PCB Laminate Materials: Dielectric Properties, Tg and Reliability

A laminate is chosen by its specifications, and the specifications that matter most are the dielectric constant, the glass transition temperature and the reliability data that sits behind them. Brand names are convenient shorthand, but the same family of materials is available from several suppliers with slightly different filler systems and slightly different behaviour, and the differences are what determine whether a stack-up performs as designed.

Dielectric Constant and Loss

The dielectric constant sets the trace width required for a given impedance and the phase shift a given physical length introduces. Its absolute value matters less than its consistency, because a material whose dielectric constant varies across the panel, or between panels, produces an impedance that varies with it. Woven glass laminates are the usual source of that variation, since a trace can sit predominantly on a glass bundle or predominantly in resin depending on its position.

Dissipation factor determines how much of the signal becomes heat per unit length. Standard FR-4 sits near 0.02, mid-range materials around 0.004 and the best PTFE grades well below 0.001. Moving down that range costs money and buys link margin, and the decision should be made by calculating the loss of the intended trace at the intended frequency rather than by choosing the best available material. Our PTFE article describes the low loss end of the range.

Glass Transition Temperature and Decomposition

The glass transition temperature is the point at which the resin changes from a rigid glass to a softer state. Below it the material behaves as a solid with a modest expansion rate; above it the expansion rate increases sharply and the material becomes noticeably more compliant. That matters during assembly, because a laminate whose Tg is below the reflow temperature spends part of the process in the softer state, and the expansion in the z direction then strains every plated hole in the board.

The decomposition temperature is the separate figure that indicates how much heat the material can take before it begins to char, and the two are often confused. A high Tg material resists expansion; a high decomposition temperature resists thermal damage during desmear and during repeated reflow. A board that must survive several assembly cycles benefits from both, which is why a material specified for high reliability usually carries both figures on its data sheet.

Reliability Data and What It Tells You

The reliability figures that matter are the thermal cycling performance and the conductive anodic filament resistance. Thermal cycling performance is measured by cycling plated holes between temperature extremes until they crack, and the number of cycles to failure depends on the expansion mismatch, the board thickness and the plating quality. It is the most useful single figure for a thick or high reliability board.

Conductive anodic filament growth is a failure mechanism specific to laminates: under a combination of moisture, voltage and a small separation, copper can migrate along the glass fibre bundles and form a conductive path between two nets that are close together. It appears after years in the field, which is why materials are tested for it and why designers working on high voltage or high reliability boards specify a material with good resistance to it.

Other Properties That Affect Processing

Moisture absorption affects the bake before assembly and the behaviour of the material in a humid environment. A higher absorption means a longer bake and a greater risk of delamination if the bake is skipped, and it also changes the dielectric constant in service. Halogen-free materials absorb more than their brominated equivalents, which is one of the trade-offs behind that specification.

Coefficient of thermal expansion matters in every direction. In the z direction it determines how much the board expands through its thickness during reflow, which is what strains a plated barrel. In the plane it determines how the board behaves when components expand at different rates, and how much the board itself will move when it is soldered to something with a different expansion.

Choosing a Material for a Real Design

The order of the questions is what makes the choice manageable. First, what is the maximum temperature the board will see, which sets the Tg and the decomposition temperature. Second, what is the frequency and the longest trace, which sets the dissipation factor. Third, is impedance controlled, which sets how tight the dielectric constant variation has to be. Fourth, what is the reliability requirement, which sets the thermal cycling and conductive anodic filament expectations.

Answering those four usually identifies two or three candidate materials, and the final choice is then made on availability and price rather than on performance. What should be avoided is choosing a material by reputation and then discovering that its dielectric constant differs from the one used in the calculation, because that error propagates into every controlled impedance trace on the board. Our design release checklist places that verification in the sequence.

Comparing the Main Laminate Materials

The available laminate materials fall into four bands. Standard FR-4 is the default for digital and power work, with a dielectric constant near 4.4 and a dissipation factor around 0.02. High Tg FR-4 uses the same chemistry with a modified resin system to raise the glass transition temperature, which improves its behaviour during assembly without changing the electrical figures much. Mid-loss materials, usually hydrocarbon or modified epoxy systems with ceramic filler, bring the dissipation factor down to roughly 0.004 while remaining processable on a standard line.

At the top of the range sit the PTFE based laminates, with a dissipation factor below 0.001 and a dielectric constant near 2.2, which are used where the link budget cannot close on anything else. Between the bands there is a cost step of roughly two to five times per band, and each step buys a specific electrical improvement rather than a general one. Choosing a band therefore means deciding which of the figures on the data sheet is actually limiting the design.

Matching the Material to the Application

A digital board running at a few hundred megahertz and a power converter both sit comfortably in the first band, and paying for a mid-loss material would buy nothing. A board carrying several gigabit links over long traces usually justifies the second band because the loss is measurable and the link margin is finite. Radar and satellite hardware sit in the fourth band because phase stability and loss determine whether the system works at all, and our development flow article describes how that decision is validated.

The application also sets the mechanical requirement. A large backplane needs a stiff material with a low in-plane expansion, a thin flex circuit needs a film, and a power module may need a substrate that conducts heat. The electrical figure that dominates in one case is irrelevant in another, which is why a single ranking of laminate materials, from best to worst, does not exist. What exists is a set of requirements and a set of materials, and the matching is the whole of the design decision.

FAQ

Is a higher Tg always better? Not always. A higher Tg usually comes with a higher cost and sometimes with a higher dielectric constant, so it is worth paying for only when the assembly temperature or the reliability requirement demands it.

What is a mid-loss material for? It fills the gap between standard FR-4 and PTFE, providing a dissipation factor low enough for high speed digital and much radar work at a cost and processability close to FR-4.

Why does dielectric constant variation matter if the average is correct? Because the impedance of each individual trace follows the local value, not the average. A variation across the panel produces a board where some traces are inside the tolerance and others are not.