Cross-Section Analysis
How a Microsection Reveals Hidden Defects
Most plating and soldering defects live inside a feature where no surface inspection can find them. A barrel crack, a thin wall on one side of a plated hole, a resin-recession crack, or a starved fillet hidden behind a component body all sit below the surface. Cross-section analysis solves this by physically cutting through the feature of interest, mounting the cut sample in clear acrylic or epoxy, and progressively grinding and polishing until the inspection plane reaches the center of the feature. The result is a mirror-bright face that, under reflected-light microscopy at 100x to 400x, shows the true internal structure with micron-level measurement resolution.
The procedure is standardized in IPC-TM-650 method 2.1.1 for boards and method 2.1.1.2 for evaluating plated holes. A typical sequence is rough cut on a precision saw 2 to 3 mm away from the target, vacuum or pressure encapsulation to fill voids and prevent edge rounding, coarse grind through 240 and 320 grit silicon carbide, fine grind through 600 and 1200 grit, then polish with 6, 3, and 1 micron diamond. Stopping within 25 microns of the centerline is critical: a hole barrel measured off-center always reads thicker than the true wall, so a grinder watches the hole diameter widen to its maximum as the cue that the plane is centered.
Once polished, the sample is often etched to bring out features that a bright polish hides. A microetch separates copper grain structure, distinguishes electroless from electrolytic copper, and reveals the scalloped intermetallic compounds at a solder-to-copper interface. The same mount supports both optical microscopy and, where finer resolution is needed, scanning electron microscopy with energy-dispersive spectroscopy to confirm layer composition.
Plating and Intermetallic Measurements
tavg = ( ∑ ti ) / n (n ≥ 3 readings around the barrel)
Off-center magnification error (apparent vs. true wall):
tapparent ≈ ttrue / cos(θ), θ = angular offset from hole center
Intermetallic growth with aging:
x(t) ≈ x0 + k × √t, k ≈ 0.02 to 0.1 μm / √hour at 100 °C
Where ti = individual wall measurements, θ = how far the polished plane sits from true center, x(t) = Cu6Sn5 thickness, x0 = as-reflowed thickness (≈ 1 μm), and t = thermal aging time. A reflow joint shows 1 to 3 μm intermetallic; beyond 5 μm the joint embrittles.
IPC Acceptance Limits Read from a Cross-Section
| Measured Feature | Class 1 | Class 2 | Class 3 | Why It Matters |
|---|---|---|---|---|
| PTH avg copper wall | ≥ 20 μm | ≥ 20 μm | ≥ 25 μm | Thermal-cycle barrel fatigue |
| PTH thin-area copper | n/a | ≥ 18 μm | ≥ 20 μm | Local crack initiation site |
| Wrap plating at knee | ≥ 5 μm | ≥ 5 μm | ≥ 5 μm | Foil-to-plating continuity |
| Solder fill (through-hole) | ≥ 50% | ≥ 75% | ≥ 75% | Joint strength and conduction |
| Intermetallic (Cu6Sn5) | < 5 μm | < 5 μm | 1 to 3 μm target | Joint embrittlement |
| BGA ball voiding | ≤ 30% | ≤ 25% | ≤ 25% | Current density, thermal path |
Frequently Asked Questions
How thick must the cross-section sample be ground before final polish?
IPC-TM-650 method 2.1.1 requires grinding to stop within 25 μm of the target plane before fine polishing. A typical sequence is 240 to 320 grit to reach about 0.1 mm short of target, then 600 and 1200 grit, then 6, 3, and 1 μm diamond. Watch the hole diameter widen to its maximum: that confirms the plane is centered, since an off-center barrel reads falsely thick.
What plated through-hole copper thickness does IPC Class 3 require?
IPC-6012 Class 3 requires ≥ 25 μm (0.001 in) average copper wall with a 20 μm thin-area minimum, read directly from the microsection at 200x to 400x. Class 2 allows 20 μm average. Class 3 also demands at least 5 μm of wrap copper over the foil at the knee and caps allowable etchback, all of which the polished section reveals.
How is intermetallic compound thickness measured in a solder joint cross-section?
After polishing, the interface is lightly etched (2% nital or 5% HCl-alcohol for tin-lead) to reveal the Cu6Sn5 and Cu3Sn layers. At 1000x or on an SEM, the scalloped layer is measured at several points and averaged. A healthy reflow joint shows 1 to 3 μm; thickening toward 5 μm or a Cu3Sn layer over 1.5 μm with Kirkendall voids signals over-aging and embrittlement.