July - August 1999
eMail the Editor
Dr. Gerald K "Skip" Fehr on IC Assembly & Packaging Trends
An Expert
LOOKS AT THE ISSUES™
|
Q |
How will packaging at the wafer level affect the role of packaging foundries? |
A |
Wafer-level packaging further closes the gap between wafer processing, packaging foundry work and board mount processing. Power, pad layout and flexibility will have to be considered much earlier and planned with the total processing in mind. The packaging foundry needs to add the new, required technology to their capability. At present, there are several technologies being touted for wafer-level packaging. I believe that the foundry must partner with a customer to pick one that will be used and then perform the required development.
It is very difficult to justify developing a process without a customer in mind. Development of standards that can minimize tooling changes for each die design would be most helpful. |
Q |
There seems to be a lot of consolidation in the packaging foundry industry. What's the cause and do you see more? |
A |
There are many reasons that mergers and acquisitions are taking place. One of the more common involves foundries who want to increase their customer base. They can accomplish this by offering more services, adding test or tying into a wafer fab, for example. Economics is a second common reason. The pricing for many packages has dropped very quickly over the past year as a result of over-capacity in the industry, and everyone trying to gain market share to hold their volumes constant. The stronger companies will purchase the weaker ones to gain volume.
I expect consolidation to continue throughout the next year. Getting bigger is not necessarily a good thing, however. The key to success in theis industry is good service, which results in high yields, on-time delivery and fair pricing. |
Q |
How will CSPs affect packaging foundries, in terms of equipment, materials, processes, etc.? |
A |
CSPs will affect design, wafer fab and board/module assembly as well as the package foundry. There are several different technologies being pushed for building CSPs. The foundry must develop at least one to show its capability.
The most common one will be what is essentially the "mini-BGA," which provides for some flexibility in die size and pad layout. Added to this will be the requirement to offer other approaches that a customer wants and that business warrants. This means more technology is used and special processes are required. Processes affected by CSPs will make flip chip more common, with smaller package ball pitch, and there will be mass separation techniques of the CSP from the matrix substrate. Equipment will be affected in two ways: More will be sold as volume grows, and the new process requirements will require the development of more flexible and newer machines. Since portable computing is the primary CSP market, there will be great pressure to cost-reduce the CSPs quickly. This will put pressure on the materials used, with a demand for lower-cost substrate materials such as snappable mold compounds, etc. |
Q |
Will it become necessary for packaging foundries to be more specialized by, serving certain niches, or more general, expanding into many package types? |
A |
I don't think a foundry can survive being just a niche player. They will have to offer many of the specialized packages along with the standard packages. Customers want to be able to buy most of their packaging needs at just a few places (one-stop shopping) to reduce their overhead. Packaging foundries now have to be agile players who can move from product to product to address market demand. |
Q |
What are the key challenges packaging foundries will face in the future? |
A |
Packaging needs from new, emerging markets, such as portable computing devices or information appliances, will drive the new package designs. This will require an ever-expanding technology base of new processes, materials and equipment. The electrical and thermal requirements will continue to increase as the devices operate at faster speeds with lower voltages. Keeping up with the technical requirements will be a large challenge in itself. Smaller and smaller packages with more and more functionality will be the trend. This will require reduced communication loops with the customers and end users to speed time-to-market. Pricing pressure will continue. Design for manufacturing, including working with materials suppliers will be pivotal. Time-to-volume is becoming the same as time-to-market, so cycle time, including communications and setup time, will face the same pressures as pricing. |
Q |
Is there an increasing trend by customers to move to a dedicated line? |
A |
I don't think there is a trend toward customer-dedicated lines in packaging foundries. These make the customer responsible for filling the line-or paying for it, anyway In return they get a lower price benefit (if fully used). Dedicated lines tend to be put in place when capacity is tight or when the design must be protected and the customer wants a guaranteed volume. A special, single-customer package may also require a guaranteed amount of business and special equipment (dedicated line) to get a foundry to run the package. Presently we are in an overcapacity mode and thus there isn't any real, driving reason for the customer to pay for a dedicated line. |
Q |
A few years ago the concept of "lights-out assembly," where the machines would essentially run themselves without human intervention, seemed to be close at hand. What happened? |
A |
I don't think true lights-out assembly is being considered anywhere. This type of facility requires a very long, steady runner, such as a memory product. The amount of maintenance required to keep today's very sophisticated equipment at peak performance also limits a true lights-out factory. Today's equipment, however, does lend itself to greater automation, with more process steps being hooked together. The equipment is better able to measure its output and shut itself down when something goes wrong, unlike the past, when it would just continue to build "bad" product. Automation is more of a cost reduction and productivity effort. In my opinion we will see more integrated lines with better equipment, but we will not see a "lights-out" assembly line for some time, if ever. |
Q |
Cost per lead has been declining for years. Will this continue? |
A |
The cost per lead (ball) has been on a downward curve for years and will continue to do so. In times of low volume, prices drop faster than when most factories are sold out. Overbuilding capacity has the same effect. This will continue to be true in the future. Cost-reduction programs that have always been in place such as reducing materials cost, automation to reduce labor and overhead and lower-cost designs will continue to serve as the most effective ways for packaging foundries to meet the prices and remain profitable. When the price-reduction curve becomes too steep for some companies, they will either go out of business or merge with stronger companies.
When the price reaches a point where no one can make money building the product, and many of the packaging foundries drop that package, the price will stabilize - as it seems to have for TSOPs. The BGA and CSP packages will be no exception to these past trends. In addition, devices are becoming increasingly more complicated, with higher power, and generally require more thermal and electrical capability from the package. This performance increase enables packaging foundries to add value and capability to packages without reducing prices quite as fast as formerly. |