With the advent of the European Union’s Restriction on Hazardous Substances (RoHS) legislation, lead was removed from solder. This required PCB assembly temperatures to be increased to accommodate the higher melting point of the lead-free solder alloys. The assembly temperature increased from a traditional level of 230°C, up to a maximum of 260°C, although some assembly houses are able to assemble at a more modest 245°C.
The additional 30°C has been demonstrated to negatively impact the dielectric material within most types of PCBs used in electronic products. Many of the commercially available dielectric materials are not guaranteed to remain robust in HDI applications when exposed to multiple 260°C assembly and rework thermal excursions. The higher assembly and rework temperature are increasing the risk of material damage.
The lead-free assembly process typically takes two or three thermal excursions to 260°C. There are commonly two thermal excursions in a reflow oven for double-sided assembly and one for localized attachment, plus a wave soldering if connectors are installed. Rework can add another two or three thermal excursions (usually at higher temperatures for longer times); there is one for BGA removal, another for BGA replacement and the third for a hand touchup. This means that the typical PCB assembly (PCBA) may be required to “survive” six high-temperature thermal excursions.
It is not fully understood whether materials used in HDI applications can withstand that amount of heat multiple times without experiencing some level of material degradation. Consequently, material damage is becoming more common in HDI applications where the assembly temperature is applied with 260°C as the upper requirement. In a recent study, 15 out of 24 electronics-industry-approved, lead-free compatible materials exhibited material damage after six assembly cycles to 260°C in a conventional reflow oven.
Thermal cycle testing was done on representative coupons to determine PTH reliability and material robustness. The thermal cycling method used is known as interconnect stress testing (IST). In this method, the thermal cycling of IST coupons ranges from ambient to 150°C in three minutes and returned to ambient in approximately two minutes, whilst constant monitoring of the resistance in copper circuits is recorded. The thermal excursions precipitate barrel cracking that result in increases in the resistance in the test circuit. A 10% increase in a circuit’s resistance is considered a failure (IPC standard). Any material damage can artificially extend thermal cycles to failure, due to the stress-relieving effects. What was found: Most of the time, material damage stress relieves the interconnections--for example, in the central zone of the PTH barrel--thus extending the thermal cycles to failure. This was not commonly found when coupons are exposed to 6X230°C preconditioning, but it is more common when similar coupons are exposed to 6X260°C.
For example, coupons were tested as received without preconditioning and achieved a mean of 500 thermal cycles to failure. When similar coupons are exposed to 6X230°C preconditioning, they failed with a mean of 375 cycles to failure. When the coupons from the same group were exposed to 6X260°C they would fail with a mean of 500 cycles to failure, which was counterintuitive. When the 6X260°C coupons were microsectioned, they exhibited material damage while the coupons tested as received and those exposed to 6X230°C showed no material damage. What was called for was a means of finding material damage electronically.
One finds material damage in IST coupons through the use of DELAM circuits that are built into the coupons. What we do is measure changes in capacitance. The design of the coupon is such that there is a flooded ground plane wherever there is a ground plane on the circuit board. With this design, one can measure the capacitance between these ground plane layers. Typically, the capacitance is measured to be 100 to 400 picofarads between ground planes in IST coupons. What we do is to measure the capacitance before preconditioning, after preconditioning, and at end of test. Greater than a -4% reduction in capacitance suggests that there is significant material damage. A cross-section is then processed on the coupons in question to confirm or refute the presence of material damage. If there material damage is confirmed, then the threshold stays at -4%. If the microsection refutes the presence of material damage, the threshold is reduced to -6%. In this way, one knows generally at what layers the material damage occurs and which coupons to microsection. Typically, one or two coupons show material damage per group of six. Rarely does one find 100% of the coupons showing material damage. By processing the coupons in this way, we greatly increase the ability to find and understand how, when, and where material damage occurs.
Based on our DELAM method, the material damage may be classified in four category types: adhesive delamination, cohesive failure, crazing, and material decomposition. Each of these types has characteristics that are unique unto themselves. This column is focused on adhesive delamination; we will cover the other three types of material damage in future columns.
