is perhaps thirty micrometres thick out of a panel measured in hundreds of millimetres, so the section proves what is true at the section line and nothing about the rest of the panel. Reading more into it than that is the most common mistake made with the technique.

Understanding that limit is what makes the section useful, because the correct question is never whether the board is good but whether a specific suspected defect is present at a specific location.

Choosing the Right Location

A section is only as good as its position. For a hole quality question, the cut should pass through the centre of the hole and perpendicular to the grain of the panel, so that the plating on both sides is cut at the same angle. A section that clips the edge of a barrel shows a misleadingly thick or thin wall.

For a registration question, cutting through the corner-most via and the adjacent inner layer pad shows the worst case rather than the typical case. A section through the centre of the panel will look acceptable even when the corners are at the limit.

For a delamination or blister question, the location is chosen by the visible defect, and the section should include both the defect and its boundary, because the boundary carries the information about which interface separated.

Reading the Plating on a Hole Wall

Copper on the hole wall should be continuous, uniform and of the specified thickness from the surface to the inner layer connection and through to the other side. Thickening at the surface and thinning toward the middle is a signature of throwing power, and it means the middle of the barrel is the weakest point rather than the average.

Our hole copper notes describe the distribution that plating processes typically produce. The measurement should be taken at the thinnest point rather than at a convenient point, since the specification applies to the minimum.

Look also at the inner layer connection, where the copper of the barrel meets the copper of the inner layer pad. A connection that shows a wedge, a crack or a line of separation at the interface is the defect that produces an intermittent open after thermal cycling, and it is invisible to any electrical test performed before the stress.

Dielectric Thickness and Layer Registration

The section gives the dielectric thickness between each pair of copper layers at the cut, which can be compared against the stack-up the design assumed. A difference matters most for impedance-controlled layers, where a thinner dielectric raises capacitance and lowers impedance.

Registration is read by comparing the position of each inner layer pad relative to the hole. The measurement that matters is the remaining annular ring on the side where the offset is greatest, since that is the side that will fail first under thermal stress.

Layer registration also has a direction. A pattern that is offset uniformly indicates a tooling or scaling issue, while a pattern that is rotated indicates a registration problem in a different axis, and the two have different remedies.

Copper Thickness and Its Measurement

Copper thickness on the cross section is measured in the plane of the slice, so a trace with a trapezoidal cross section has a thickness at its base that is greater than at its top. Which value is compared against the specification should be stated, because the difference can exceed twenty per cent on a heavily etched layer.

Surface copper includes the plating added during processing, so a section through a finished board reports the sum of the base foil and the plating. Comparing that number against the foil weight ordered produces a disagreement that is really a definitional difference.

Our plating thickness notes set out where each measurement is normally taken and which value is quoted. Agreeing the convention before the section is prepared avoids a debate afterward about what the measurement means.

Surface Finish and Soldermask at the Interface

The section shows the surface finish as a thin layer over the copper, and its thickness and continuity can be judged where the magnification allows. A finish that is present on the pad and absent on the trace beside it suggests a process issue rather than a design one.

Soldermask appears as a distinct layer with a defined edge, and the section shows how far the mask encroaches over the pad and how well it is cured at the interface with the copper. An undercut void beneath the mask edge is a place where process chemistry can be trapped and later released.

The section cannot judge solderability, which is a surface property rather than a cross sectional one. Our judging PCB quality notes cover the surface checks that complement the section.

Preparation Artefacts That Mislead

Grinding and polishing can smear copper across a gap, close a crack with debris, or round the edge of a soft layer. A crack that appears after polishing but not before is a preparation artefact, and it is the reason a section is interpreted by someone who prepares sections routinely.

Plating can also be pulled from the wall during grinding if the mounting resin is harder than the copper, producing a gap that looks like a void. Comparing the same feature at two magnifications is a quick way to distinguish a real defect from a preparation one.

Where a result is surprising and consequential, the standard practice is to prepare a second section at the same location. A defect that appears in one section and not in its neighbour is either a local defect or an artefact, and the second section is the cheapest way to tell which.

Coupons and the Aftermath of Testing

The most informative sections are usually of coupons that have been stressed first: thermal cycling for hole reliability, reflow simulation for delamination, or mechanical bending for flexible circuits. A coupon that has been through the same stress as the product shows the failure mode rather than the as-built condition.

Our test coupon notes describe the structures worth including. The section then shows which interface failed and at what number of cycles, which is the input a reliability model needs.

Keeping the section photomicrographs with the lot record turns the exercise into data. A stack of images from successive lots shows a drift in plating distribution long before it produces a failure.

Deciding Whether to Cut the Board

A section destroys the sample, so the decision should be taken deliberately. If a spare board or a coupon from the same panel exists, cut that. If not, accept that the section answers the question at the cost of one board from the lot.

The alternative to a section is a non-destructive examination, of which the most useful for holes is a thermal stress followed by an electrical continuity check on a daisy chain. Where the design includes such a chain, the stress test replaces the section for production monitoring and the section is reserved for analysing a failure.

When the question is about corrosion, contamination or a surface condition, no section will help. The correct tools there are ion chromatography, energy dispersive spectroscopy and optical inspection, and ordering a section instead wastes both time and the sample.

FAQ

How long does a microsection take? Several hours for a mounted, ground and polished sample with photomicrographs, and longer where a second section is needed. It is not a technique that answers a question the same morning.

Can a section prove that a board is good? No. It proves that the material along one line through one sample is as expected, which is useful evidence and not a guarantee about the lot.

What does gopcb provide for a failure investigation? We provide mounted and polished sections with photomicrographs at several magnifications, measurements of plating and dielectric thickness, and the coupon and lot records from the same panel. Where the failure is thermal, we run the cycling and report the cycle count at which the first crack appears.