Copper Migration In FPC Flex Circuits
A flex circuit that passes its first test can still fail in the field months later because of a process that leaves no mark at assembly. Copper from the conductors dissolves, moves under the influence of an electric field and moisture, and deposits as a dendrite that grows across the gap until two nets touch. The fault appears as an intermittent signal, then as a dead function, and the failed unit often shows nothing unusual under a low power microscope.
This article explains how a copper migration failure develops inside a flex circuit, what test conditions expose it, and which design and material decisions slow it down. The focus is on flexible constructions, where thin dielectric layers, adhesives and repeated bending make the problem more likely than on a rigid board.
How Copper Migration Starts In A Flex Circuit
The mechanism is electrochemical. A thin film of moisture forms on or inside the dielectric, an electric field is present between two conductors at different potentials, and the metal at the anode goes into solution as ions. Those ions travel toward the cathode, deposit as metal, and grow a filament back toward the anode. The process needs all three conditions at once, which is why a dry board with no bias never shows it.
Flexible constructions invite the mechanism. The dielectric layers are thin, the adhesive between the coverlay and the base film can absorb moisture, and the traces are often closely spaced at the connector or under a stiffener. Contamination left from plating or from handling provides the ionic species that carry the current through the surface film, and the small gap between conductors keeps the field strength high.
The Test Conditions That Reveal It
The standard exposure is a steady bias at 85 degrees Celsius and 85 percent relative humidity, held for 500 to 1000 hours with the traces energised at their working voltage. The bias is what separates this test from a damp heat storage test, because without a field there is no driving force for ion transport. Insulation resistance is logged through the run, and a drop of two or three orders of magnitude marks the onset.
A part that survives 1000 hours at those conditions has demonstrated a useful margin, but the test is a comparison rather than a guarantee. Two material systems can both pass and behave differently over a ten year field life, so the result should be read alongside the applied voltage, the trace spacing and the expected humidity. Accelerating the test by raising the voltage changes the mechanism and is a poor substitute.
Design Rules That Slow Electrochemical Migration
Open the spacing where the voltage is highest. The rate of electrochemical migration depends strongly on field strength, so the most effective single change is to increase the conductor gap in the region where the bias and the contamination coincide. A spacing that is comfortable for a signal net is often inadequate for a net that sits at supply potential for the whole life of the product.
Keep the field away from the edge of the flex and from the bend region, where the coverlay is most likely to be stressed or lifted. A trace that runs to the edge has less dielectric around it and gathers contamination more easily. Where a high potential net must cross a tight area, a grounded guard trace between the two conductors interrupts the field and is a more reliable measure than a marginal increase in spacing.
Material And Adhesive Choices
The adhesive that bonds the coverlay is often the weakest link, because it is the layer that absorbs moisture and holds ionic residues. An adhesive with a low moisture uptake and a clean cure reduces the supply of mobile ions, and the same reasoning applies to the base film. Where the application is severe, a construction with no adhesive layer at all removes the problem rather than managing it.
Surface finish and cleanliness matter as much as the polymer. Residue from plating, from flux or from handling provides the ions that make the surface film conductive, and a final clean with a suitable rinse removes them before the coverlay is laminated. The steps in the process are worth checking here; the fabrication sequence and the dimensional behaviour of the laminate both determine how much residue survives into the finished part.
Insulation Resistance Measurement And Bend Life
Insulation resistance is the parameter to measure, and the measurement has to be made at the working voltage rather than at a low test voltage, because the mechanism depends on field strength. A reading taken immediately after a humidity soak and a reading taken after a period of drying will differ, so the conditions should be recorded with the number. Trending the resistance over the soak is more informative than a single pass or fail value.
Bend life interacts with all of this. Each flex cycle opens and closes micro cracks in the conductor and in the surrounding dielectric, and moisture reaches the trace along those paths. A construction that survives 10000 bends in a dry state may show migration at 3000 cycles once the sample is humidified, so the mechanical and the electrical tests should be run on the same samples rather than on separate ones.
Field Failures And What They Look Like
A migration failure rarely arrives as a hard short at the start. It begins as a rise in leakage that changes the bias point of an analogue node or the timing of a digital one, and it may only show when the product is warm or after a period of high humidity. The intermittent behaviour is what makes it difficult to attribute to the board rather than to the component on it.
When a suspect board is examined, the dendrite is often at the interface between the coverlay and the base film rather than on the exposed surface, so a surface inspection alone will miss it. Sectioning the board across the suspect gap exposes the growth, and the location is usually where the spacing is tightest or where the coverlay is stressed. For flex assemblies that must survive handling and cleaning before they reach that point, a protective coating adds a barrier but does not replace the design measures above.
Process Control and Verification
On a design of this kind, bend life is the item that decides how the rest of the board is arranged. Running a first article through the same checks as the production panel confirms that the two agree, and that comparison is the cheapest form of process control available at prototype stage. Keeping a sample from the panel turns a dispute into a measurement, because the same coupon can be re-examined by both parties without rebuilding the batch.
FAQ
Does conformal coating stop copper migration completely? No. A coating slows the arrival of moisture and blocks surface contamination, but a flex circuit that is bent repeatedly can crack the coating and reopen the path.
What trace spacing is safe for a biased flex circuit? There is no single figure. The safe spacing depends on voltage, humidity and the dielectric, so it should be validated on the actual construction at 85/85 with bias applied.
Why does a board pass the 85/85 test and still fail in the field? The test is a fixed exposure. A product that sees condensation, contamination or repeated bending experiences conditions the test does not reproduce, so the result is best treated as a comparison between designs.
