Emphasis has shifted from upstream processes to optimization of back-end operations for BGA packages
BY J.R. SCHENK
The electronics industry's demands for ever-increasing circuit densities, higher levels of integration and improved cost/performance capabilities have led to an escalating proliferation of ball grid array (BGA) designs in the past five years. Array technologies rapidly are evolving from specialized techniques used in a limited percentage of devices to more widely considered alternatives for many types of designs (from microprocessors to memory arrays to multi-chip BGA modules). According to a recent report, array packages will grow from 2 billion in 1998 to 12.8 billion in 2003 – a 44 percent annual growth rate.1
 Figure 1. A magnified shot of BGA balls being inspected. |
While BGA semiconductor designs already are revolutionizing many of the computing, communications and consumer products that pervade our daily lives, the manufacturing processes required to assemble and package grid array integrated circuits (ICs) also have had to undergo sweeping changes. In particular, significant new demands have been placed upon the final steps of the back-end process, involving package/mark inspection, grid array ball inspection and transfer to tape-and-reel for subsequent use by automated printed circuit board (PCB) assembly systems. These automated tray-to-tape systems are evolving in response to the combined demands for higher sustained throughput rates and the need for flexibility to handle and inspect a variety of package sizes and ball array dimensions.
BGA Technology Trends
There are two major trends in BGA design that directly impact the tray-to-tape inspection and handling processes. On one end of the spectrum is the movement toward creating larger grid array devices that leverage the technology's ability to integrate a high degree of on-chip functionality with relatively dense input/output (I/O) capabilities. On this end of the size range, it is not uncommon to see new BGA designs measuring 45 x 45 mm or larger. On the other end of the BGA technology curve, grid array's inherent opportunities for miniaturization and higher circuit densities are leading to devices with much smaller ball sizes and tighter pitch dimensions. For example, some medium sized 20 x 20-mm BGAs are pushing ball sizes down below the 500-micron level toward the 200-micron range.
The BGA design landscape is diverging into two distinct “sweet spots” in which the technology can be used either to design new functionality and more I/O into larger devices or to shrink existing functionality and I/O requirements into much smaller devices. While it is likely that, in the future, we will see a blending of these trends with ultra-small ball sizes used in even large packages, the need to optimize production throughput and yield is leading most designers currently to opt for one end of the spectrum or the other.
Another key trend in the BGA market that significantly impacts the design requirements for back-end tray-to-tape systems is the continuing movement by many companies toward outsourcing their BGA assembly operations to specialized contract manufacturers. While such outsourcing offers the benefits of targeted experience with BGAs and refined production capabilities, it requires that these contractors manage a wide variety of package sizes, ball dimensions and inspection requirements. From an equipment utilization and return-on-investment perspective, this need to support high-mix environments necessitates that tray-to-tape systems have substantial flexibility to handle this spectrum of outsourced products.
Current BGA Tray-to-Tape and Inspection Challenges
These trends are challenging current tray-to-tape systems with regard to the capabilities of the machine vision inspection systems. The tradeoffs between overall field of view (FOV) and per-ball pixel resolution are being pushed to their current limitations by the movement toward smaller balls and finer pitch.
In most cases, a minimum of 10 pixels per ball is required to conduct an adequate inspection of the grid array. This means that inspection of arrays using a 300-micron ball diameter would require 30 micron/pixel resolution, which, with today's camera technologies, limits the FOV to approximately 30 millimeters square. Therefore, a typical 27 x 27-mm package size with 300-micron balls is within the practical FOV, though a movement upward to 35-mm2 or 45-mm2 packages or a reduction in ball sizes to 200-micron dimensions will stretch beyond current camera FOV capabilities. While it is possible to inspect larger grid arrays in a multi-step camera process, the throughput penalty is significant when compared with acquiring the entire array within the camera's FOV in a single pass. The ultimate solution for this challenge most likely will lie in the current work being pursued by machine vision camera suppliers to provide higher resolutions and expanded FOVs (Figure 1).
For many systems, achievable throughputs also can be limited by the overall speed of calculations used in the inspection process. With more image data being acquired and tighter tolerances needed for acceptable parts, the quick analysis and determination of pass/fail results are becoming significantly more complex. While most systems already have migrated to standards-based computing environments – such as PC platforms with MMI technology that can be optimized for both speed and extensibility – some proprietary legacy system designs are reaching the limits of their calculation capabilities. Although proprietary image analysis capabilities appeared to provide a performance edge a few years ago, that advantage has turned out to be short-lived; high-speed commercial microprocessors and computing systems continually are boosting performance while providing a smooth migration path of backward compatibility.
