1. Heavy copper does not change the pressure, it changes how far the resin has to travel
On a standard multilayer, inner copper is thin. After etching, the circuitry is close to a pattern sitting flat on the laminate surface. During lamination the prepreg resin only has to level out that small step. Flow distance is short, resistance is low, and a fill failure would be a surprise.
Replace that inner copper with heavy copper and the pattern stands up into walls. The gaps between traces stop being shallow depressions and become deep, narrow channels — and the resin has to flow into every channel and fill it completely, inside the limited window between softening and full cure. SKPCB's heavy copper capability runs to more than 12 oz, with 16 oz and 18 oz already built in production (frozen capability wording), on top of multilayer capability at 40 layers in standard production and up to 64 layers. The thicker the copper and the higher the layer count, the longer the resin's path and the deeper the channels it has to fill.
So heavy copper lamination fails in pairs:
- Not enough resin, or resin that cannot reach → gaps left at the channel floor, starved trace sidewalls. This shows up as voids and incomplete fill.
- Resin arrives but stress is uneven → the thermal expansion mismatch between heavy copper and resin gets locked into the board at cure. This shows up as delamination, local expansion/shrinkage mismatch, and secondary cracking the next time the board is heated.
These are not two severity levels of one problem. They need different detection methods and different acceptance criteria. That is the fundamental reason heavy copper lamination cannot inherit standard multilayer acceptance terms.
2. First, separate three defect names that get used interchangeably
In engineering conversation these three terms are routinely substituted for one another, but their causes, locations and detection methods all differ. Agreeing on terminology is a precondition for agreeing on criteria.
1. Void
Resin has not filled a space it was supposed to fill, leaving a closed or partly closed gap. On heavy copper the typical locations are the floor of the channels between thick inner traces and the junction where a trace sidewall meets the adjacent layer. A void does double damage: it is a weak point in dielectric strength, and it is somewhere for moisture to sit — which then vaporises under reflow heat and becomes the starting point for delamination.
2. Resin starvation / incomplete fill
Resin arrived, but not in a quantity sufficient to form a continuous, dense dielectric layer. The difference from a void: a void is "one empty spot", starvation is "thin and loosely packed throughout". Starvation is harder to spot visually than a void because it has no defined cavity boundary. Usually it can only be confirmed by looking at the resin-to-glass distribution on a microsection.
3. Delamination
The bond between layers fails and the interface opens. On heavy copper, delamination mostly does not crack open during lamination itself. Lamination buries the stress; later thermal processes (HASL, reflow, rework heating) release it. That is the awkward part: first inspection after lamination can pass cleanly, and the problem surfaces on the assembly line.
With those three separated, the inspection methods in the next section have something to map onto — because no single method detects all three.
3. Which inspection methods actually confirm heavy copper lamination
The key point: a press log only proves the cycle ran as programmed. It does not prove the inside of the board filled. Internal condition has to be read out by inspection. SKPCB's quality system accepts to IPC-6012 Class II and Class III, runs 10+ inspection gates, and performs 100% electrical test (flying probe or test fixture) on all orders. Heavy copper lamination judgements normally land on the following methods:
Microsection / cross-section analysis
The core judgement method for heavy copper lamination, with no substitute. It is the only way to directly see whether the channel floor filled, how resin and glass cloth are distributed, and whether the layer interfaces are continuous. The cost is that it is destructive — it samples, it does not screen. Which makes the agreed sample *location* more important than the sample count: it must be taken where the heavy copper circuitry is densest and the channels are deepest, not from an open area at the board edge. Prepare and examine per the IPC-TM-650 method family.
X-ray inspection
Non-destructive, good for registration and larger internal anomalies. But it is not sensitive to small voids at the channel floor — heavy copper absorbs X-rays strongly, and contrast in heavy copper areas gets suppressed. So X-ray can screen for big problems; it cannot replace microsection for fill judgement.
Ultrasonic / acoustic scanning
More sensitive to interfacial bond failure (delamination) than X-ray, because it looks for acoustic impedance discontinuities and a delaminated interface is exactly that. A good non-destructive cross-check in the delamination direction.
Thermal stress testing and reliability testing
Specifically there to detonate, early, the stress that lamination buried but has not yet expressed. The board is taken through a controlled thermal process, then re-inspected for delamination or cracking. SKPCB's quality wording includes per-batch reliability testing. For heavy copper multilayer this gate matters more than visual inspection — it targets exactly the "passes first inspection, fails on the assembly line" mode from section 2.
Visual and automated optical inspection (AOI)
Catches anomalies visible on the surface and edges (edge delamination, measling, outer-layer defects). It cannot see internal fill. Its role is upstream screening, not a lamination quality criterion.
100% electrical test
Catches continuity and insulation failures. If a void-driven dielectric weak point has already developed into an actual short or leakage path, electrical test will find it. A void that has not yet broken down is invisible to it. Which is why "electrical test passed" does not mean "lamination accepted".
4. What the design and ordering side can do to lower lamination risk
Lamination quality is not decided by the lamination step alone. A substantial part of it is fixed at design.
1. Keep copper distribution balanced; don't put all the heavy copper planes on one side
The more uneven the copper distribution within a layer, the larger the variation in the space resin has to fill, and the more asymmetric the post-cure stress distribution. Board warpage and interlayer stress both tie directly to this. For the systematic approach, see multilayer copper balance and warpage control.
