By Keith Gurnett and Tom Adams
As worldwide electronics manufacturing moves slowly and unevenly into lead-free materials and processes, most attention goes to components and the bonding of those components to the printed wiring board-especially to the effects of the relatively high reflow temperatures on components and joints.
Much less attention goes to the printed wiring board itself. Some years ago, when lead-free manufacturing was still more theory than practice, experts often assumed that “new” higher-temperature materials would replace FR-4, the green-colored glass-fiber-epoxy laminate found in most boards. The FR refers to “flame retardant.”
Engineers who have long experience with FR-4, however, prefer it to expensive higher-temperature materials for many applications. The result is the continuing dominance of FR-4 in the vast majority of lead-free processing being done today.
This means that systems designers, more or less, are pushing FR-4 to its thermal limits. Not much research has been done regarding the effects of higher reflow temperatures on laminates.
Gail Tennant, team leader in regional supplier engineering at electronics manufacturer Celestica in Toronto, outlines the situation.
According to Tennant, there is confusion in terms of what processing temperatures the printed wiring board will be subjected to during lead-free soldering. “One of the key things that we’ve seen in the industry is that there are not really any specifications on what lead-free means,” she says. “Some suppliers think it means a processing temperature of 245 degrees Celsius, but other suppliers are testing to 260°C.
“Overall, lead-free means that the thermal window for the processing of boards is much smaller than it was with leaded solders,” Tennant continues. “Problem areas extend to rework, where a higher temperature is needed to remove the bonded joint from the board.”
Tennant is part of an industry consortium to establish the ground rules. “We are working to define the tests that will qualify laminates for lead-free temperatures,” she explains. “The tests will let us compare one laminate to another.”
One problem area is the long-established glass transition temperature (Tg) of a material. The Tg is less useful for FR-4 laminates in lead-free processing because all of the temperatures involved are above the Tg.
Instead, Tennant explains, the industry is moving toward a decomposition temperature because such a measurement would more accurately describe the behavior of boards at higher temperatures. The decomposition temperature is the temperature at which a board, after a stated time, loses about 2 to 5 percent of its mass through volatility.
The types of damage that an FR‑4 laminate may experience in the tin-lead process temperatures are fairly well understood from long experience with leaded solders, but the same types of damage are likely to be more frequent or more severe under higher lead-free temperatures. Some types of damage can involve little-known mechanisms because of limited experience in volume production with FR-4 at elevated temperatures.
Board assemblers want to pay attention to six areas. The first is separations between layers of the laminate. Delaminations occur in response to high temperatures, and may be enhanced if contamination is present within the layers, or if the laminate has picked up moisture during previous handling. Moisture can be removed by baking the laminate before reflow, but baking is too expensive for most applications.
One problem with delaminations is the difficulty involved in identifying them. Some delaminations, and especially those very close to the surface, cause a visible bubble. Others create no bubble and are not visible optically, but can be found by sectioning the laminate or by using acoustic microscopy.
Delaminations present more of a threat to long-term reliability than some other defects because there may be no electrical signature at the time they occur. More to the point, the delamination may not immediately cause an electrical open. During service, however, the delamination is likely to experience mechanical movement and can collect moisture, making a field failure more likely.
Second, board assemblers must pay attention to distortions in copper barrel plating. Temperatures much above the Tg of FR-4 cause the coefficient of thermal expansion of the laminate to increase dramatically. Because the layered structure of the laminate constrains movement in the x and y dimensions, the greatest movement is in the z dimensions. This movement can “stretch” the copper barrel plating in drilled holes.
Plating a hole with either electro or electroless copper is inherently difficult because the narrow diameter of the hole limits current flow and chemical reactions. This means it is quite a feat to obtain a consistent and even coating. Whether the copper is plated onto the walls of the hole or grown by chemical reaction, the most frequent anomaly happens when the metal thins near the middle of the holes.