 Figure 2. The parallel processing tray-to-tape system. |
Another major challenge for tray-to-tape systems is the requirement to continually increase throughput capabilities to keep up with upstream process improvements. Although most of the preceding BGA operations (e.g., die attach, reflow solder, encapsulation, etc.) can be sped up through use of parallel process flow designs, the final back-end steps of inspection and taping have been more constrained. For the most part, these limitations are the results of the linear sequencing used for traditional systems in which parts are presented one at a time for inspection and ultimately are fed onto a single taping reel. Although, in concept, the final inspection and taping process could be conducted in parallel by deploying completely redundant systems, this alternative has lacked economic practicality, because of both the significant expense involved and the limited floor space in most production environments.
The taping and sealing process itself also is being challenged by the larger overall device sizes used for many BGA components. With larger and heavier parts, it becomes inherently more difficult to maintain consistent and proper alignment within the tape pocket, thereby requiring better insertion process control and dimensional tolerances. In addition, the larger pockets can be significantly more challenging to seal properly, consequently increasing the risk of downstream component damage or process problems on the PCB assembly line. Finally, larger BGA components require more frequent changing of reels because of their higher processing speeds and their allowance for fewer components on each reel.
Emerging Next-generation Capabilities
Since all of these BGA trends are continuing to escalate, tray-to-tape system designers need to provide innovative ways of improving performance and flexibility while handling a wider range of part sizes and types. A new generation of tray-to-tape systems is emerging to deal with these throughput demands and high-mix process requirements by fundamentally changing a number of the underlying handling and inspection processes.
Parallel Pick, Presentation and Inspection Operations: A high degree of parallel processing is one of the key innovations that will be used for boosting sustainable throughput in next- generation tray-to-tape systems (Figure 2). Traditional tray-to-tape machines typically have made use of single-device pick-up heads to move parts one at a time from the tray to mark-inspection, then to ball inspection, and subsequently to taping. While some high-end current-generation machines already use multiple heads for moving individual devices between sequential functions, new system designs will become even more efficient by building in the ability for multiple heads to pick and move multiple parts at the same time.
In addition, these new systems can provide an expanded level of options for tailoring the multi-part handling capabilities to specific process requirements. For example, a system can be configured to pick up many components from the tray simultaneously and then rapidly step each device through the vision inspection station, without having to go back and individually pick more parts, until all are completed.
Minimizing Unnecessary Movements: Leading-edge current systems already are designed to minimize unnecessary movements through techniques such as avoiding the requirement to flip devices between inspection processes. For example, by mounting one camera above the mark-inspection station and another below the ball-inspection station, the need for flipping can be eliminated. In new-generation machines, incorporating options for in-tray mark inspection can further reduce unnecessary movements. In these instances, the entire tray can be scanned and inspected prior to picking and moving any parts, improving overall throughput while simultaneously eliminating the wasted motion of picking and processing parts that fail the mark inspection.
Improved Taping and Sealing Capabilities: New-generation systems also will improve the taping process itself by enabling multiple parts to be placed into the tape in a single operation. By using tightly integrated capabilities for dynamic adjustments to alignment and pitch, placement heads will be able to deliver multiple parts to tape rapidly, while maintaining consistently uniform and highly precise in-pocket positioning registration. In addition, these new machines offer significantly improved control over the sealing process to ensure consistent final quality even when taping two components at once.
The Bottom Line
While the demand for BGA designs has been escalating steadily, much of the process optimization efforts thus far have focused on the upstream operations, putting increasing pressure on the final inspection and tray-to-tape processes – pushing them to keep up with overall throughput rates. However, significant emphasis now has shifted toward balancing the whole production environment by optimizing these back-end operations as well.
Over the next few months, many of the innovations discussed in this article will be emerging in new-generation tray-to-tape machines, which will offer higher sustained throughput capabilities and enhanced flexibility to handle a wide range of BGA package sizes and ball dimensions. By boosting the overall efficiency of BGA assembly processes and simultaneously improving costs through better equipment utilization, these new handling capabilities also will help make the use of BGA designs an even more viable option for a wider range of applications.
Reference
1. “Advanced IC Packaging Markets and Trends,” Electronic Trend Publications, Nov. 1999.
J. R. SCHENK, engineering manager, can be contacted at Ismeca, 2365 Oak Ridge Way, Vista, CA 92083; 760-305-6200; Fax: 760-305-6294; E-mail: [email protected].