Reliability Study of Bottom Terminated Components (Part 1)
Bottom terminated components (BTC) are leadless components where terminations are protectively plated on the underside of the package. They are all slightly different and have different names, such as QFN (quad flat no lead), DFN (dual flat no lead), LGA (land grid array) and MLF (micro lead-frame. BTC assembly has increased rapidly in recent years. This type of package is attractive due to its low cost and good performance like improved signal speeds and enhanced thermal performance.
However, bottom terminated components do not have any leads to absorb the stress and strain on the solder joints. It relies on the correct amount of solder deposited during the assembly process for having a good solder joint quality and reliable reliability. Voiding is typically seen on the BTC solder joint, especially on the thermal pad of the component. Voiding creates a major concern on BTC component’s solder joint reliability. There is no current industry standard on the voiding criteria for bottom terminated component. The impact of voiding on solder joint reliability and the impact of voiding on the heat transfer characteristics at BTC component are not well understood. This paper will present some data to address these concerns. We will present our study on the thermal cycling reliability of bottom terminated components, including non-symmetrical LGA and QFN components. Two different solder process conditions and different voiding levels were included in the study, and the results will be discussed. The paper also covers our thermal modeling study of the heat transfer characteristic of BTC component.
BTC is known as bottom terminated component or bottom termination component. It is a leadless component for which the termination is protectively plated and is on the underside of the package. Common BTC components include QFN (quad flat no lead), LGA (land grid array), MLF (micro lead-frame), DFN (dual flat no lead), etc. BTC components are available in different sizes, lead counts, and designs. Most parts are unique from supplier to supplier with various pad designs.
Most BTC components typically have a large ground or power termination along with smaller signal terminations. Voiding is commonly seen at the solder joint of BTC components, especially at the thermal pads. In many cases, large voids and many voids which can exceed 25% of the area can be seen at the thermal pads of BTC components. Voiding causes many concerns for the solder joint reliability of BTC components. The following questions are typical when dealing with BTC: Will the excessive voiding decrease the solder joint reliability? Will the voiding impact the heat transfer and thermal behavior of the BTC component? In this paper, we will present our study on the thermal cycling reliability of BTC components. Voiding and reliability data will be compared. We will discuss the heat transfer characteristic of BTC component using our thermal modeling study.
The company's Bottom Terminated Component Test Vehicle, Rev 1.0 was used in the study (Figure 1a). It is a double sided board with the dimension of 8” x 11” x 0.093” [203mmx279mmx2.4mm]. Thirteen different BTC component types from various suppliers were designed into the test vehicle (Figure 1b). It also had other components such as BGA, SMT connector, and chip components among others.
Figure 1: Company Bottom Terminated Component Test Vehicle a) Topside b) Various BTC Pad Design.
Table 1 summarizes the component details tested in the reliability study. Both QFN and LGA component types were included in the reliability testing. We also included a BGA and FQFP component for data comparison. For the QFN component type, we had both symmetrical QFN (Figure 2) and non-symmetrical QFN components in the study (Figure 3). For the LGA component type, some LGA components had the same pad sizes for both signal and ground pins (Figure 4). Some other LGA components had various pad sizes for signal and ground pins (Figure 5).
Figure 2: Symmetrical QFN Component Images. a) QFN 88 b) Dual Row QFN132
Figure 3: Non-symmetrical QFN Component Images.
Figure 4: LGA Images_ Identical Pad Size for Signal and Ground Pins.
Previous studies [1-2] showed that more solder generally resulted in less voiding. To study the impact of voiding on reliability, we built boards using two different process conditions to simulate different voiding levels in the solder joint. In one process condition, boards were built using a regular stencil which resulted in more voids in the solder joint. In the other process condition, pre-tinned components were used. This condition typically had less voiding in the solder joint.
Figure 5: LGA Images_ Non-identical Pad Size for Signal and Ground Pins.
Table 1: Component Details
Thermal Cycle Test Conditions
The thermal cycle testing was performed in an air-to-air thermal cycle chamber, from 0 to 100°C with 10-minute dwell time at each peak temperature, and a temperature ramp rate and cooling rate of approximately 10-15°C per minute. The chamber profile of temperature versus elapsed time is shown in Figure 6.
Figure 6: Thermal Cycle Temperature Profile - 0 to 100°C
Continuous monitoring was done on daisy chained components. Non-daisy chained components were cross-sectioned after 1,000 cycles, 2,000 cycles and 3,000 cycles.
Results and Discussions
Pre-tinned BTC components generally had less voiding than non-pre-tinned BTC components. However, voiding varied depending on the component type. Most of the components had an average void % of less than 25% except the dual row QFN132 component and FQFP176 component (Figure 7a). Maximum void percentage was seen up to 30%, 40%, 50%, depending on the component type, its pad size and process condition (Figure 7b).
Figure 7: Solder Joint Voiding Comparison for Sample Built with Pre-tinned BTC vs. Regular BTC. a) Average Void% b) Maximum void %.
Pre-tinned components did not eliminate voids. However, the size of the voids in the pre-tinned samples was typically smaller than the voids of solder joints using non pre-tinned component (Figure 8). The shape of the voids for pre-tinned sample was more defined and circular as compared to the non-pre-tinned solder joint sample (Figure 9).
Figure 8: X-Ray Images of QFN 3550 Solder Joint Assembled with Non-Pre-tinned Component vs. Pre-tinned Component.
Figure 9: X-Ray Images of QFN132 Solder Joint Assembled with Non-Pre-Tinned Component vs. Pre-Tinned Component.
Solder Joint Microstructure – Time Zero Analysis
Cross sections of the BTC solder joints were performed to evaluate the solder joint microstructure and inter-metallic formation at time zero (prior to thermal cycle testing). Non-pre-tinned components had a standoff height of about 2mil to 3.5mil, depending on the component type. The pre-tinned component standoff height was between 4-7mil. It was approximately twice of the non-pre-tinned component standoff [Table 2]. The BGA196 component had a standoff height of around 11mil.
Table 2: Solder Joint Standoff Height of BTC Assembled Using Regular Process and Pre-tinned Components
Good solder joint wetting and normal IMC layers were observed for all the components. No cracked solder joints were found at time-zero for all the samples (Figure 10, Figure 11 and Figure 12).
Figure 10: Cross Section Images of a QFN component at t=0_ Non-Pre-tinned Component.
Figure 11: Cross Section Images of a QFN Component after Reflow_ Using Pre-tinned Components.
Figure 12: Cross Section Images of BGA Component after Reflow.
In Part 2 of this two-part article series, failure analysis after thermal cycle tests will be discussed, as well as the impact of solder voids in the thermal pad of the BTC.
Editor's Note: This paper has been published in the technical proceedings of IPC APEX EXPO.