Intermetallic Growth And Solder Joint Reliability

A solder joint is not a single material. Between the solder and the copper, the nickel or the gold that it wets, a layer of intermetallic compound forms. It is the reason the joint holds: a metallurgical bond exists where a layer of a new phase has grown between the two metals, and without it there would be only contact. The same layer is also the reason joints fail, because it is brittle and it grows with time and temperature.

This article explains what the intermetallic layer is, how it grows, why thickness matters, how the different metal systems behave, and how the growth is controlled and measured.

What The Layer Is

When tin based solder wets copper, the tin and the copper react to form a compound of a defined composition rather than a mixture. At a copper interface the first layer is a thin, rounded structure and above it a coarser, more irregular one. Their thicknesses are measured in fractions of a micron after reflow and grow with further exposure to heat.

The layer is continuous and adherent when it is thin, and it is what makes the joint strong. Its mechanical properties, however, are those of a ceramic rather than of a metal: it is hard, it has little ductility and its coefficient of expansion differs from that of the solder on one side and the metal on the other. Those differences are what turn a thick layer into a failure site.

How It Grows

Growth is a diffusion process, so it is driven by time and by temperature. The rate rises steeply with temperature, roughly doubling for every 10 to 20 degrees of additional exposure, which is why a joint that spends an extra minute above the melting point during reflow ages far more than the profile suggests, and why a second reflow pass has a disproportionate effect.

The second driver is the supply of the reacting metal. A joint on a copper pad has an effectively infinite reservoir of copper, so the layer grows as long as the joint is hot. A joint on a plated pad has the metal available only in the plating, and once that plating is consumed the growth stops and the interface changes character. That is the reason the thickness of the plating matters.

Why Thickness Matters

A joint with a thin intermetallic layer deforms by yielding in the solder, which is ductile and can accommodate the strain of a thermal cycle. As the layer grows, the fraction of the joint that is brittle rises, and at some point the fracture path follows the interface instead of the solder. The result is a form of embrittlement of the interface, and the failure appears as a joint that cracked at low load, often after thermal cycling rather than at first assembly.

The layer also consumes material from the joint. On a small joint, such as a fine pitch ball, the metal that becomes intermetallic is a significant fraction of the total volume, and the remaining solder becomes a smaller and less compliant part of the joint. A joint that is expected to survive many cycles therefore has to be designed with enough solder volume that the intermetallic does not dominate it.

Different Metal Systems

Copper and tin form one family of compounds and nickel and tin form another, with different growth rates and different mechanical properties. A joint on a nickel barrier, such as the one on an electroless nickel immersion gold finish, has a slower growing intermetallic than one on bare copper, which is one of the reasons for the nickel layer. Gold is a special case, because it does not form an intermetallic with tin in the same way but dissolves into the solder and forms compounds when its concentration is high enough, which is why the gold layer on a board is kept thin.

The alloy choice sets the temperature at which the growth happens. A tin lead eutectic joint is molten at 183 degrees, while a lead free joint is molten at about 217, so the lead free joint spends more time at high temperature during processing and ages more. The differences between the two systems are described under lead free versus leaded solder.

Controlling The Growth

Three levers control the growth. The first is the profile: a shorter time above liquidus reduces the ageing, and a peak that is no higher than necessary does the same. The second is the number of thermal cycles the assembly sees, which includes the second reflow, any rework and any baking. The third is the material system, which is chosen through the pad finish and the plating thickness rather than through the solder.

The design contributes through the solder volume and through the joint geometry. A joint that is thin and wide has more interface per unit of solder than a tall one, so it ages faster in relative terms. The rules that determine the joint geometry are described under PCB design and fabrication, and the process that produces the joint is set out under PCBA development process.

Verification And Failure Analysis

The layer is measured on a section. The joint is mounted, polished and etched with a reagent that reveals the phases, and the thickness of each layer is measured at several points along the interface. The measurement is a research tool as much as a production one, and it is used to qualify a profile or to explain a field failure rather than to check each lot.

Where a field failure is being investigated, the section is combined with an elemental analysis of the fracture surface. A fracture that follows the intermetallic indicates an aged joint, while one that passes through the solder indicates an overload. That distinction decides whether the corrective action belongs in the profile, in the material or in the mechanical design of the product.

Where It Shows Up In The Field

The classic case is a product that passes every test at manufacture and fails after a year in service, with a joint that parted at low mechanical load and shows no sign of melting or of a manufacturing defect. The failure is often triggered by a mechanical event, such as a thermal cycle in a daily on and off pattern, or by a single overload during handling that a sound joint would have survived.

A second case is a board that has been reworked more than once. Each rework adds a thermal cycle and a fresh supply of copper at the interface, and the joint that has been reworked three times may have twice the intermetallic thickness of a joint that was made once. That is a reason to limit the number of rework operations on a high reliability board and to record them, because the joint that fails in the field is often the one that was reworked last.

FAQ

Is any intermetallic layer acceptable? Yes. A joint without one is not a joint. The concern is the thickness and the fraction of the joint that it occupies, not its presence.

Does a second reflow double the layer? It does not double it, but it adds significantly, because the growth depends on the time above the melting point and the second pass adds a further period at high temperature.

Can the growth be stopped? It can be slowed by using a nickel barrier and by keeping the thermal exposure to a minimum, but it cannot be stopped. The practical objective is to keep the layer thin enough that the joint still fails in the solder rather than at the interface.