The Evolution Revolution in Flux
07/01/2007
Fluxes Have Grown Up
BY JIM HISERT AND ANDY C. MACKIE, Ph.D., Indium Corp.
Fluxes - the brownish, fairly liquid material that allows molten solders to wet onto other metal surfaces - have come a long way from their rather primitive SMT forbears in solder paste, wave soldering, and rework to fulfill diverse needs in the semiconductor market. Consequently, the performance demands for these functionalized fluids have increased significantly.
A series of viscous, tacky, ball-attach fluxes has been developed for application on BGA substrates in a pattern of precisely controlled deposits. Spheres of solder alloy are placed into these flux deposits and the final unit is reflowed to form a finished BGA package. Flux choices for BGA manufacturers are primarily governed by reflow profile and atmosphere type, substrate land metallization, sphere alloy, flux application method, flux cleaning method, sphere size, and placement method. Current practices use water-soluble fluxes that allow cleaning after reflow. However, it seems inevitable that the drive for cost reduction and process simplification will result in the use of no-clean fluxes.
Pin-transfer Flux Process
The pin-transfer method is the most common process used to selectively apply flux deposits to BGA substrate land pads. Flux is dispensed into a shallow reservoir with a fixed depth. A squeegee passes across the surface of this flux once or twice, to ensure that the surface is smooth, for reproducible deposits. Next, a specially designed head with a set of spring-loaded pins is lowered into the flux reservoir. After a short time delay, the pin-transfer head is lifted from the flux reservoir and aligned just above the substrate. The pattern of the pins exactly matches the corresponding substrate pads. The spring-loaded pins are now lowered to meet the substrate, then raised again, transferring a large portion of the flux from the pin to the substrate.
 Figure 1. Low voiding that is typical of advanced fluxes in semiconductor packages. |
Many factors must be taken into consideration during this process. The pin properties, such as diameter and point geometry (the shape of the pin end), play an important role in the amount of flux they will carry. The depth of flux into which the pins are lowered is also an important control variable. Furthermore, the rheological properties of the flux and their variation over time play a significant role in determining the volume of flux deposited onto the BGA substrate. Humidity may also be crucial. A 35-55% relative humidity (RH) is the preferred range, especially when using water-soluble fluxes.
Why Pin Transfer?
The use of such an old technology for semiconductor packaging in the third millennium may seem surprising. However, pin transfer lends itself readily to the flux-application process as it combines several factors that make it preferable over printing:
Gasketing: Due to fine pitch and small size, the typical plastic BGA substrate tends to be thin, allowing it to flex a little. If flux printing were to be used, a dedicated board-support unit - or fixture - might be necessary to allow the substrate to gasket completely against the stencil. The pin-transfer method, however, lends itself easily to boards with uneven topographies. The use of spring-loaded flux-transfer pins allows for small variations in board co-planarity.
Aspect Ratio: A small, fairly deep flux deposit is usually needed. For stencil printing, this would require a thick stencil (considerably thicker than a SMT solder paste printing stencil). The level of process control with such a high-aspect-ratio printing process leads to unacceptable variation in the size of the flux deposit.
 Figures 2a and b. Lead-free spheres reflowed under air and nitrogen. |
Shear Thinning: Although ball-attach fluxes are fairly viscous and tacky, repeated printing will cause the viscosity to drop significantly (so-called “shear thinning”). Combined with issues related to the release of flux from the thick stencil, this is a headache for process control if poorly formulated fluxes are used.
Pin transfer is therefore an important process for BGA manufacture, and it is with this in mind that the following study was conducted.
Comparing Fluxes
Four different ball-attach fluxes were prepared using the pin-transfer method. The BGA substrates used were FR-4 laminate with copper (Cu) area-array pads. The spheres used to demonstrate attachment were 0.026-in. diameter (0.66-mm) SAC387 (95.5Sn3.8Ag0.7Cu).
Wetting
Using standard flux chemistries, lead-containing solders with high tin content, such as Sn63 (63Sn37Pb), will wet readily onto standard BGA substrate metallizations. For SAC solder balls, however, wetting often poses a challenge, due to higher surface tension of the high-Sn alloys,3 and the predominance at the solder surface of intractable tin oxides, rather than a SnPb-oxide mix. Added to this, the predominant use of air reflow atmosphere must be considered. Since typical lead-free temperatures are ~40°C higher than in standard Sn/Pb reflow, improving wetting of the solder becomes a complex task.
