HAST Test Setup: 5 Rules for PCB Reliability

A HAST test puts a board or an assembly under high temperature and high humidity while it is electrically biased, and it is used to find humidity driven failure modes in days rather than in the months a steady state humidity test would take. For a fabricator or an assembler, it is the test that exposes process weaknesses before a customer finds them in the field.

The test is only as good as its setup. Bias, temperature, humidity and duration interact, and a run that is too mild proves nothing while one that is too aggressive produces failures that have no field equivalent. The useful HAST test is the one whose conditions can be defended as an acceleration of a real service environment.

What a HAST Test Does

The test accelerates the failure mechanisms that depend on moisture and electrical potential. Water vapour diffuses into the laminate and into the solder mask, ionic residues dissolve, and the applied bias drives metal ions through the electrolyte that forms. The result is leakage, dendrite growth or a conductive anodic filament along a fibre bundle.

Those mechanisms are the same ones that cause field returns in humid climates, which is why the test is used as a qualification tool rather than as a routine inspection step. For the assembly process, the value is that the test measures the residue left by flux, cleaning and handling rather than the residue the process claims to leave. A board that passes in dry conditions and fails here is telling the process something that a visual inspection can never show.

Why HAST Replaced Longer Humidity Tests

The older steady state test, run at a mild temperature and high humidity for a thousand hours, reproduces the physics of moisture absorption but takes six weeks. HAST raises the temperature above the boiling point of water in a pressurised vessel, which increases the diffusion rate by orders of magnitude.

The trade is that the acceleration model has to be justified. Above the glass transition temperature of the laminate the material behaves differently, and a test run too far above it can produce failures that would never occur in service. Choosing the condition therefore starts with the laminate and the design, not with the capability of the chamber.

Bias Voltage and Test Board Design

Bias voltage is what separates HAST from a plain damp heat test, and it has to be applied between the features whose isolation is being evaluated. A test board with adjacent traces, closely spaced vias and an interdigitated pattern of opposite polarity gives the most information per sample.

The board should represent the production process, including the same solder mask, the same surface finish and the same cleaning step, because the failure path runs along the surfaces those steps create. A test vehicle made with a different process measures that process, not the one in production. Cleanliness measurements that predict the outcome are described in ionic contamination testing.

Temperature, Humidity and the Acceleration Model

Temperature humidity conditions are chosen so that the acceleration factor against the intended service environment can be calculated and defended. The humidity matters as much as the temperature, because it sets how much electrolyte is available, and the two together decide which mechanism dominates.

The chamber has to hold both values across the load rather than at the control sensor. Steam and vapour distribution differ between a full chamber and an empty one, so a condition that is verified empty may not be the condition the boards actually see. Chamber qualification with a mapped load belongs in the test plan. General humidity control practice is described in temperature and humidity control.

Duration, Sample Size and Sampling

Duration is set by the failure mechanism and by the confidence the test has to provide, and it is normally read from the acceleration model rather than chosen for convenience. Sample size follows the failure rate that must be detected, so a test intended to demonstrate a very low failure rate needs a large number of biased features.

Sampling practice often uses a stepped approach, with readings taken at intervals rather than only at the end. A test that reports only pass or fail at the end loses the information that shows how quickly the population degraded, and that information is what distinguishes a marginal process from a robust one. Where the test is repeated on a periodic basis, the same biased test vehicle should be used every time, because a change of pattern changes the stress at the failure sites.

Failure Criteria and Continuous Monitoring

The failure criterion should be an electrical measurement, usually insulation resistance below a defined limit or a leakage current above it. Continuous in situ monitoring is preferable to periodic measurement, because a failure that heals when the bias is removed will be missed by a manual reading.

Setting the limit too loosely turns the test into a pass for everything, and setting it too tightly converts normal variation into failures. The limit belongs in the plan, agreed before the run, and it should be stated in the same terms the design uses for the finished product. Related test practice is described in electrochemical migration testing.

Failure Analysis After the Test

Failure analysis is where the process learns something. A microsection through the failure site shows whether the path runs along the resin and glass interface, which points at CAF, or across the surface between two conductors, which points at residue, flux or contamination.

The analysis has to happen before the sample is dried out and handled, because the evidence is a thin film of electrolyte and corrosion product. Photographs, a section and a residue analysis together allow the result to be tied back to a specific process step rather than to the test chamber. Where the failure is a dendrite, the analysis should also identify the metal involved, because copper and tin point at different sources.

Records, Correlation and Reporting

The report should record the exact conditions, the loading pattern, the bias scheme, the sample construction and the monitoring method. Without those, a HAST test cannot be compared with a previous run or with supplier data, and the result becomes an anecdote rather than evidence.

Correlation with field returns is the final step. When a HAST test produces a failure that matches a returned unit, the test has earned its place as a process control; when it produces failures that never appear in service, the conditions need revisiting. Thermal cycling data provides the other half of the reliability picture, as described in thermal cycling testing, and reference methods are published by IPC.

FAQ

How long does a HAST test run? Typical durations are measured in tens of hours rather than weeks, with the exact figure derived from the acceleration factor for the intended service life. The duration should be justified in writing, not inherited from a previous project.

Is HAST a substitute for thermal cycling? No. The two tests accelerate different mechanisms, and a board can pass one while failing the other. Where both moisture and thermal fatigue are risks, both tests belong in the qualification plan.

Why do samples have to be biased during the test? Because moisture alone rarely causes the failure. The electrical potential drives the electrochemical reaction that dissolves metal and deposits it elsewhere, so a bias free test would miss the mechanism entirely.