Software Automation for Advanced Packaging

Because of the industry transition from 200 to 300 mm wafers, chip manufacturers have undertaken the most significant and extensive retooling effort in the industry's history. Not only has this effort reflected the need to process larger substrates, but also the need to support a new kind of manufacturing infrastructure where automation and standardization have become elemental driving forces.

The requirement for full factory automation in both software and hardware is driven by several undeniable truths. First and foremost, fully loaded 25-wafer carriers weigh in excess of 18 lb. Couple this with their ungainly dimension and you create a work hazard simply by physically moving them on a daily basis. Second, the larger 300 mm wafers create roughly 2.25× more die per set of 25, increasing a carrier's value proportionately. This creates another reason to keep people away from handling carriers. Third, because feature sizes are going nowhere but down, contamination must be eliminated every way possible. These three reasons justify the need for the physical, hardware side of 300 mm automation. The forth issue — misprocessing — becomes key, again because of the wafers' higher value. Misprocessing is minimized via the software side of 300 mm automation. Finally, software provides the vehicle for fast ramps and iterative yield improvements. Figure 1 illustrates the evolution of the 300 mm standards.

In recent years, as factory automation has slowly migrated into 200 mm fabs, IC makers had the flexibility to define their own unique automation specifications. This, in turn, forced suppliers to develop a wide array of compliance options, thereby diluting their ability to deliver good automation solutions on time. This wasn't the case for 300 mm, where the need for automation solutions drove the industry to aggressively develop standards. Well-conceived and complete standards would ensure simplicity for tool suppliers and shorten ramp-ups for new fabs. Figure 2 illustrates the various 300 mm standards.

The standardization solution for 300 mm fab automation is well on its way through varied pilot line implementations. Full factory automation is within reach. Standards have become mature, and consensus will be reached between customer and supplier on the definition and implementation of the standards. Figure 3 shows 300 mm automation requirements.

1. The evolution of 300 mm standards development. (Source: SEMI, Applied Materials)

The demand for smaller, faster and higher-performing devices to support growth markets such as cellular phones, handheld computers, MP3 players and other tightly designed appliances is promoting the need for flip-chip packaging schemes. It has been reported that flip chip will realize a 28% compound annual growth rate through 2004, making it one of the fastest growing market segments.

2. The need for 300 mm automation drove the industry to aggressively develop standards. (Source: SEMI, Applied Materials)

Flip-chip technology, interestingly enough, has been around for 40 years. It is the application of flip- chip-in-package or direct chip attach that drives the need for wafer bumping. In broad terms, wafer bumping can be defined as the process by which solder or gold bumps or balls are formed to the surface of a device, enabling interconnect with flip-chip packages. In contrast to wire bonding, wafer bumping is akin to traditional fab processing, requiring photolithography, metalization and inspection technologies.

Because of alignment of packaging technologies to more traditional processing technologies, there is a need to implement software automation. Also, hardware automation for packaging lines is necessary to support general 300 mm substrate issues as well as specific packaging-related substrate issues. Advanced packaging manufacturing lines (APMLs) need to follow the same automation roadmap as the front end. From a software perspective, all accepted 300 mm fab standards have application in process-related APMLs (Table ).

3. The standardization solution for 300 mm fab automation is well on its way through varied pilot line implementations.

SEMI task forces are in place to guide the test, assembly and packaging (TAP) segments in 300 mm software requirements. The goal is to utilize the experience of the front end to minimize problems and execute a faster transition.

There are few automated test equipment control standards with widespread industry use. This includes the current Tester Specific Equipment Model (TSEM, SEMI E30.3). This increases integration costs for IC manufacturers and restricts the market for test equipment suppliers. The TSEM standard requires clarification, supplemental information and removal/modification of specific capabilities to enable widespread usage.

E30.3 describes the TSEM requirements. International SEMATECH guidelines for tester interface communication have been published in Technology Transfer #01044113A-ENG. The handler guidelines are published in #01044111A-ENG2Handler Interface Communications. Similarly, #01044112A-ENG2 Prober Interface Communications provides the prober interface guidelines.

SEMI task forces have been established to modify SEMI E30.1, E30.2 and E91 for 300 mm requirements. In support of APML, they have been active in developing the 300 mm standards Control Job (E94), Process Job (E40), E87, E90, and E39 for the prober, tester and handler cells.

The basic 300 mm requirements are the same as those addressed in the E94, E40, E87, E90 and E39 standards. Individual task forces are being created to define specific equipment models for making this applicable to a particular cell or step where the tool is being used.

The first fully automated 300 mm APML facilities are coming on-line this year, with some of the largest North American integrated device manufacturers and Taiwan subcontracting houses taking the lead. Although our plans are not finalized, the Advanced Packaging & Interconnect Alliance (APiA) expects to have a 300 mm APML up and running for demonstration before the end of the year.

The APMLs have lagged the front end in terms of automation, primarily because of the cost structure and performance requirements of semiconductor devices. Conventional packaging processing historically has been sufficient to meet the needs of packaging by protecting the chips without degrading performance. The new chips that are designed to meet performance and size requirements demanded by end-user product manufacturers are challenging the capabilities of packaging processes and requiring a new manufacturing paradigm. Packaging now has a significant impact on both the performance and overall cost of this new generation of semiconductor devices. More attention (and funding) is being given to the APMLs for this reason. The beginnings of the recent growth cycle are showing that APML capital equipment spending is growing faster than front-end-of-line (FEOL) spending.

Adoption and integration of automation into APML, particularly for 300 mm, is expected to benefit from the front end's cycles of learning. Of course, disparate standards and the general lack of automation experience vis-à-vis the front end may work to counter this expectation.

There are a variety of hardware-driven automation issues that APMLs must consider. The key issues are handling different substrates sizes and materials (ceramic, SiGe, GaAs, InP and quartz); increasing numbers of bumps per die; handling of probed wafers; varying bump density; and finally, handling of prototype, engineering and production of level volumes.

Another challenge is the handling of thin wafers. Wafers must be processed through backgrind to thicknesses as little as 120 µm to meet the requirements of certain end-user applications such as smart cards. This thinning process is done in the APML and provides handling challenges that require new techniques in both software and hardware.

At some point, the wafers must be transferred to a film-frame and handled (automatically) in that format through the dicing, inspection and possibly wafer sort processes. Film-frames also may be the solution of choice for thin-wafer handling. This new format will place challenges on both software and hardware as well as on device tracking, carrier ID reading, robotics, etc.

Many APML facilities are subcontract houses, which brings on a different set of issues:

• Shipping. Wafers must be automatically handled from and into shipping containers.
• Bridge tools. Many of the subcontract houses also do 200 mm wafers. Automation needs to accommodate.
• Lower-cost tools. Subcontract houses need lower-cost tools to meet their business goals, and automation features are often the victim of cost reduction.

The benefits of automation will manifest itself in several ways: 1) increased ability to subcontract manufacturing by relying on automation standards; 2) increased ability to track wafers and die for process feedback and quality control; 3) minimized handling to improve yields; 4) maximized throughput; 5) enabling new, lower-cost technologies such as wafer-level packaging (WLP) and wafer-level test; and 6) increased ability to integrate technologies into process modules (integration of metrology and inspection technologies is a good example of this opportunity).