Contact Us
  • Home
  • BLOG
  • Why Do Plated Hole Barrels Crack After Thermal Cycling?

Why Do Plated Hole Barrels Crack After Thermal Cycling?

The barrel of a plated hole cracks because the laminate around it grows through the board thickness three to four times faster than the copper inside the hole, and the copper that has to absorb that movement is the thinnest plating on the board. Copper expands at about 17 ppm per C. FR-4 expands at 50 to 70 ppm per C below Tg and at 200 to 300 ppm per C above it, so most of the movement happens in the last 70 C before the solder melts.

The arithmetic is simple enough to do on the stack you are buying. Take a 1.6 mm board, Tg 170, heated from 25 C to 240 C. The resin stack grows about 42 um, the copper in the hole grows about 6 um, and the barrel is asked to stretch the 36 um difference over its own length. That is 2.3 percent strain. Plated copper yields at roughly 0.1 percent, so every reflow pass pushes the barrel past yield, and the barrel does not fail from one big push. It fails because the same plastic strain is repeated until work hardening has taken the ductility out of the copper.

This is why two boards from the same stack, with the same drill file, behave differently in the field. One has 25 um of ductile copper in the hole and survives a thousand cycles at minus 40 to plus 125 C. The other has 18 um, a nick at the knee and a void in the middle of the barrel, and it opens after three reflow passes and a few hundred field cycles. The crack is not in the laminate. It is in the copper, and the laminate only supplies the displacement.

Where the crack starts

Three places, in this order of frequency. The knee, where the barrel meets the internal land, because the pad is anchored in resin and the barrel is free, so the whole displacement is concentrated into a few tenths of a millimetre. The middle of the barrel on a thick board, because acid copper throws less into the centre of a deep hole, so that is where the plating is thinnest. And the corner where the barrel meets the surface pad, which is also the point where a press-fit pin or a rework iron adds its own load on top of the thermal load.

On a polished cross section the crack reads as a hairline at 200x that runs from the inside of the copper outward. At 50x the same barrel looks continuous. Any acceptance decision made from a photograph of the hole rather than from a section is a guess, and it usually guesses in favour of the supplier.

What IPC-6012 asks for, and what the class actually buys

The copper minimum is set by class. Class 2 requires an average of 20 um in the hole, 18 um allowed in thin areas. Class 3 requires 25 um average, 20 um minimum, with 12 um where the plating wraps onto the surface land. Cracks in the barrel and at the corner are not allowed in Class 2. Class 3 keeps a narrow allowance for corner cracks, defined in Table 3-10 of the standard, and that table is routinely read as permission for any small crack, which it is not.

Below about 20 um the fatigue life falls faster than the thickness does, because the strain per cycle is fixed by the laminate and the thinner copper simply reaches its limit sooner. Going from 20 um to 25 um is a 25 percent thickness increase and a much larger increase in cycles to failure, which is the most direct way to buy cycles on a high layer count board.

Aspect ratio decides how much copper lands in the middle

Plating thickness is not uniform down a hole. The electrolyte reaches the mouth of the hole more easily than the middle, so the mouth is thick, the middle is thin, and the difference grows with aspect ratio. A 1.6 mm board with a 0.25 mm hole is 6 to 1 and plates evenly. A 3.2 mm board with a 0.15 mm hole is 21 to 1, sits on a high layer count line, needs longer plating time and a tighter bath, and still gives a middle of the barrel that is thinner than the mouth.

That is the number to put on the drawing next to the plating callout. A note that says 25 um without saying where it is measured has produced a million arguments and no reliability improvement.

Three process choices that make it worse

Drill and desmear quality. A dull bit tears resin away from the glass bundle instead of cutting it, desmear then etches an uneven surface, and the plating follows that surface. A nick or a resin smear at the knee is a crack starter that no amount of extra plating will close, because the copper there is not bonded to anything.

Thermal excursions above Tg. Every pass above Tg spends fatigue life, and a 20-layer board takes several press cycles before it ever reaches assembly. Six reflow passes on a thick backplane are not the same as two, and the difference shows up as corner cracks on the first panels of a build rather than in the field.

Press-fit and rework in a thin barrel. A press-fit pin loads the barrel in hoop tension, and a 0.6 mm hole with 20 um of copper is not the place to save plating time. Desoldering a through-hole part with a vacuum iron pulls solder out of the barrel and can lift the plating at the knee, which is a crack that appears on the next thermal cycle, not during the rework.

What SUPERB controls on the panel

Hole wall copper is held at 20 um and above, and the plating thickness is reported from a cross section rather than from the bath log. Aspect ratio is 10 to 1 on the standard line and up to 25 to 1 on the high layer count line, mechanical drilling starts at 0.15 mm with hole position at plus or minus 0.075 mm, Tg 130, 150 and 170 laminates are stocked, back-drill stub is held at 0.25 mm or less, and panels ship with a section report so the knee and the middle of the barrel can be looked at instead of assumed.

The honest boundary: a cycling qualification at minus 40 to plus 125 C for 1000 cycles belongs in the customer reliability plan, and we supply the coupons and the section for it. What we control is the copper, the drill and the laminate that decide whether that qualification passes. Limits are on high reliability PCB, the full window on PCB capability, and the laminate data on FR-4 process parameters.

Need a quote or a review of your stack against these limits? pcba@superb-tech.com · WhatsApp +86 133 9608 1443