I’m trying to gain a better understanding of multilayer PCB structures. I’m struggling with understanding the roles of “prepreg” and “core.” Although I know they are involved in bonding the layers together, I’m uncertain about why both are necessary and how they differ from each other. Could someone clarify their distinct functions and why they are both necessary?
Example: 8 layer Stack-up
Soldermask
Layer-1 SIG
Prepreg
Layer-2 GND
Prepreg
Layer-3 SIG
Core
Layer-4 GND
Prepreg
Layer-5 GND
Core
Layer-6 SIG
Prepreg
Layer-7 GND
Prepreg
Layer-8 SIG
Soldermask
The key distinction lies in their composition and manufacturing process.
The core consists of a layer of FR4 sandwiched between copper layers, crafted in a specialized core facility. This FR4 layer is formed between smooth copper foils to precise thickness specifications.
On the other hand, prepreg comprises uncured FR4 used by PCB manufacturers to bond etched cores or copper foils to etched cores. As a result, prepreg thickness varies depending on the height of the etched boards it connects.
In applications where the dielectric properties are crucial, such as high-frequency transmission lines and antennae, having signal and ground layers adjacent to a core yields superior repeatability compared to when the fields traverse prepreg.
@BerndKruger That was helpful, but I’m not sure I quite understood correctly. Is the following a fair re-wording?
Core and Pre-Preg as basically the same materials, except that Core has already been baked into a more rigid and permanent shape with consistent composition. Pre-Preg isn’t baked yet, so it will still mold itself around the layer it is attaching to. Because the epoxy/glue will move more than the glass, it won’t be quite as consistent, particularly right near the interface. Because the glass and epoxy have different electrical characteristics, this can matter for extremely precise designs, such as antenna or high speed. If the pre-preg is thick compared to the copper traces, it should be pretty similar after baking.
Is the copper thickness on the other side of a pre-preg also less consistent, either because of lower standards or because the pre-preg itself offers less physical resistance to encroachment?
The biggest problem is just that it flows. As the press heats up the prepreg gets soft and flows anywhere it can. Some mfg’s have problems with air getting trapped for example. Since it can flow, if there are large differences in the amount of copper different areas have that can be a problem. you can add pours or thieving to compensate.
A conventional 8 layer stackup without advanced construction (like buried vias or HDI) will normally have 3 cores with 4 prepreg layers alternating between the cores. An alternate 8 layer structure is to use 4 cores with 3 prepreg layers but note that requires more fabrication steps and generally costs more than a 3 core stucture.
• A core is fully cured while pre-preg is partially cured – referred to as “B” stage- and needs to see more heat to fully cure.
• FR4 is a thermal reactive resin. In the case of pre-preg, as the temperature starts to increase the viscosity drops allowing the resin to encapsulate/fill in around the traces on the inner layers. When the resin hits it’s Tg temperature it hardens and no longer flows. The tricky part in lamination is controlling the temperature rise so the resin flows evenly across the panel without keeping it below the Tg for too long so it runs out and causes resin starvation. Bring the temperature up too fast and it hardens around the edges trapping volatiles towards the center and causes delamination.
• Pre-preg comes in a wide range of glass styles and resin content. This is important for maintaining dielectric thickness as a signal layer requires more resin to fill around the traces than a plane layer and can press out thinner than desired.
• Cores come in a limited range of glass styles so it’s common for us to make cores in-house for special applications.
• Vacuum lamination is pretty much an industry standard as it eliminates air and volatile entrapment. The resin contains solvents/volatiles which out-gas during lamination so a bigger problem than air. The press cycle usually includes a “Bump” where platen pressure is released for a moment and the volatiles are removed.
And a couple more subtle “Pre-Preg versus Core” differences. Cores generally have quite a range of thicknesses, from more than 0.062" down to maybe 0.005" This allows you to design a strong board; while the thinner prepreg sheets (1 to 5 mils or so) allow you to tweak the thickness/stackup, achieving the desired thickness and impedance. A one mil sheet of prepreg allows you some “tuning” of the impedance. Lastly, it’s generally a good idea to match pre-pregs to cores. If you’re using Rogers 9999x core for example, there is usually a corresponding pre-preg listed for use along with it. All this in addition to everything listed above. A great PCB is always a Goldilocks recipe, “a little of this and not too much of that”!
When considering PCB prepreg vs core, it’s important to note their unique characteristics in multilayer PCB construction. A core is a thicker, more rigid layer that provides the primary structural support of the board. In contrast, prepreg is a thinner, uncured fiberglass material used to bond layers together during the lamination process.
