Improving package-on-package reliability: a multi-material approach
As package-on-package (PoP) configurations continue to gain popularity – particularly in the high-functionality handheld sector – increasing concerns regarding their thermal cycling and drop test reliability are driving the industry to develop more robust performance solutions. In the early days of PoP when device pitches were relatively large (0.5mm/0.65mm, bottom/top packages) by comparison, thermal cycling and drop test performance were top priorities, yet seemingly manageable. Today, as PoP configurations become even more miniaturized in terms of pitch requirements (0.4mm/0.5mm, bottom/top packages) and the package footprints have become larger (14 mm x 14mm), the need for more robust materials systems is greater than ever.
Mark Currie, Henkel Corporation, Irvine, CA USA
In developing PoPs, material requirements were important, but the process of building PoPs also had to be established. Driving process deliverables to ensure good performance is certainly part of the equation for modern, larger footprint PoPs with tighter pitches but improving on the material platforms for the interconnect solution is also imperative. Initial reliability improvement for PoPs was largely focused on top-level package stability and finding a way to provide underfill-like protection. Not only are next generation materials such as epoxy flux for ball attach and underfill important, but new lead-free alloys developed specifically for package-level applications and low warpage mold compounds are also improving PoP reliability performance.
Top package challenges
For PoPs, top-level package stability is challenging at best. While the bottom-level package follows standard surface-mount processes and can be reflowed and then underfilled as per traditional methods, the top-level package presents a dilemma as traditional, capillary underfill processes simply aren't conducive to the level two package assembly. Historically, the top package has been mounted using tacky flux, where the spheres are dipped in the flux material and then mounted on top of the level one package. While the fluxing action encourages solder joint formation during reflow, the lack of additional protection (i.e., underfill) presents reliability issues.
Recent materials advances, however, offer a potential solution to the PoP predicament. A new reflow cured encapsulant called epoxy flux, has found use in several applications for semiconductor packaging and printed circuit board assembly, with PoP processes benefiting from the new material as well. Formulated to incorporate both a fluxing component to enable solder joint creation and an epoxy system to encapsulate individual solder bumps for added protection, this new material affords the device support PoP packages – particularly the level two component – require. Additionally, because the materials are cured during the standard reflow process, they also improve process efficiency.
Tacky flux and underfill alternative
To analyze the attachment and protective performance of epoxy flux materials for PoP assembly, a study was conducted that evaluated four different top level attachment materials: two no-clean tacky fluxes (Tacky Flux A and Tacky Flux B), a SAC 305 solder paste and an epoxy flux.
Devices manufactured with the above materials were then tested for drop reliability and, as indicated in Table 1, epoxy flux delivered the best performance with the most number of drops to first failure. These results suggest that the added epoxy encapsulation of the spheres in combination with the flux for solder joint formation offers superior performance to that of flux alone.
|Figure 1. Both devices with tacky flux showed cracks at the top interconnect.|
Using a dye and pry method, additional testing was conducted on a subset of failed devices that had electrical failures on the bottom interconnection only, as well as packages that showed failures on both the top and bottom interconnections. The PoPs assembled using the tacky fluxes had cracks in the top interconnection. Figure 1 shows full cracks (left side image) on the devices that had electrical failures on the top and bottom interconnections and partial cracks (right side image) in the device where electrical failures were detected on the bottom interconnect only. The solder paste-assembled PoP that exhibited electrical failures on the top and bottom had solder joint interconnect cracks on the top package (Fig. 2). The epoxy fluxed devices, however, showed no cracking on the top interconnect. (Fig. 3)
|Figure 2. SAC305 device also revealed cracks on the top interconnect only.|
Ball attach material
The success with epoxy flux on large format BGAs and CSPs in combination with its above stated positive PoP reliability results, led researchers to evaluate its performance as a ball attach medium as well.
|Figure 3. Devices attached with epoxy flux had no cracks.|
A recent study examined three solder sphere alloys, which were all SAC and which are referred to here as SAC-1, SAC-2 and SAC-3. The alloys' shear strength was tested against four different fluxes: one water wash flux (Flux A), a no-clean flux (Flux B) and an epoxy flux (Flux C). Each material was analyzed using a single ball shear test at a shear height of 30micron and a shear speed of 0.5mm/s. For all three SAC alloys, the epoxy flux offered superior solder joint strength as compared to the other fluxes Table 2 illustrates the epoxy flux's effectiveness as a ball attach medium.
