Moving automation from the front end to the back end
01/01/2002
BY DAVE HUNTLEY
 |
Twenty years ago, there was little focus on software and automation in wafer fabrication. The frontiers of science were being pushed back to keep up with Moore's Law and enabling technologies were bound to succeed. Software engineering was a mere afterthought until it became increasingly important to have an efficient and automated manufacturing process.
Now that the fab automation segment has matured significantly, test, assembly and packaging (TAP) automation is catching up with fab automation. According to the Semiconductor Equipment Monitor (Oct. 21, 2001), TAP represents only 17.8 percent of the capital budget required to manufacture semiconductor devices. This explains why the emphasis has been on fab automation, rather than on TAP. But with the advent of new packaging technologies, such as flip chip and chip scale packages, the complexity and cost of assembly underscores the need for more advanced automation.
A Fast-paced World
In the fab, it was all big dollars, far-reaching global strategies and elegant concepts that rarely resulted in on-budget (and on-time) automation projects. TAP businesses do not accommodate such thinking as well, and automation suppliers are expected to provide point solutions for the TAP manager's problems, which are constrained by pressing schedules. Automation suppliers are often charged with finding solutions that fit a long-term plan.
Short-term needs (such as wafer mapping, material tracking, equipment monitoring and production scheduling) are usually clearly defined. An automation supplier can usually get the job done quickly and at a reasonable cost. This approach can lead to a mish-mash of solutions that fail to integrate with one another. However, having worked in both fab and TAP environments, I believe that a "rapid prototyping" approach is a quicker, more robust route to the final goal of full automation than a more measured (and sluggish) approach typically found in fabs.
Putting It All in Context
It makes sense to provide solutions within the context of a wide framework. This is one reason why, during the evolution of fab automation during the early 1990s, SEMI published SEMI E98 (Provisional Standard for the Object-Based Equipment Model), which provides tool-to-tool level integration, and SEMI E81 (Provisional Standard for CIM Framework Domain Architecture), which offers integration from the factory level to the tool.
Such documents provide a framework to locate a component being developed and the technology with which it integrates other components. It's important to be familiar with these documents, their origins and their implementation in order to provide solutions that will deliver short- and long-term value.
TAP offers some exciting new challenges. In the fab, there are wafers; they come in different sizes, but they are all just wafers. In TAP, there are wafers of various sizes, as well as strips, trays, reels, individual devices, carriers and handlers, and consumables (such as epoxy, wire and mold compound). This makes the back-end automation proposition significantly more interesting and challenging.
TAP is a Growth Market
While the PC will remain a steady driver, high-growth products will be wireless appliances, such as cell phones, which demand smaller form factors and mixed signal operation. The PC market calls for faster, more complex logic devices (the fab specialty), while the wireless appliance market will need smaller, lighter, mixed signal devices (the TAP specialty). As a result, TAP represents an increasing portion of the cost and value of the completed device.
To accommodate this demand, the landscape of back-end processing is changing at a fast rate. Not many years ago, you could enumerate the steps: mount, saw, second optical inspection, die attach, cure, wirebond, mold, singulate, test. Today, with chip scale packages, flip chip, strip test, and so on, the process flow isn't as well defined. Therefore, TAP must be much more flexible than the fab.
Over time, large automation companies will move into TAP and form alliances with, or acquire, the smaller companies already there providing solutions. In the meantime, TAP is on fertile ground in which those smaller companies can apply the skills learned from the fab. AP
Dave Huntley, president, can be contacted at Kinesys Software Inc., 6 Petaluma Blvd. North, Suite B8, Petaluma, CA 94952; 707-766-8855; Fax: 707-766-8196; E-mail: [email protected].