How material properties turn into process decisions on a high-frequency board, and the eight checks worth running before you commit a stack-up to a supplier.
On a high-frequency board the substrate is not a passive backdrop. It sets how fast a signal travels, how much of it arrives, and how long the finished assembly keeps working. Two laminates can carry identical copper artwork and behave like two different circuits, because the dielectric constant and the loss tangent underneath that copper are doing as much work as the traces themselves.
So the difference between a capable supplier and an ordinary one rarely shows up in a catalogue. It shows up in whether the engineering team can tell you what a high-frequency laminate will actually do inside your stack-up, and whether the factory can keep the material behaving that way once it leaves the warehouse. The sections below work through both halves of that capability: understanding the material, and controlling it through production.
Anyone can forward a datasheet. The useful question is what the numbers mean for one specific design. A team that only repeats vendor figures cannot tell you that a dielectric constant drifting across a panel is what will shift a filter’s centre frequency, or that a Z-axis expansion coefficient is the reason through-holes crack after the second reflow. Both facts come from the same habit: reading material properties as consequences rather than as specifications.
In practice that means knowing which property governs which failure, and being able to name the trade-off before the layout is frozen. The rest of this article follows those properties in roughly the order they cause trouble.
Dielectric constant and dissipation factor define a high-frequency material, and both are moving targets. The headline value on a datasheet is measured at one frequency and one temperature. Real circuits work across a band and a temperature range, so the useful question is how far the numbers travel.
RO4350B is quoted at a dielectric constant of 3.48 at 10 GHz, which is where most design work starts. The value drifts as frequency rises, and on a narrowband filter that drift is enough to push the passband off target. A team that knows the direction and size of the drift can compensate in the layout, or say early that the specification needs margin. The same figures for a pure PTFE laminate such as RT/duroid 5880 are far flatter across frequency, which is part of why that material appears so often at millimetre-wave bands.
Temperature moves the dielectric constant too, and the size of that movement depends on the material family. PTFE-based laminates such as the RO3000 series hold their electrical properties over a very wide temperature range. Ceramic-filled hydrocarbon grades such as RO4350B are strong here as well, and they cost less while processing more like ordinary FR-4. The choice is rarely about which material is better in the abstract. It is about how much temperature the application will really see, and whether the programme can absorb a PTFE process route.
A high-frequency board has to hold its electrical behaviour through soldering, thermal cycling and years of service. That requirement is carried by two mechanical properties that stay invisible until they fail.
Coefficient of thermal expansion is quoted separately for the X and Y axes and for the Z axis, and the Z figure is the one that decides whether a board survives assembly. Copper expands at roughly 17 ppm/°C. When the substrate expands much faster in the same direction, every temperature cycle pulls the plated barrel of a through-hole away from the copper it is bonded to; do that enough times and the barrel cracks. Low-loss laminates are engineered so that in-plane expansion sits close to copper, leaving a through-hole stress that is manageable instead of cumulative. The numbers for the common Rogers grades are collected in this explanation of CTE in Rogers laminates.
Thermal conductivity decides how much heat leaves a die through the board rather than through a heatsink. It is easy to treat as a footnote, but on a power amplifier it is a design parameter like any other. RT/duroid 6002, for example, conducts 0.60 W/m·K, several times the figure for a standard low-loss PTFE laminate. That difference changes the junction temperature a design can hold, and therefore how much output power is available before derating begins.
The properties above do not stay in the engineering department. Each one pushes a specific demand onto storage, drilling, lamination or surface preparation. The table below maps the eight that come up most often to the process response they require.
Read it as a checklist rather than a specification. Any laminate you are considering will hit three or four of these rows, and the ones that hit the most rows are the ones that need the most experience behind them.
| Material Property | What It Decides | Process Response |
|---|---|---|
| Frequency dependence of Dk | Whether the passband stays where the design put it | Layout compensation at the target band; tolerance review against real laminate data |
| Temperature stability of Dk and Df | Whether performance drifts between the bench and the field | Material family chosen against the operating range, not the room-temperature figure |
| Z-axis CTE | Plated through-hole life under thermal cycling | In-plane expansion matched to copper; controlled lamination profile |
| Thermal conductivity | Junction temperature under load | Copper area and thermal vias; metal backing where the material alone is not enough |
| Moisture absorption | Whether the stack-up blisters during lamination or reflow | Humidity-controlled storage, first-in-first-out stock, pre-bake before lamination |
| Surface hardness | Hole wall quality, edge burrs, tool life | Drill parameters, entry and backup material, tool replacement by measurement |
| Copper foil profile | Skin-effect loss at high frequency | Low-profile or reverse-treated foil with controlled micro-etching |
| Inert surface chemistry | Whether copper bonds to the hole wall at all | Plasma or chemical activation before electroless copper |
Performance is lost before machining even starts if the material is handled badly. Two controls do most of the work.