 Figures 3a, b, c. All spheres soldered well and cleaned effectively at their respective reflow temperatures with 3 different alloys. |
To study the effect of flux type on wetting of substrates, four ball-attach fluxes that were deposited by pin transfer and inspected post-reflow were used. Although this is a simple study, it highlights the wetting ability of the different available fluxes. Two of the chemistries were water-soluble (Flux A and Flux B), the other two were no-clean (Flux C and Flux D). The overall wetting of the water-soluble fluxes was better than the no-clean fluxes, since no-clean fluxes are generally less active than water-soluble fluxes. Between the water-soluble fluxes, Flux B formed bumps with slightly better shear strength. Of the two no-clean fluxes, Flux C formed the highest shear-strength joint.
Cleaning Without Staining
No-clean fluxes rely on rosin or resin to encase the reactive remnants of activator materials, which are fairly ionic. If cleaning is desired for increased reliability, a water-soluble flux is commonly chosen.
Tests revealed a downside that seemed to be a trade-off with the better-wetting fluxes. The two sets of simulated BGAs using no-clean flux appeared clearer than those made using the water-soluble flux. However, room-temperature water with only slight agitation was used to wash the water-soluble flux residue. This partial washing was conducted to show the ease of cleaning between the two water-soluble chemistries.
Between the water-soluble fluxes, Flux A cleaned slightly better than Flux B. Experience has shown that if BGA packages are reflowed in air using a standard lead-free reflow profile, virtually all traces of flux residue can be removed from BGA packages with de-ionized (DI) water using a jet at above 60 psi for 1 minute or longer. Note that, in general, air reflow at excessively high temperatures will severely impact the cleanability of any flux.
Further Testing
Water-soluble Flux B was chosen to undergo additional testing. In a standard SIR test per IPC-TM-650, IPC J-STD-004 sets a resistance requirement for >108 Ω after 176 hours at 85°C/85%RH. Flux B passed this criterion with a resistance of >109 Ω at the end of this time period.
An X-ray image was also taken to check for voiding in the material. Figure 1 demonstrates the low voiding that is typical of advanced fluxes for the semiconductor packaging industry.
All the materials studied were optimized for reflow in air. However, in another study the effect of nitrogen atmosphere reflow was also examined. Figures 2a and 2b are images of lead-free spheres reflowed under air and nitrogen (at around 100-ppm O2), respectively. Although the use of a low-oxygen atmosphere produces a somewhat more aesthetically pleasing solder joint, reflow of the simulated BGA packages in both air and nitrogen yielded solder joints that were robust and fully wetted to the substrate pads.
 Table 1. Results of a wetting-spread test. |
To follow up on this observation, a wetting-spread test (Table 1) was conducted to quantify the difference between air and nitrogen reflow for Flux B. It was concluded that spheres reflowed in nitrogen looked a little better but there was no significant difference in solder wetting.
Solderability of both low- and high-temperature alloys in air is also a potential parameter for concern. The flux in question was able to perform its function over a wide variety of reflow temperatures. Newer fluxes for ball attach are being challenged to work well in air under standard Sn/Pb reflow profiles, and under high-lead temperatures. This requires a flux to have a relatively low activation temperature, yet still remain active without “burning out” at temperatures over a range of 130°C or more. The samples shown in Figures 3a, 3b, and 3c all soldered well and cleaned effectively at their respective reflow temperatures with 3 different alloys.
Conclusion
Technology needs are driving the evolution of fluxes for different applications in semiconductor packaging. Currently, there is no single pin-transfer ball-attach flux that clearly outperforms all other fluxes in each of the many desirable performance characteristics. However, there are fluxes that excel in certain characteristics based upon what is most important to each application. Therefore, customers should work with their supplier to decide which flux will best meet their process needs.
Acknowledgments
The authors would like to thank Dr. Ning-Cheng Lee and Dr. Yan Liu of Indium Corporation for their advice during the data gathering and drafting of this document. We would also like to thank Grace Soh (Indium Corporation, Singapore) for her work, some of which was used in this paper.
References
- N-C. Lee, “The Relationship of Components, Alloys and Fluxes,” Printed Circuit Design & Manufacture, October/November 2005.
- G. Soh, “Interconnect Flux,” Indium Corporation Internal Training Presentation, September 2006.
- A. C. Mackie, “Rethinking the Importance of Reflow Atmospheres in the Lead-free Era,” Circuits Assembly, March 2003.
JIM HISERT, technical support engineer; and ANDY MACKIE, Ph.D., product manager for semiconductor packaging materials, may be contacted at Indium Corporation, 1676 Lincoln Ave., Utica, NY 13502; 315/853-4900; E-mail: [email protected], [email protected].