CAF Resistance Testing: Test Method and Failure Criteria

CAF, or conductive anodic filament, is a failure mechanism in which a copper salt filament grows along the resin and glass interface between two conductors under the combined influence of moisture, bias voltage and time. The filament does not form inside a single conductor; it grows from the anode toward the cathode and eventually bridges the two, producing a resistive short. The defect is usually latent and only appears after months in the field.

Testing for CAF resistance therefore has to accelerate a slow electrochemical process without changing its mechanism. That requirement shapes the coupon geometry, the conditioning environment, the applied bias and the duration of the test, and it also shapes the failure criteria used to judge the result. A test that is too aggressive produces failures that would never occur in service, and a test that is too gentle produces nothing at all.

What Causes CAF Growth

The mechanism needs three ingredients at the same time: a path along a fibre-resin interface, moisture that dissolves ionic species, and a potential difference between two conductors. The filament then grows by electrochemical dissolution at the anode and deposition further along the path. Removing any one of the three ingredients stops the process completely.

Hole walls and the resin-rich regions between drilled holes are the usual locations, because drilling damages the interface and opens a path for moisture. Poor lamination, incomplete resin cure and contamination from drilling or plating all make that path easier to establish. A board can pass every electrical test and still carry a latent CAF path.

Coupon Design

The standard coupon is a set of parallel plated holes connected into two interleaved combs, so that adjacent holes belong to opposite polarities. A typical pattern uses holes on a 0.9 mm pitch with a 0.35 mm finished diameter, which leaves the hole walls about half a millimetre apart. The holes are connected on both surfaces and, in some constructions, through the internal layers as well.

The coupon must be built with the same laminate, the same drilling parameters and the same plating process as the production board. A coupon with a different construction measures the material rather than the process, which is useful for laminate qualification but not for monitoring a line. Isolation resistance between the two combs is measured before the test begins, so the initial value has to be well characterised. Coupons are normally taken from the panel border, and the sample position should be recorded, because the middle of a panel and its edge do not always behave the same way.

Humidity and Bias Conditions

The standard humidity test condition is 85 degrees Celsius and 85 percent relative humidity, with a bias of 10 to 100 volts applied between the combs. The combination accelerates growth by orders of magnitude while keeping the failure mechanism electrochemical rather than thermal. Lower humidity and temperature extend the test duration dramatically, so most programmes use the standard condition and vary only the voltage.

Bias voltage is the most sensitive parameter. Doubling the voltage roughly halves the time to failure, and it also changes the location of the failure because a higher field concentrates growth at the narrowest gap. Comparisons between test runs are therefore valid only when the voltage has been held constant.

Insulation Resistance Measurement

Insulation resistance is measured at intervals during the soak, either continuously on a monitored channel or at defined points such as 24, 96 and 500 hours. Each measurement must be taken with the bias removed, because the applied voltage itself would dominate the reading. A single reading has little value on its own; the useful information is the trend, and a falling resistance is the earliest sign that a filament is forming.

The failure threshold is usually defined as an insulation resistance below 10 megohms, or a decay of more than two orders of magnitude from the initial value. Both criteria are normally applied together, because a board can start at a low value and remain stable, which points to a process issue rather than to a CAF failure.

Failure Criteria and Interpretation

A confirmed failure requires a visible filament after the electrical result, found by microsectioning along the failed path. The filament appears as a thin copper-containing deposit along the fibre-resin interface, usually running from the anode toward the cathode. The section plane is chosen to follow the suspected path rather than simply through the middle of the hole. Confirmation without a section is not reliable enough for a qualification report.

Where several coupons fail at the same location, the cause is usually a design or process feature rather than a material defect. Where failures are scattered across the coupon, the laminate or the lamination cycle is the more likely source. That distinction is the main reason for testing a population rather than a single coupon.

Process Factors That Improve Resistance

Drilling parameters have the largest single effect, because a clean hole with minimal resin damage closes off the interface path. Reduced feed per revolution, a spindle speed matched to the hole size, and controlled entry and exit material all reduce the damage at the hole wall. Back-up material should be changed often, because a worn backing board no longer supports the exit side and burr damage increases.

Resin content, glass style and cure also matter. A laminate that wets the glass bundles well resists filament growth for longer, and a fully cured resin dissolves fewer ionic species. Desmear and plating chemistry then decide whether the remaining path is sealed or left open to moisture.

Test Duration and Sampling

Standard durations are 96, 168, 500 and 1000 hours, and the choice depends on whether the test is a qualification or a routine monitor. Qualification of a new laminate usually runs to 500 or 1000 hours, while production monitoring uses shorter runs with tighter acceptance limits.

Sample size must be large enough to catch scattered failures. Thirty coupons per condition is a common minimum for a qualification, and the result is reported as a failure rate rather than as a pass or fail on individual boards. A single passing coupon proves almost nothing about the process that produced it.

Reporting and Records

The record should state the coupon design, the laminate and prepreg used, the drilling and plating parameters, the conditioning condition, the bias voltage, the measurement intervals and the failure criteria. The conditioning chamber should also be calibrated, since a five percent humidity error changes the time to failure appreciably. Without those details a CAF result cannot be compared with any other result.

When a failure occurs, the section and the resistance trace should be kept together, because they explain each other. That package is what allows a supplier discussion to stay technical rather than becoming contractual. Retaining the failed coupons themselves is also worth the storage cost, because a later question about the failure mode is nearly always asked.

FAQ

How long does a CAF test take? Between 96 and 1000 hours depending on the purpose, with qualification runs at the longer end of that range.

What humidity is used for CAF testing? 85 percent relative humidity at 85 degrees Celsius is the standard condition, applied together with a fixed bias voltage.

Can CAF be detected by functional test? Only after the filament has bridged the conductors. Before that point the board passes, which is why coupon testing and process control are used instead.