2. Mixed copper weights are easier to get right than uniform heavy copper everywhere
Not every layer needs heavy copper. Using heavy copper only on the layers that carry current or heat, and standard weights elsewhere, materially reduces the resin fill difficulty. For how to mix a stackup, see heavy copper multilayer mixed weight stackup.
3. Don't push heavy copper spacing to the limit
Tighter spacing means narrower channels and harder resin entry. On heavy copper, spacing is a process feasibility question, not only an electrical one. Note too that heavy copper etch compensation shifts the finished trace width and spacing — see heavy copper etch compensation review.
4. State board thickness together with layer count, not layer count alone
SKPCB's thickness range is 0.2–6.0 mm. On heavy copper multilayer the copper volume itself pushes finished thickness up, so estimating finished thickness as "layer count × standard thickness" at design time often does not survive lamination. For thickness tolerance logic see PCB finished thickness tolerance control.
5. Choose material for the downstream thermal history, not just initial properties
The value of high-Tg and high-CTI materials on a heavy copper stackup is mostly in resisting later thermal shock. SKPCB's stocked high-reliability materials include Isola 370HR, ITEQ IT-180A/IT-158 and Shengyi S1000-2, with high-Tg >170°C and high-CTI >600V options available. Cross-checking a specific grade against the IPC-4101 specification system with your manufacturer is far more effective than writing "high Tg" on the drawing.
6. Leave margin on PTH aspect ratio
SKPCB's stated maximum is 10:1 for PTH. Because heavy copper multilayer boards tend to finish thicker, shrinking hole diameters touches that boundary quickly — and hole problems compound with lamination quality problems.
5. Put these items into the acceptance terms
The common dispute on heavy copper lamination is not "it was built badly", it is "nobody agreed what good was". Write these six into the technical agreement or PO attachment:
- Acceptance class: state IPC-6012 Class II or Class III explicitly. For industrial and power applications, Class III's interlayer requirements sit closer to the real risk. SKPCB builds to both.
- Microsection sample location: specify "taken from the densest heavy copper area with the smallest spacing", not just a sample count. Wrong location, and no number of sections proves fill quality.
- Whether re-inspection after thermal stress is required: the most worthwhile cost on a heavy copper board. First inspection after lamination alone will not find delayed delamination.
- Finished thickness and tolerance: give it alongside layer count and per-layer copper weights. Don't make the manufacturer back-calculate it.
- Material grade: name the grade rather than describing it as "high Tg", and state whether equivalent substitution is allowed.
- Defect terminology: define void, starvation and delamination as separated in section 2, so both sides stop using one word for different things.
For reference: SKPCB publishes a 99.8% outgoing pass rate (defined as good units shipped ÷ total units produced × 100% per production batch) and a scrap rate of under 0.3% (scrap quantity ÷ total units produced × 100%). Those are batch statistics, not an individual commitment on one heavy copper order — which is exactly why the six items above still need to be agreed line by line.
6. In one sentence
Heavy copper lamination quality is not determined by what the press was set to. It is determined by whether the copper-to-resin volume ratio left the resin space it could actually fill, and whether the right inspection method was used to read out the internal condition. Voids, starvation and delamination are three different failures needing three different judgements, and the delayed kind of delamination can only be caught by re-inspection after thermal stress. The most effective thing the design side can do is balance copper distribution, mix copper weights, and stay off the spacing limit.
Further reading
- Heavy copper multilayer mixed weight stackup
- Lamination void and delamination prevention
- Multilayer lamination registration control
- Multilayer copper balance and warpage control
- Capability pages: heavy copper PCB, multilayer PCB
- Quality: quality assurance
FAQ
Q1: Can X-ray find voids from heavy copper lamination?
Large anomalies yes, small voids at the channel floor unreliably. Heavy copper absorbs X-rays strongly, so contrast is suppressed in heavy copper areas — precisely the areas you most need to see. X-ray suits registration and gross-defect screening; fill judgement needs microsection.
Q2: Inspection passed completely after lamination. Why did it delaminate during assembly?
Because the thermal expansion mismatch between heavy copper and resin was locked into the board at lamination, but not yet released. HASL, reflow or rework heating provides the release condition, and the delamination then expresses itself. The defence is thermal stress testing before delivery to detonate it early, not stricter visual inspection after lamination.
Q3: Does thicker copper always mean more lamination problems?
Correlated, but not the only variable. What really sets the difficulty is the combination of copper weight, circuit density and spacing — heavy copper with generous spacing can be easier to build than medium copper at limit spacing. So give all three at quotation and DFM stage; copper weight alone is not enough to judge risk.
Q4: Uniform heavy copper or mixed copper weights — which is more robust?
Mixed is usually both more robust and cheaper. Heavy copper only on the layers carrying current or heat, standard weights elsewhere, reduces resin fill difficulty and uneven-copper stress at the same time. See heavy copper multilayer mixed weight stackup.
Q5: Microsection is destructive, so how is every board covered in volume production?
By a combination rather than one method: microsection validates the process window and batch representativeness, electrical test provides 100% screening (SKPCB runs it on all orders, flying probe or fixture), plus per-batch reliability testing and multiple in-line gates (10+ inspection gates). Microsection proves "built this way, the process is correct"; full screening and batch testing prove "this batch actually ran that way".
Q6: What is the minimum information needed to assess heavy copper lamination risk at order time?
Layer count, copper weight per layer (inner and outer distinguished), minimum trace width and spacing, finished thickness and tolerance, material grade or equivalence requirement, IPC acceptance class, and whether post-thermal-stress re-inspection is needed. Without per-layer copper weight and minimum spacing, the risk cannot be assessed.