PCB Riveting and Fusion Technology in Multilayer Lamination

Before a multilayer board is pressed, the inner layers have to be held in the right position relative to each other. If they move during the lamination cycle the board comes out with layer-to-layer misregistration, and misregistration on a dense board ends in open vias and shorted planes. The oldest way to prevent that movement is PCB riveting, and the way most boards are held today is prepreg fusion. Knowing how each method works explains why the newer one displaced the older one, and where the older one is still the right answer.

What Riveting Does

Riveting places a metal or plastic rivet through a registration hole in the inner layer stack and upsets it so the layers cannot slide. The rivets are added after the inner layers have been imaged, etched and inspected, and they stay in the board through pressing and drilling.

The method is well understood and needs no special lamination equipment, which is why it survived for so long on thick boards and on boards built from many thin inner layers. It has real disadvantages as well. Rivets cost money, and the cost grows with the number of them. The upset head creates an indentation in the outer surface of the stack, which becomes a visible defect if the board is thin. The rivet hole itself consumes area on every inner layer, and on a dense design that area is expensive. Tooling plates wear out faster when rivets are part of the stack, because the rivet heads press into the plate surface.

How Prepreg Fusion Replaces It

Prepreg fusion uses the material already in the stack. Prepreg arrives in the B-stage, partly cured and still thermoplastic at a temperature well below the full lamination temperature. A heated tool applied at defined points around the panel melts the prepreg locally and, when the tool is withdrawn, the resin cools back to a rigid state, bonding the inner layers at those points.

What that buys is a stack with no added parts. There is no rivet to buy, no rivet hole to reserve, and no rivet head to mark the surface. The bonded points are inside the prepreg, so the surface of the inner layer is not disturbed. The bond is strong enough to survive the rest of the multilayer lamination cycle, including handling and pressing, and the equipment needed is a heated tool, a positioner and a controller, which is cheaper and easier to operate than the accumulated cost of rivets and the slower riveting operation.

The Elements That Decide Whether Fusion Works

Four things govern the quality of the fusion step, and a shop that gets all four right sees the effect in registration yield rather than in a defect report.

The first is the accuracy of the positioning system. The fusion operation happens after the inner layers are stacked, so any error in the stacker is locked in by the bond. A positioning system that is stable, repeatable and free of backlash is a prerequisite; a system that drifts will produce bonds in the right place and a stack that is still misregistered.

The second is the design of the bonding point. The shape of the tool tip can be round, square or rectangular, and the area of the resulting bond matters more than the shape. A bond that is too small does not hold the layers firmly, so the stack can shift in the press. A bond that is too large forces resin out of the region, and that resin movement shows up later as a white spot, a weak bond between the layers, or a dry region that delaminates after thermal cycling.

The third is the flatness of the equipment. Fusion applies force as well as heat. If the tooling plate or the support is not flat, the pressure is uneven across the panel, the stack tilts, and the layer-to-layer registration that the fusion step was meant to protect is lost. Uneven pressure also bends the panel, which is the same failure mode that riveting produces when the rivets are too tight.

The fourth is control of temperature and time, and this is where the process window is narrow enough that it pays to characterise it rather than to use a nominal setting.

Bonding Geometry: Round Against Rectangular

Comparing a round bonding tool with a rectangular one shows the trade-off directly. The rectangular tool covers roughly twice the area of the round tool at a comparable size, and the measured bond strength is correspondingly higher: in the same stack positions the rectangular joints produced noticeably higher pull values than the round ones.

The same comparison shows the cost of the larger area. The rectangular tool forces considerably more resin to flow out of the bonded region than the round tool does. When the resin flow is excessive, material can be pushed up at the edge of the panel and the edge can be left under-pressed, which is the defect that appears as a soft edge or a delaminated border.

The choice therefore depends on the board rather than on a preference. On a small board with little space for bonding points, round joints may not hold the stack well enough, so rectangular joints are used and the position of each one is chosen so that resin flows inward rather than to the edge of the panel. On a large panel with generous space, round joints keep the resin flow low and the process clean.

Temperature and Time Windows

The fusion temperature controls how far the resin expands away from the tool. At a low temperature the fused area is uneven and cracks form, and the bond is unreliable. At a high temperature the resin expands too far, the polymer degrades, and the result is a bond region that is weak for a different reason. In practice, around 285°C for the same dwell and the same stack gives a fused area that spreads uniformly with no crack risk, while 270°C leaves the area uneven and 300°C spreads it excessively.

The dwell time behaves in the same way. A dwell that is too short leaves the fused region uneven and prone to cracking. A dwell that is too long keeps the resin flowing after the bonded area has already reached the right size, weakening the bond. Around 15 seconds at the same temperature gives a uniform fused area, while 12 seconds is short of the window and 18 seconds overshoots it.

The stack-up matters as much as the settings. With the same temperature and dwell, a thinner prepreg such as 2116 fuses into a uniform area with no cracking, while a thicker 7628 leaves a cracked fused region. The reason is thermal: a thinner prepreg conducts heat through the bond more evenly and needs less resin flow to close the joint, so thinner prepregs are the natural choice for the fusion step.

Registration Is the Point of Both Methods

Riveting and fusion exist to protect the same thing, and the process controls that support them are similar: stable tooling, controlled temperature and pressure, and an inspection step that catches a drifting stack before a full lot is pressed. Our multilayer PCB manufacturing flow treats the bonding step as a monitored process with its own settings per stack-up, not as an operator preference.

The inner layers themselves have to be right before either method can help. Registration budgets start with the artwork and the etch compensation, and a stack that is already at the tolerance limit cannot be rescued by bonding. That is why the review of the fabrication data before the order is released covers the layer count, the prepreg sequence and the registration targets together.

Choosing Between Them

Fusion has displaced riveting on the majority of multilayer work because it is cheaper, cleaner and easier to control, and because it leaves no artefact in the finished board. Riveting remains useful where the stack is very thick, where the prepreg construction does not fuse reliably, or where the shop already has the riveting tooling and the volumes are small.

Either way, the decision belongs with the fabricator, because it depends on the prepreg types available, the tooling on the line and the registration capability of the shop. What the designer controls is the space reserved for the bonding points on the inner layer artwork, and that space should be agreed with the fabricator at the start rather than added later. The design rules for those inner layers sit alongside the rest of the documentation, and our quality management process keeps the panel-level and the board-level checks in one record so a registration trend is visible before it becomes a yield loss.

FAQ

Does fusion work on any prepreg? Thinner prepregs fuse more reliably because the heat passes through them evenly. Thicker constructions need higher settings and are more likely to show cracked fusion areas.

Can rivets and fusion be used on the same board? Yes, and some thick stacks use both, with the fusion points holding the layer pairs and the rivets holding the assembled stack.

What is the sign of a fusion process that is out of control? A fused area that is uneven in size from point to point, or a rise in layer-to-layer misregistration after pressing, both point at the bonding step rather than at the drill.