FR-4: Used by many known by few

The non-conductive parts of a PCBs buildup, the laminate, is usually made of bundled fibreglass strands that are embedded in epoxy resin. The most common type of laminate grade is FR-4, Flame Retardant 4, that has properties most suitable for a wide range of PCB applications. FR-4 is by far the most common laminate, but it is far from the simple, uniformly-used, monolithic, well-behaved material we might be tempted to assume.

A cross section of a four-layer board. We can clearly see the glass fibre weave embedded in the resin.

FR-4 is in fact a category of materials that has been defined by NEMA from the 1960s to today; the relevant standard specifies FR-4’s performance requirements, which has evolved significantly over the years. Even without many PCB designers’ awareness, significant research and development is constantly invested in this, deceivingly basic, building block in order to meet the increasing performance demands of the electronics industry: chemistry and raw material improvements and, in particular, gains in reducing product variability.

Let’s look at those properties we’d like FR-4 to meet for us:

• Have consistent, uniform, and predictable properties even after several cycles of heating and cooling where the material softens, flows, and then hardens again;
• soften enough to fill tiny gaps in copper, but not too much to escape completely under pressure;
• create strong and lasting bonds, much like glue, between the layers of the buildup (that is, between laminates, copper, and soldermask); and
• be available as rigid cores and as very thin sheets and in varying flow properties, and with a long shelf-life.

It’s actually quite a lot to ask for from a material!

In order to manufacture reliable PCBs, the quality of raw materials must be good, and the manufacturing process specifically customised for the laminate used. Plus, the appropriate material properties must be chosen to account for the future use of the board, if, for example, it’ll go through many heat-cool cycles or operate in high-temperature environments. Below we will tell the qualitative story of FR-4 so that PCB designers are aware of potential trouble points, and be able to discuss them with their manufacturers in order to choose the best materials for their application.

Looking at the cross section of a four-layer board can reveal a lot of detail. We can see the individual glass fibres that make a bundle, and that these bundles make the weave. Then, one or more of these weaves are embedded in resin. During the buildup process of the board’s layers, copper is added to create the circuit.

How is FR-4 used?

FR-4 comes in two primary forms:

• Core: cured laminate with (one or) two sheets of copper on its surfaces; this is the base for making a two-layer PCB.
• Prepreg (short for pre-impregnated): partially cured epoxy resin laminate sheets; the curing and bonding happens with the heat during pressing.

Cores and prepregs come in many thicknesses, allowing manufacturers to create infinitely many buildups (Eurocircuits offers over 975 predefined buildups!). Now here’s a very simplified description of how PCB buildups, or stackups, are made.

For two-layer boards manufacturers use ready-made cores that go through chemical and mechanical processes to make up a finished PCB. For four or more layers the boards are created by stacking etched (and drilled if needed) cores with prepreg sheets (usually two) between otherwise adjoining copper layers.
This buildup is heated and pressed, causing the uncured resin in the prepregs to melt and squeeze into gaps in the copper and holes (like into blind- and buried vias). This creates adhesion that shouldn’t ever come apart under the material’s specified usage and conditions and lifetime. Finally, the stack then goes through the same chemical and mechanical process as a two-layer board.

The resulting stack can now be a building block for manufacturing a buildup with additional layers, stacked together with further pressings. More about this process is here.

You’ll sometimes hear of low-flow, or no-flow, prepreg that, as the name suggests, only flows to the extent that it bonds layers, but isn’t meant to fill voids. This type of prepreg doesn’t shrink (vertically) or expand (horizontally) much during the pressing of the stack. An example where this is useful is with flex and semi-flex boards where we’d like to maintain a mechanical boundary without trimming.

What may go wrong?

Of course it’s the responsibility of the manufacturer to make sure that their process is fine-tuned to the materials that they use in order to produce reliable boards.

It actually starts with logistics since prepreg has a limited shelf-life, after which it degrades, so manufacturers must make sure that the materials are ‘fresh’ and within specification limits. Then, the copper patterns in the individual design matter: more copper, fewer gaps, thicker layer; less copper, more gaps, thinner layer. Bear in mind that the resin must flow into all those tiny gaps in the copper – like between differential-pair tracks – in order for the board to function correctly.

Not getting all this right will cause problems: gaps that cause impedance mismatches, delamination of copper from the laminate, inconsistent hole plating, and more, as we’ll discuss next.

Some of the faults that can occur in FR-4: measling (gaps between bundles in intersection), crazing (gaps between glass fibres), blistering (‘larger’ gaps in the resin), and delamination (gaps that cause ‘wide’ gaps between the weaves). On the right are some gaps between the laminate and the copper around a through-via.