Higher reliability lead-free alloys
While the advantages realized with epoxy flux technologies will certainly extend the reliability for PoP devices, at the foundation of performance improvement is the strength of the solder joint. Epoxy flux can serve to strengthen the interconnect when used as a ball attach material, but incorporating an alloy that enhances thermal cycling reliability and drop test performance is also part of the long-term solution for PoP viability.
Unfortunately, the prevailing board-level alloys (SAC305, SAC387 and SAC405) commonly accepted and used in the industry are not necessarily ideal for package-level manufacture. In fact, reliability tests clearly show that the drop-test characteristic life of traditional SAC solder alloys is significantly shorter than traditional SnPb alloys.
Specifically targeting the handheld market and its unique requirements, package manufacturers have made slight process adjustments to accommodate low Ag solder and the traditional board level SAC materials, but neither option has been able to provide both the drop test reliability and thermal cycling performance consistent with that of SnPb alloys.
To address this problem, development of an alternative lead-free solder sphere alloy was initiated to address these industry-wide challenges. The tough criteria for the new alloy were as follows:
• Better drop test performance than SAC105 and SAC305;
• Superior thermally cycling reliability to that of SAC105 and SAC305;
• Compatibility with various surface finishes;
• Solidus/liquidus temperature cannot exceed 225°C; and
• Reliable with different sphere diameters
Focusing on the elemental breakdown inside the developed alloy variations, joint integrity was evaluated through IMC (intermetallic) structure, aging and solder grain structure analysis to predict failure modes. Once potential alloys were selected, thermal cycling and drop test evaluations were conducted on devices with different component side surface finishes and with varying sphere diameters of 0.4mm, 0.3mm and 0.25mm. A single alloy composition was determined and its drop test performance proved to be significantly more robust than that of SAC105 or SAC305 (the current, prevailing alloys used in package-level manufacture) (Fig. 4).
|Figure 4. Newer generation package level alloy exhibits a 20% improvement in drop test performance over that of SAC105 and 40% over SAC305.|
Warpage control mold compounds
Last but certainly not least in the PoP reliability equation, is controlling bottom- and top-level package warping through the use of low-warpage mold compounds. It is not uncommon with PoP devices for the top package warp upward ("smiling face") and the bottom package to warp downward ("crying face"), which often results in stretched, broken, or cracked solder joints.
Mold compounds that are formulated to specifically address warpage through modifications to mold compound properties such as CTE (coefficient of thermal expansion), Tg (glass transition temperature) and shrinkage percentage may alleviate the warpage issues that plague PoPs. While much progress has been made in this area, there are a relative few materials suppliers who have the expertise required to design such mold compound materials specifically destined for PoP packages. But, as miniaturized devices become the norm and die sizes become smaller, mold compound properties and performance will factor greatly in overall reliability.
While making any one of the aforementioned materials alterations would certainly yield reliability improvements for PoPs, the tighter pitches of next-generation packages will dictate a multi-material approach to ensure that required performance metrics are realized. By enhancing protection at the package-to-package connection as well as at the sphere-to-package interface through epoxy flux technology, stress is minimized and thermal cycling and drop test performance improved. Likewise, alloy selection and mold compound properties will largely impact the overall reliability of modern, miniaturized PoP packages. Like most things in electronics, there are many variables to ensuring success and PoP devices are no exception. However, incorporating proven, high reliability materials at all layers of the device – from the package level to the board level – will go a long way toward advancing electronics technology, especially for the handheld sector.
1. B. Chan, et al., "Epoxy Flux Technology – Tacky Flux with Value Added Benefits," ECTC Conference, May 2009.
2. B. Toleno, D. Maslyk, "Process and Assembly Methods for Increased Yield of Package-on-Package Devices," APEX 2008.
Mark Currie received his Phd in electronic assembly and BEng in manufacturing engineering from the U. of Salford, UK, and is global product manager at Henkel Corporation, 15350 Barranca Parkway, Irvine CA 92618 USA; ph.: +1 949 789 2500; email firstname.lastname@example.org.