Adhesive delamination is a breakdown between two laminated interfaces. This is the type of damage seen when two laminated surfaces come apart. There are three stages of dielectric material: A, B and C. The A-stage material is the uncured liquid epoxy. The B-stage material has been manufactured with glass fibers embedded in the epoxy, with or without copper foil on the outer layer, and the epoxy is partially cured and therefore a little bit tacky to the touch. The C-stage material is a fully cured epoxy, copper foil and fiberglass boards. During the fabrication of PCBs, fabricators use mostly C-stage and B-stage materials. During lamination, the B-stage layers become fully cured and as such are used to “glue” the other B-stage and C-stage layers together. The most common types of adhesive delamination are the breakdown between the B-stage and C-stage material, B-stage material and copper, and less frequently between epoxy and the glass bundles as a group.
In adhesive delamination, it is the laminated surfaces that come apart. One cause of adhesive delamination is weaknesses between the epoxy and the oxide coating on the copper. The copper is frequently oxidized to improve adherence between the copper and the B-stage epoxy. Years ago, the amount of oxide coating played a large role in adhesive delamination. If the amount of oxide coating was too thin, then there was a tendency toward adhesive delamination. If the oxide coating was too thick, there was also a tendency to delaminate. This was one of the reasons for IPC’s TM 650 method 22.214.171.124 Time to Delamination (TMA Method) – 12/94 test to see if the oxide coating would delaminate or not. The T260°C test works by bringing a small sample of the material to an isotherm of 260°C, measures the thickness of the sample and hold it there until a delamination causes the thickness to increase, or for 10 minutes, which was end of the test.
Another test, IPC TM 650 126.96.36.199 Thermal Stress of Laminates – 12/94 was the solder float test. In the test a sample of the board or a coupon is floated in solder for 10 seconds, then microsectioned and examined for material damage. This T260 test and solder float test became the quintessential material tests for material reliability. We find that neither of these tests anticipates delamination of materials. These tests were useful when materials had a low Tg, but do not anticipate material damage in higher-Tg materials available today. Today’s oxide treatments are greatly improved and tend not be so prone to delamination.
One of the considerations related to adhesive delamination is vapor pressure. What happens is that the water and other volatiles trapped in the material exert pressure when the water vaporizes into steam. The vapor pressure of water is 300 pounds per square inch (PSI) at 230°C and is 700 PSI at 260°C. So logically, the water vapor should produce huge amounts of pressure internally in the PCB. But we forget to take into account the amount of water available. Given that the amount of water is so low in circuit boards, as low as 0.12% of the weight of the board (polyimide could be as high as 4%), all the water vaporizes and then the pressure stops rising at such a significant rate. This limited amount of water limits the pressure that the water exerts on in the board.
Generally the water vapor and the pressure generated from the water is not enough to cause the PCB to undergo adhesive delamination. There must be other factors at play besides the vapor pressure of the water or other volatiles. Let’s say you have adhesive delamination in a group of coupons that were not dehydrated by baking before preconditioning. To save them, you decide to bake the rest of the coupons at 105°C for four hours in an attempt to drive the water out of the coupons before preconditioning. The odds are that the coupons will still delaminate. I have never found that baking to remove water tips the scale and saves coupons from delaminating. In fact, the bake, if it is too aggressive, at too high a temperature or for too long, may force the coupons to delaminate sooner. The tendency to delaminate after aggressive baking may be due to thermal aging of the epoxy when exposed to high temperatures for long periods of time.
Figure 1: Cross-section showing adhesive delamination.
Adhesive delamination looks like a blister on a cross-sectional view (Figure 1). It is long and tapers to a point at the two ends. The delamination is along laminated surfaces like the interface between the B-stage and C-stage, B-stage and copper or along the glass bundles. Delamination along glass bundle is not the individual glass fiber but between the glass bundles as a group and the adjacent epoxy. Crazing is a separation between individual glass fibers. The concepts associated with crazing will be covered in another column. Figure 2: Animation depicting adhesive delamination. Click here to watch animation.
In adhesive delamination, the separation is along the laminated interface. Frequently this delamination is deep within the board and it is not visible by an external examination. Most often, this sort of material damage occurs during assembly and rework. This is more common with the thermal excursion associated lead-free assembly and rework. In the animation, notice that the delamination occurs during the heating cycle.
This article originally appeared in the February 2013 issue of The PCB Design Magazine.
Paul Reid is program coordinator at PWB Interconnect Solutions, where his duties include reliability testing, failure analysis, material analysis, and PWB reliability consulting.