Solder Alloy Choice and Thermal Fatigue Life

A solder joint that is cycled between two temperatures will eventually crack. The mechanism is not a mystery: the board and the component expand by different amounts, the joint is the only thing connecting them, and the alloy is asked to absorb the difference by deforming.

The rate at which thermal fatigue ends in a crack depends on the solder alloy, on the joint geometry, on the temperature range and on how long the product spends at each extreme. Understanding the mechanism makes it possible to choose the alloy and the joint for the life the product needs.

Why the Joint Moves

The coefficient of thermal expansion of a laminate is far higher than that of a ceramic package, and both are higher than the copper planes that stiffen the assembly. When the temperature changes, the two ends of a joint move relative to each other, and the distance moved depends on the difference in expansion and on the size of the component.

That is why a large package with a small pitch is the hardest case. The outermost joints see the largest relative movement, and they carry the highest strain. A ball grid array with a big body and a stiff substrate is the archetypal example.

How Deformation Becomes a Crack

The alloy does not deform elastically at these strains. It creeps, and the creep happens preferentially at the grain boundaries. Over many cycles the boundaries are damaged, small voids form and join, and a crack starts at the point of highest strain, usually the corner of the joint where it meets the pad.

The crack then grows along the interface or through the bulk of the joint. It does not fail the connection when it forms; the joint remains electrically continuous until the remaining area is too small to carry the current and the load, and the failure appears suddenly after a long period of apparently normal operation.

Accelerated Test and What It Measures

Thermal cycling tests accelerate the process by using a wider temperature range and shorter dwell times than the product will see. The result is a life in cycles under the test conditions, and the relationship between that life and the life in the field is the subject of the acceleration model.

The model assumes a failure mechanism that does not change with the test conditions. Where the acceleration is too aggressive, the mechanism does change, and the test measures something that will not happen in the field. Choosing the test conditions is therefore a technical decision rather than a matter of shortening the schedule.

Alloy Choice and Its Effect

Different alloys have different creep behaviour, and the difference shows up in the fatigue life. A lead free alloy generally creeps more slowly at a given temperature, which can be an advantage at high temperature and a disadvantage where the joint has to accommodate a large strain.

The melting range matters as well. An alloy close to its melting point deforms more readily, so a joint that runs hot has a shorter life than the same joint at a lower temperature. Managing the temperature of the assembly is as effective as changing the alloy, which connects the choice to thermal management.

Joint Geometry and Stand Off

Joint geometry decides how the strain from thermal fatigue is distributed. A joint with a larger stand off, produced by more paste or by a taller pad, is more compliant and survives more cycles than a thin, flat joint.

Geometry also sets the location of the failure. A joint with a sharp corner concentrates the strain at that corner, and a fillet that is too small does the same. The design of the pad and the amount of paste therefore have a direct effect on the fatigue life, quite apart from their effect on the electrical connection.

Underfill and Mechanical Support

Underfill is a material placed between the component and the board that shares the load with the solder joints, and it is the most direct answer when a thermal fatigue requirement cannot be met by geometry alone. It reduces the strain in the joints significantly, at the cost of an extra process step and of the loss of reworkability.

The underfill material has to be chosen with the joint and with the expected temperature swing. A stiff underfill transfers more of the load to the component and the board, while a compliant one allows more movement. The thermal expansion of the underfill also matters, since a mismatch with the alloy reintroduces the problem it was meant to solve.

Life Prediction in Practice

A prediction is only useful if it is anchored to a measurement. The usual approach is to test the actual assembly under a defined cycle, record the number of cycles to failure, and then use the model to estimate the field life with an appropriate margin.

The test should use the real materials, because the alloy, the surface finish and the pad geometry all influence the result. A test performed on a generic coupon is evidence about the coupon rather than about the product, and the difference between the two has ended more than one qualification. The requirements that come out of it belong with the other specifications in the manufacturing tolerances.

Design Measures That Help

Several design choices extend the life of a joint without changing the alloy. Reducing the size of the component or the distance between the outermost joints lowers the strain; adding a compliant layer or a thicker board changes the stiffness of the assembly; and keeping the operating temperature lower reduces the creep rate.

Where the product is expected to see many cycles, the arrangement of the joints and the layout of the copper around them also matter. A plane that stiffens the area under a large package reduces the deformation at the board side, which is one of the reasons that plane design and joint reliability are connected rather than separate subjects.

Field Life and the Operating Profile

The life that matters is the life under the conditions the product actually sees. A device that is powered on once a day experiences one large cycle per day, while a device that is switched continuously experiences many small ones, and the two damage the joint in different ways.

Recording the operating profile, including the temperature swing and the dwell at each extreme, is what makes the acceleration model meaningful. Without it, the test result is a comparison between designs rather than a prediction, which is still useful but is not the same claim. The measurement discipline is the same one used across quality control for any product requirement.

FAQ

Does a lead free joint last longer than a leaded one? It depends on the temperature. Lead free alloys generally perform better at higher temperatures and leaded alloys have an advantage in some low temperature, high strain applications.

How long should a thermal cycling test run? Long enough to produce a statistically meaningful number of failures under a mechanism that matches the field. The duration follows from the model, not from the calendar.

Can a cracked joint be detected before it fails? Sometimes, by measuring the resistance of a chain of joints during the test. The change is small until the crack is large, which is why the test is usually run to failure rather than to a limit.