Improving Thermal Cycling Behavior of QFNs with the Soldering Alloy

BY STEVEN TELISZEWSKI, INTERFLUX ELECTRONICS N.V. 

■ QFN 9x9mm, component without leads.

Introduction 

Due its specific properties and benefits the QFN (Quad Flat No-leads) is gaining importance in electronics assembly industry. The QFN offers a small footprint in combination with a very thin profile and low weight. It also combines an interesting I/O distribution along the perimeter of the package and has good thermal and electrical behavior. These features make the QFN one of the most popular semiconductor packages on the market today. 

In practice however, the QFN package is facing some challenges. 

In the electronics assembly production process, the QFNs are being connected to a PCB board by means of a solder paste that will form solder joints when passing through a reflow oven. The QFN does not have leads like a QFP or PLCC which results in a very low standoff height (gap between PCB board and QFN). This means that the interconnection material or the solder joint is small and very thin. This causes a lower long term resistance to thermo-mechanical stress of the solder joints in the field. 

Thermo-mechanical stress is caused by differences in thermal expansion (CTE) of the different materials used in the PCB and packages, when the electronic unit is heating up and cooling down (thermal cycling). In the field, this heat can be generated by the electronic component itself and by external influences. After a certain number of thermal cycles, the solder joints that connect the electronic components to the PCB are known to start cracking. A cracked solder joint usually leads to malfunctioning of the electronic circuit in the field. Beside the costs of repair, it can in some cases lead to dangerous situations for people using or depending on the electronics. 

■ Crack in QFN solder joint due to thermomechanical
stress.

QFN solder joints have been reported to start cracking up to three times faster than those of a QFP leaded package. 

Obviously a key parameter in this matter is the solder joint itself. The mechanical properties of this solder joint are mainly determined by the soldering alloy it is made of. Other parameters are the base materials the solder joint is connected to, the elements that are introduced into the solder joint by the finishing (e.g. Au from NiAu), the three dimensional structure of the solder joint, the heating and cooling curve the solder joint has seen in the soldering process,… The extent of importance of the soldering alloy parameter however is not fully understood yet as not enough test data is available. 

To get a better understanding, a case study was created that evaluates the thermal cycling behavior of three current soldering alloys in combination with different types of QFNs. 

Design of experiment 

The most obvious tool for evaluation of thermal cycling behavior is a thermal cycling test. For that purpose the industry standard JEDEC JESD-A104 condition G was chosen. Test vehicles were submitted to a -40°C / + 125°C temperature cycle of one hour. Dwell time at the extremes is 10min and ramp rate in between is 10°C/ min. The oven for thermal cycling is a Weiss VTS 7027-10. 

The PCB base material is Isola 370HR. The PCB consists of 6 layers and is about 1,6mm thick. Cu- thickness of the Inner layers is 25μm and the outer layers have 35μm Cu. Board finish is NiAu. 

■ Test vehicle and thermal cycling test chamber

Thermal via’s are 0.45mm plated holes filled with conductive fill. 

There are 4 sections on the PCB board. Each section is populated with 4 identical pieces of a different type of QFN 9X9mm. A version with a low CTE mold (7 ppm) and a high CTE mold (15 ppm) in combination with a large and a small mass plane (die paddle) were chosen. In the results, this will level out failures that are only related to a specific design of QFN. QFNs are daisy chained to enable continuous measuring. For that purpose an Analysis Tech Model 128/256 STD event detector was used. When a an open circuit is detected 10x sequentially, it is considered a failure. 

Three current soldering alloys were chosen for the solder pastes. 

A first choice is SnAg3Cu0,5 as it is by far the most used lead-free soldering alloy on the market today. SnAg3Cu0,5 is known to have rather poor thermal cycling behavior. 

Mainly because of cost driven reasons there is a tendency on the market towards reducing the Ag- content in the SnAgCu alloys. Therefore, the SnAg0,8Cu0,7 alloy was the second alloy chosen to evaluate. 

As the soldering temperatures of SnAgCu alloys can be harsh on temperature sensitive components and base materials, there is another tendency on the market towards low melting point alloys. These alloys allow for lower soldering temperatures and indulge temperature sensitive materials. 

Furthermore, they can substantially reduce energy consumption and manufacturing cost. LMPA-Q is a new low melting point alloy with enhanced mechanical properties developed by Interflux Electronics. As it has shown very good results in earlier thermal cycling tests, it was chosen as the third alloy to complete the test row. 

Each test vehicle was soldered with a dedicated reflow profile for the used solder paste. For the SnAg3Cu0,5 and SnAg0,8Cu0,7 a measured peak temperature around 240°C was used. For the LMPA- Q alloy this peak temperature was around 215°C which resulted in a reduction of energy consumption of about 20% in the soldering process compared to the other two alloys. 

■ Weibull plot of the registered failures.

Evaluation of the test results 

The SnAg0,8Cu0,7 alloy was the first alloy to show failures in between 100 and 200 cycles, quickly accumulating to about 25% of all QFNs and to 90% of all QFNs after 3000 cycles. 

The SnAg3Cu0,5 looks a bit better. It started failing after 500 cycles, quickly accumulating to about 25% of all QFNs and 65% of all QFNs after 3000 cycles. 

The LMPA-Q alloy lasted longest before showing any failures. First failures were registered after 800 cycles, accumulating to about 10% and 25% after 3000 cycles.

Conclusion 

Test results show a very clear influence of the used soldering alloy in thermal cycling behavior of QFNs that are reflow soldered to a PCB board. 

The reported rather poor thermal cycling behavior of Sn(Ag)Cu alloys in this matter was confirmed. SnAg3Cu0,5 has better thermal cycling properties than the SnAg0,8Cu0,7, which might be contributed to the Ag content. 

The low melting point alloy LMPA-Q clearly possess better thermal cycling properties than traditional Sn(Ag)Cu alloys. It can be a viable solution for applications where failures by thermo-mechanical stress are an issue. 

www.lmpa-q.com 

Steven Teliszewski has more than 20 years of experience in the field of electronics manufacturing. After his engineering studies, he started working in the production of a Belgian OEM where he was responsible for the soldering processes and the technical and ISO-coordination. At the manufacturing company of soldering chemistry, Interflux Electronics NV in Gent(B), he got the opportunity to give worldwide support to the Interflux subsidiaries, customers and distributors. Many years of hands-on experience in a wide variety of applications, processes and their parameters created the basis for an expertise in analyzing and solving reliability problems. This expertise is being shared on a regular basis for many years now on technical seminars and fora worldwide. 

0

Start typing and press Enter to search