Materials such as RO4835 absorb moisture from the air, and absorbed water turns to steam inside the stack-up during lamination or reflow. What appears afterwards is blistering or delamination, at the point where the board is most expensive to replace. The standard response is a storage area with controlled temperature and humidity, first-in-first-out rotation so nothing sits indefinitely at the back of a rack, and a bake matched to the material type and how long it has been in store.
Panels stored on edge or under heavy stacks take on internal stress and warp, and warp present before lamination does not come out afterwards. Dust and oil are the second problem: both interfere with dry film adhesion, and a pattern that lifts during development reappears later as an open or a short. Keeping panels flat, clean and separated costs almost nothing and removes a whole class of defects.
There is no single process window that suits every high-frequency laminate. PTFE is soft and chemically inert. Ceramic-filled hydrocarbon grades are hard and abrasive. Running both on the same line means running each with its own parameters.
Pure PTFE laminates such as RT/duroid 5880 are soft enough that ordinary drilling leaves smear and fuzzing on the hole wall. The fix starts with drill geometry, feed and retract speed, and the choice of entry and backup material, and it ends with activation. Plasma treatment or a dedicated chemical desmear is not optional here, because PTFE will not bond to electroless copper on its own. Skipping that step produces holes that pass electrical test and fail in the field. The process sequence is set out in these notes on processing RO5880 boards.
RO4350B and similar grades carry a hard ceramic filler. That filler is what gives them a stable dielectric constant and good thermal behaviour, and it also wears drills and routers faster than glass-reinforced epoxy does. Tool life has to be managed by measurement rather than by habit, because a worn tool raises hole wall roughness and produces the burrs and white edge defects that show up on routed outlines.
High-frequency laminates use low-profile copper to reduce skin-effect loss, and a smoother foil is harder to bond. Cleaning and micro-etching before pattern transfer have to land inside a narrow window. Too little leaves contamination that stops the dry film adhering; too much roughens the surface and gives back the loss the foil was chosen to avoid.
One of the clearest signs of material fluency is a hybrid stack-up: a Rogers core carrying the RF layers, bonded with ordinary FR-4 prepreg for the rest of the structure. It can cut material cost substantially without giving up signal performance, but only if the engineer understands that the two materials have different dielectric constants, different expansion coefficients and different lamination temperatures. Warpage and delamination are the usual outcomes when those differences are not designed around. The same thinking applies to advising on grade. Where a design needs RO4350B-class RF behaviour but has to run on an FR-4 line, a grade such as RO4835 can be the answer. The mechanics of mixing a high-frequency core with FR-4 are covered in this guide to RO3003 and FR-4 hybrid boards.
Material knowledge and process control are two halves of one capability, and a design is only as good as the weaker half. Understanding what a laminate does across frequency and temperature fixes the specification; controlling moisture, tool wear and activation decides whether the finished board matches it. Buyers who can ask about both, and get specific answers, are in a far better position to judge who should build their next board.
Frequently Asked Questions
Because the dielectric sets propagation speed, loss and delay for every trace on the board. The same artwork on two different laminates produces two different circuits, and no amount of copper adjustment fully compensates for a dielectric that was chosen badly.
Narrowband structures move. A filter's centre frequency or an antenna's bandwidth shifts away from the design point, and the correction has to be built into the layout rather than discovered on the first prototype.
The Z-axis figure, because it acts in the same direction as the plated through-hole. When the substrate expands much faster than copper in that direction, each thermal cycle works the barrel loose, and cumulative damage eventually opens the connection.
Absorbed water turns to steam during lamination or reflow, and the pressure it generates separates the layers as a blister or a delaminated patch. It appears after assembly, which is the worst moment to find it.
Ask which properties of your chosen laminate will drive the process, what they do about moisture and tool wear, and how they activate PTFE surfaces for plating. Specific answers to those three questions tell you more than any capability list.
Send us your frequency plan, stack-up and operating temperature range and we will tell you which material properties will drive your process before the design is frozen.
Get a Fast Quote Fast Reply on WhatsApp