Solder Joint Cooling Rate and Grain Structure
A solder joint is not a homogeneous piece of metal. It is a composite of tin-rich grains, particles of intermetallic compound and the layers that form where the alloy meets the copper or the component finish. The size and distribution of those features depend on how the joint cooled, and the cooling rate is a consequence of the profile, the board and the component. That is why two boards built with the same alloy and the same peak temperature can behave differently in thermal cycling.
What the Microstructure Is Made Of
A tin-silver-copper joint consists of grains of tin with a body-centred tetragonal structure, with particles of silver-tin and copper-tin intermetallic dispersed through them. Along the interface with the copper pad, a layer of copper-tin intermetallic forms during soldering, and its thickness grows with time and temperature. The interface adjacent to a component termination forms a similar layer with whatever metal the termination carries.
The structure starts forming as soon as the alloy is liquid and continues as it cools. The result is a joint whose properties vary from the interface to the centre, and whose grain structure records the thermal history. That history includes the reflow profile itself and every subsequent thermal excursion, including the second reflow pass, the wave if there is one, and the rework.
Cooling Rate and Grain Size
A fast cooling rate produces fine grains and small intermetallic particles, which gives a stronger but more brittle joint with a higher yield strength. A slow cooling rate produces coarse grains, which is softer and more ductile but which can fail by fatigue earlier under cyclic loading. The difference between a cooling rate of 2 °C per second and one of 0.5 °C per second is visible in the grain size under a microscope.
The cooling rate is set by the rate at which heat leaves the joint, which depends on the thermal mass around it. A joint on a pad connected to a large ground plane cools slowly because the plane holds the heat; a joint on an isolated pad on a thin board cools quickly. This is why the same profile produces different microstructures at different points on the same board, and why a joint that fails in a thermal test is often the one on the heaviest copper.
Intermetallic Growth and Time at Temperature
The intermetallic layer at the interface grows with the square root of time at temperature, so the growth is fastest at the beginning and slows as the layer thickens. A joint that spends 60 seconds above liquidus has a thinner layer than one that spends 120 seconds, and the difference is measurable. The layer is necessary for a good bond, and it becomes a problem when it occupies a significant fraction of the joint.
The consequences of a thick intermetallic layer are a loss of ductility and an increased sensitivity to thermal stress. In extreme cases the layer becomes the site of a brittle fracture, and the failure is sudden rather than progressive. Where a joint sees multiple reflows, the layer thickness should be checked by cross-section on a sample, because the contribution of each pass is cumulative and the limit is set by the thinnest acceptable joint, not by the average.
Effects of the Board and the Component
The board determines the heat path. A board with heavy copper planes conducts heat away from the joint quickly during cooling, which increases the cooling rate, and it also holds heat longer during the soak, which increases the time at temperature. The net effect on the microstructure is not obvious, which is why the joint of interest should be measured rather than the profile averaged over the board.
The component contributes mass and, in some cases, a heat path through its termination. A large component body holds heat and slows the cooling of its own joints, while a component with a metal case can draw heat away faster. The orientation of the board in the oven and its position relative to the conveyor also matter, because the airflow around an edge is different from the airflow in the middle.
Measuring Microstructure
Microstructure is measured by cross-sectioning a joint, polishing it and examining it under a microscope, either optically or with a scanning electron microscope. The features to record are the grain size, the intermetallic particle distribution, the thickness of the interface layer and the presence of voids or cracks. The preparation matters: polishing that smears the soft tin will obscure the structure, so the last steps should be light and the etchant chosen for the alloy.
The measurement is destructive and slow, so it is reserved for the first article, for a qualification, and for investigating a failure. A sample from each board is not necessary; a sample from a representative position on a board from each lot is enough to detect a change in the process. Where a change is suspected, the sections from the current lot and from a reference lot should be prepared together, so that the comparison is not distorted by the preparation.
Consequences for Mechanical Behaviour
A fine-grained joint resists fatigue crack initiation better and propagates a crack more slowly, provided the crack path stays within the joint. A coarse-grained joint deforms more before failing, which can be an advantage where the joint has to absorb thermal expansion. The intermetallic layer behaves in the opposite way to both: it is hard and brittle, and a crack that enters it propagates quickly.
The practical consequence is that the cooling rate and the time at temperature should be chosen for the failure mode that the product will experience. A product that sees thermal cycling needs a joint that resists fatigue; a product that sees mechanical shock needs one that absorbs energy. Both are affected by the cooling rate, so the profile is a mechanical design decision as much as a thermal one. The thermal cycling test results should be read together with the microstructure rather than as a pass or fail.
Process Levers Available
The levers that change the microstructure are the peak temperature, the time above liquidus, the cooling rate and the alloy itself. Cooling rate is the one that is most often neglected, because it is not controlled by a setpoint: it follows from the board, the oven and the exit conditions. Cooling zones with controlled airflow, and a lower exit temperature, are the tools available where the cooling rate has to be changed.
The alloy lever is the most powerful and the least flexible, because changing the alloy changes the melting range and the whole profile. Where a joint has to be more resistant to fatigue, a small addition of a fourth element is sometimes used, and the effect on the melting range has to be assessed. The alloy selection guidance and the mixed alloy notes cover the consequences of combining alloys, which is a common source of an unexpected microstructure at a joint.
What the Specification Should Say
A workmanship specification should not attempt to describe the microstructure, because it cannot be inspected in production. What it can do is set the process parameters that produce an acceptable microstructure: the peak, the time above liquidus and the cooling rate, expressed as a range. Those are the parameters that can be measured on every run and that correlate with the structure.
Where the product has a critical joint, the microstructure should be verified once at the first article and after any process change, and the result recorded with the process parameters it came from. That pair is what allows a later change to be assessed without repeating the sectioning. The thermal cycling test is the functional confirmation, and the two together give a defensible qualification.
FAQ
Does a slower cooling rate always give a better joint? No. A slow rate gives a more ductile joint and a coarser structure, which resists some failure modes and not others. The right rate depends on the loading the product will see, which is why the profile should be set from the application rather than from a general preference.
Can the microstructure be changed by rework? Yes, and usually for the worse, because rework tends to produce a longer time at temperature and a slower or uncontrolled cooling. A joint that has been reworked several times will have a thicker intermetallic layer than its neighbours, which is one reason a reworked site is often the first to fail.
Is the intermetallic layer a defect? No. It is the bond. It becomes a concern only when it is thick enough to reduce the ductility of the joint, which is why the acceptance criterion is a maximum thickness rather than a requirement that it be absent.