The IPC – in IPC-A-600 and elsewhere – tells us in some detail what may go wrong with FR-4, and what we might consider ‘acceptable’ for the three PCB Classes. Here are some of these faults, shown in the graphic above:

• Measling: separation of the fibres at the weave intersection
• Crazing: voids between glass fibres within the bundle (yarn)
• Delamination/blistering: separation in the base material, between individual weave layers (plies) or between the base material, copper foil, or coating (such as soldermask)

The IPC also talks about the exposure of the weave to the surface:

• Weave exposure: the glass fibres are exposed because of an absence of resin.
• Weave texture: the glass fibres are visible through a thin layer of resin.
• Exposed or disrupted fibres

There’s a limit on allowed metallic or non-metallic ‘foreign inclusions’ entrapped in an insulating material. And, limits on ‘haloing’, the roughness of the exposed surface after drilling and routing of the laminate.

That’s quite a few things that can go wrong! Let’s now discuss the properties of these materials and what they are subjected to in order to understand how these faults may occur.

On the left is an example of a pad that has ‘lifted’ from the laminate, and on the right is an example of a void caused by delamination.

Coefficient of thermal expansion, CTE

A PCB goes through several heating and cooling cycles; the exact number depends on buildup and on the application. For manufacturing there’s, roughly, a cycle for every pair of copper layers, and for each component soldering side. Then, a cycle may happen at rework or when a module is soldered on. And, of course the product may go through several or many cycles, depending on the application and its setting. The board must therefore survive all of these cycles, and the rest of its lifetime.

Even though the PCB appears to be monolithic, it is composed of several materials, each responding to temperature in a different way. One measure for this response is the coefficient of thermal expansion, CTE, which tells us by how much the material expands for every change in temperature unit, normally 1℃. CTE is usually expressed in ppm/℃: the expansion in parts-per-million of a unit length for every degree change in temperature. (Sometimes you’ll also see the total expansion expressed as a percentage over a range of temperatures, for example, 3.3% for 50-260℃).

Uniform materials, like copper, will have a single CTE as it expands at the same rate in all directions. A composite material, like FR-4, should have CTEs defined for the three x, y and z axes (CTEx, CTEy, CTEz, respectively). If the horizontal x- and y-axes weave pattern is the same, CTEx is the same as CTEy, otherwise they’ll be somewhat different; the z-axis, however, is quite a bit different than the horizontal CTEs since the reinforcing fibres run only in the x- and y axes. The CTE should also be spec’d for below and above the glass transition temperature, Tg (more on the later).

Cross sections of vias in a two layer board (left) and a multi-layer board. The bubbles in the via are suspended in resin which is added in the cross-sectioning process; more about that here.

Let’s quickly take a simple example of how to use a CTE value. If our laminate CTEx,y is specified as 17 ppm/℃ and our material area is 50x50mm, we’d expect the material to expand in the XY-axes at a rate of 50 / 106 x 17 = 0.00085 mm/℃, or 0.00085 x 200 = 0.17mm at 200℃.

Problems start happening when materials with different CTEs are bonded together and then heated. This makes intuitive sense: imagine bonding a paper card (low CTE) to a rubber band’s surface (high CTE) and then pulling on the band’s ends… the paper will crack and tear. This ‘tear’ can also happen both ‘inside’ the PCB and on the interface – the solder joints – between the board and components.

Copper has a low CTE compared to the CTEz of FR-4. The copper rivet within the FR-4 is therefore resisting the expansion and may crack.

Since there is no glass reinforcement in the z-axis the resin’s CTE is the primary expansion factor, one that has different values below and above the glass transition temperature Tg: CTEz can be as high as 70 ppm/℃ below Tg but may increase to over 250 ppm/℃ above Tg. This leads to a very large increase in expansion when Tg is exceeded, as it almost certainly does during soldering. A particular concern is with plated holes and vias that form copper rivets – annular rings on the top and bottom connected by the barrel – through the laminate. The CTE of copper is about 17 ppm/℃ whilst the CTEz are much higher (CTEx,y are normally made so their CTE is close to that of copper). This means that the laminate expands more than the copper as the temperature rises, potentially causing the barrel to crack or the pads lift from the PCB surface. This risk is increased with repeated cycles and the length of time the PCB is at certain temperatures.

Then, there’s a possibility of CTE mismatches between the horizontal expansion of the PCB (CTEx,y) and the components soldered onto it. These components are made of different materials that also expand at different rates; a BGA, for example, is likely to have a PCB inside of the package and the interface to the main PCB is through tiny solder balls. The CTE mismatches can cause shearing and cracks in the solder joints that may cause a disconnect or reduce the long-term reliability of the product.

What’s important to remember is that composite materials may have quite a few CTE specifications and designers need to make sure that they are considering the correct ones for their application.

Glass transition temperature, Tg

FR-4 and other types of laminates have a glass transition temperature, Tg, property: the temperature at which a polymer changes phase between a glassy-brittle state and a viscous-rubbery state. At temperatures higher than Tg there’s a significant increase in the CTE, and a subsequent increase in the volume occupied by the polymer. Most commonly used FR-4 will have a Tg of around 130-140℃ with materials becoming more expensive and exotic as the Tg spec increases (180℃ is considered ‘high-Tg’, but, for example, laminates in the Rogers RO4000® family can have Tg of 280℃).