At high temperatures this is where rupture of the solder is most likely under the stress of thermal coefficient of expansion. If a rupture does occur it can be hidden when the faces of the break line come together again when the board cools and contracts. Thus, a hidden potential failure point exists.
This type of fault can have two different effects. First, it can open and close with temperature variations and cause an intermittent failure during the operational mode. The cause of the intermittent failure will be difficult to detect without temperature testing-an unusual diagnostic step for laminates. Second, the rupture can cause permanent failure and require replacement of the board.
In boards that use high-density interconnects, mechanical drills or lasers form the tiny holes. Mechanical drills tend to leave ridges, which create an uneven thickness down the hole and can accentuate rupture effects. The effect is minimal in through-holes filled with solder, but more pronounced in unfilled vias.
The third concern is conductive anodic filament (CAF) growth. This complex phenomenon results in the growth or electromigration of a copper filament that can cause a short. It is most common in dense multilayer boards, and occurs where copper barrel plating meets the glass reinforcement fibers of the glass-epoxy laminate.
The laminate is a mat of fiberglass and epoxy formed and laminated under temperature and pressure. Moisture can form an electrolytic cell that leads to copper filament growth along the interface between the epoxy and the fibers. “If you disturb the laminate with a delamination or other damage, over time you will see it grow a CAF, which creates a path for a short,” Tennant notes.
This growth typically bridges two oppositely biased copper conductors. The failure can manifest itself in four main ways: through-hole to through-hole, line to line, through-hole to line, and layer to layer. The most common failure mode is hole to hole.
The bond between epoxy and fiberglass in the basic board makeup is the key. If the higher lead-free temperatures degrade this bond, the loose bonds promote ionic migration. The chance of conductive anodic filament failure increases, and overall assembly reliability may suffer.
Various factors such as the pH level or the presence of impurities during the manufacturing phase, along with the application of voltage in the operational mode, can accelerate the CAF reaction.
The fourth concern involves micro hardening. This change in material properties, which comes from thermal excursions above the Tg of the materials, makes the laminate more brittle. Although excess heat softens the laminate and makes it somewhat rubbery, the return to temperatures below Tg can cause micro hardening and brittleness.
The effects of micro hardening are not well understood. “We are still investigating the implications of this characteristic,” Tennant reports. “However, initial finds show that the materials are more brittle. This has implications for handling and possibly for reliability.”
The fifth concern involves damage to board finishes, which protect the copper from oxidizing from moisture, grit, abrasion, and dust; create a barrier to migration; and identify the points to be soldered. The material choices include metallic coatings (matte or bright tin, tin over nickel, gold over nickel, palladium, palladium over nickel, silver) and organic solderability preservatives (OSPs). Another well-known method is hot air solder leveling, or HASL.
These coating methods are being evaluated for their performance at the higher reflow temperatures of lead-free. Probably the final protective layer is the most critical coating as it is this layer that protects the board surface from environmental deterioration during its working life. Damage during reflow may cause breaks, dimpling, and lifting that allow moisture ingress and peeling. Although these types of finish damage may be visible on open surfaces, they will remain hidden in areas under components.
In the case of HASL the solder will reflow once more during the assembly reflow phase and has the potential to break the final top coating.
The final concern is board warping. Tennant says he has already observed this phenomenon as a result of lead-free temperatures. Sometimes referred to as “bow and twist,” warping is a significant problem for fine-pitch technology where the signal lines are carried out on thin traces and the power lines on much thicker traces. Fine-pitch boards are becoming mainstream and inevitably mean multilayer circuits. Thin traces and the power lines on much thicker traces carry out the signal lines.
Bow and twist can occur even when tin-lead solder is used, but happens more readily at the higher lead-free temperatures. If the two thick-trace planes are put on one side of the stack they act like a bimetallic strip and cause bow and twist in the board. This distortion can unseat components during reflow.
Design of the board layers can limit bow and twist. Using thinner glass fibers in the glass-epoxy mix can also limit warping, although at somewhat higher cost.