The distinction between core vs prepreg is also significant when considering via placement. Vias can pass through a core and any number of prepreg layers, but they cannot pass through multiple prepregs without also intersecting a core. Therefore, the stack-up configuration, including the placement of cores and prepregs, plays a critical role in the design, particularly when planning for high-density interconnects.
By carefully selecting the appropriate materials and understanding the roles of prepreg and core, you can optimize your PCB for both mechanical strength and electrical performance. Be sure to consult the specific stack-up requirements of your design to determine the best use of core and prepreg materials.
While both core and Prepreg materials are derived from similar fiberglass and resin compositions, they serve different purposes and are essential for different reasons. Core materials are fully cured and provide a rigid base for the PCB, which is vital for maintaining the structural integrity of the boards.
Prepreg, short for “pre-impregnated,” is a glass fiber weave impregnated with a resin bonding agent. It is not fully cured and is used to bond the various layers of the PCB together during the lamination process. When heat and pressure are applied, the prepreg melts and flows to fill gaps, solidifying to form a strong bond between layers. This capability to flow and conform during lamination is essential for achieving specific dielectric properties and layer spacing, which are crucial for impedance control and signal integrity in high-frequency applications. Prepreg is available in various thicknesses, allowing for flexibility in achieving specific board specifications.
The key difference in the prepreg vs core layers lies in their material composition and thickness. Prepreg layers have more resin and less glass fiber compared to core layers. However, the most significant factor is the thickness. For instance, if you’re routing a 100Ω impedance differential pair on a core layer with the ground on the opposite side, you might need a very wide trace—up to 100 mil—because the ground is 1.6mm away.
In contrast, on a prepreg-separated layer, where the ground is only 0.2mm away, you can achieve the same impedance with a much narrower trace, around 8 mil. This difference can greatly impact your design, particularly in high-speed or high-frequency applications.
I would have naively expected a single core, with three prepreg-then-foil layers on each side. Are the pre-pregs too unstable for that? Would each of the three need to be laminated separately, so that fabrication would get time-consuming and expensive?
What about internal prepreg-foil-prepreg layers, like a pair of cores, but instead of gluing them together with a single prepreg, use two pre-pregs with copper between them?
By the end of fabrication, is the pre-preg fully cured? Does it effectively become a low-precision core, or does it remain forever only partially cured?
It’s fully cured during the lamination cycle which occurs early in the manufacturing process. It becomes a dielectric layer but wouldn’t describe it as low-precision. Good controlled impedance results rely on accurate press-out thickness of the pre-preg. Just to clarify a core is made from pre-preg so it’s all basically the same stuff. The laminate manufactures just lay-up pre-preg and copper foil in large format presses then cut it up into whatever size the PCB shop requires. This eliminates a lot of material handling issues and helps keep cost down.
No, the minimum separation from a plane is more like 3 mils. But the 1 mil sheets can be used to control exact thicknesses and Class 3 boards are required to have at least 2 I think, so instead of one 2 mil you could use two 1mil to achieve that.
% mile cose could be used. The board wouldn’t have much strength. Ten mil can actually yield a useful board if you need one thin.
I assume Class 3 requires multiple pre-pregs because that is more reliable. What sort of failure are they trying to prevent?
Is it just a way to ensure there are enough different layers of glass, or is there something about either the separate manufacture or the joint between them that also matters?
Prepreg sheets, like most materials, are subject to pinholes. To insure electrical integrity, capacitor makers use two thin sheets of dielectric on their good capacitors because it makes the odds of lining up two random pinholes extremely high. I assume IPC was trying for a similar benefit.
In general, core material is more consistent than prepreg in terms of both thickness and dielectric constant. This is because core is manufactured in a controlled factory setting to a precise thickness.
When prepreg is used in the final PCB assembly—where the overall board must meet a specific thickness—the prepreg dielectric thickness can vary more due to the cumulative effect of variations in all the other layers. Prepreg also leads to a wider variation in dielectric constant, as the supplier controls only the raw prepreg material, not the final assembly process. This variation in thickness and dielectric constant can affect the performance of your board, particularly when working with high-speed signals.
For applications that require precise controlled impedance, such as high-frequency designs, it’s often better to rely on core material. However, for less critical logic signals, using prepreg for controlled impedance can still be acceptable.
In one of our designs, where we were dealing with 5 GHz microstrip transmission lines on a 1.6 mm thick board, we opted to use Rogers 4350 core for layers 1-2 and 9-10 for more accurate impedance control. The inner layers were FR4, where we allowed the fabricator to use a mix of core and prepreg that suited their manufacturing process.