Whilst CTEs above Tg are important mostly for the PCB manufacturer, Tg itself is an important parameter for the application of the PCB. It’s where the designer needs to make sure that the environment temperature – ambient, or local to a chip, for example – stays below it with some margin (there’s no fixed margin, each manufacturer will define it for a product). Otherwise, the PCB will lose its mechanical properties and will not behave as expected.

It is important to remember that Tg is not an indicator of thermal performance or endurance; the decomposition temperature (Td, described in detail below) is a more appropriate property for this.

Glass fibre weave

A single glass fibre weaves. Each bundle is made of several grass fibre strands and weaved together (one direction is called ‘weft’, and the other ‘weave’). Notice that the distance between, and ‘thickness’ of, the bundles is not necessarily the same in the x and y directions.

The glass fibres that make up the FR-4 aren’t always the same; in fact, there are many ways to weave an FR-4! You can vary the amount of fibre strands in a bundle, the thickness of each bundle (they can be ‘flattened’), the distance between yarns, and have different parameters for the x- and y-axes (weft and warp, respectively). Then, there’s how many weaves there are per unit thickness. This means that the glass-to-resin ratio of our particular FR-4 weave greatly affects CTE – resin has a higher CTE than glass – and Tg – resin is the part that softens – values.

There’s more. FR-4 is anisotropic: exhibits different properties in different directions (wood is an example of an anisotropic material, with and against its grain). So despite what we might assume, the relative permittivity / dielectric ‘constant’, εr / DK, isn’t uniform either (this is sometimes referred to as ‘micro-DK’); how could it be when each spot has a different ratio and density of glass and resin? This, of course, could impact the resulting characteristic impedance along tracks. For low frequency application this isn’t much of an issue, but as the frequency of signals increases the fibre-weave-effect (also related is the glass-weave skew) is worsened. Of course these issues are exacerbated when there are faults in the FR-4 that we discussed earlier – blistering, measling, crazing, etc.

Other properties

There are other specifications that FR-4 laminate manufacturers provide. Decomposition temperature, Td, is the temperature at which the material loses 5% of its mass after a controlled ramp-up in temperature. It’s a measure for degradation rate. Td is an important measurement for when assembling with a process that requires high(er) temperatures, like with lead-free.

As a measure of performance of the material time to delamination is used. It is normally specified as minutes at three temperatures, 260/288/300℃: T260/T288/T300. As the name suggests, it’s the time in each temperature after which there’s delamination, a separation of bonds within the laminate, and with copper that causes the problems we’ve discussed earlier.

FR-4 is sensitive to moisture that when absorbed changes its characteristics; for example, εr / DK increases and Tg decreases. Moisture can be absorbed during production and in the field, potentially also leading to many of the issues we’ve discussed so far. (Moisture is also a contributor to Conductive Anodic Filamentation, CAF, failures which is outside the scope of this article). An FR-4 datasheet should have a ‘moisture absorption’ figure as a per-cent (as defined in IPC-TM-650 2.6.2.1) and manufacturers and assemblers should make sure that their processes don’t cause faults. After manufacturing, it’s designers who should consider, and monitor, the effects of moisture on their product in terms of component assembly (for bare PCBs), shipping, and storage.

What does all this mean to a product designer?

As much as we’re tempted to think of FR-4 as a single-variety uniform material, it isn’t: there are many differences in parameters between all the types of FR-4 that are available for designers and manufacturers to choose from.

Luckily, good manufacturers deal with many of the potential issues that may arise during the manufacturing process, and will choose materials that will withstand it. This is because they are able to control and optimise the process to avoid them, and they are responsible to give you a working, reliable board. When any of the issues associated with the type or quality of the laminate are relevant to your product, consult with the manufacturer about which material to choose from the ones that are compatible with their process.

As we’ve seen, the trouble doesn’t end with the PCB manufacturing because there’s additional heat cycles happening during assembly and in the ‘field’. So a component assembler will also need to consider the properties of the FR-4 it will assemble components on. You might want to discuss the FR-4 used by your PCB manufacturer with the assembler to make sure that it’s compatible and can accommodate the heat cycles involved. Or, choose a service that does both PCB manufacturing and component assembly for you, so they control the entire process.

After all that, consider the operating and board-local temperatures of your board, rework, daughter card assembly. Are these temperatures approaching Tg (each laminate manufacturer can tell you the margin under Tg you must stay below)? Factor all of that into your choice of laminate and discuss that with your suppliers.

At Eurocircuits

The information about the materials that we use are always at our Downloads page. A convenient place to see which materials we’ll be using for your buildup is at the Buildup Editor. There, we can see the individual cores and prepregs, their type and height. Some examples are shown below:

Additional information about how many soldering cycles our materials can endure is here, and how we check the quality of our PCBs with microsections here.