Value added from backend drives assembly automation
05/01/1998
Value added from backend drives assembly automation
Today`s semiconductor manufacturing backend is a linear extrapolation of technology in place for 25 years. C. Scott Kulicke, chairman and CEO of Kulicke & Soffa Industries, stated that it has been "predictable progress. We`ve worked hard because, obviously, it is more complex to build a quad flat pack than a dual in-line package. But, while today`s factories may be bigger and more efficient, most still run the same old way - discrete lots of a few hundred or a few thousand units carried from station to station."
International experts on test, assembly, and packaging technologies (TAP) met at the Second Annual SEMI TAP Automation and Integration Conference, to focus on technologies for semiconductor manufacturing`s backend. The conference was held in Mesa, AZ.
According to Kulicke, the industry is on the cusp of a change that will need more automation-and-integration technology for TAP, a change driven by the fact that "test, assembly, and packaging are now starting to stand in the way of utilizing the capabilities of next generation fabs," Kulicke said. "The whole question of test and assembly being a nonvalue added technology is changing dramatically."
Presentations at SEMI TAP clearly benchmarked today`s automation and integration in test and assembly as struggling with clustering sequential process steps, integrating automated vision and inspection subsystems, adopting first applications of manufacturing execution systems, and modeling tomorrow`s backend factories.
Kulicke commented that the semiconductor industry`s need for a greater degree of integration and automation in TAP is driven by a broad range of emerging technology changes. "The next five years we will see more changes in the way semiconductor devices are packaged and tested than we have seen in the past 20," he said.
Some of those changes may be:
Wire-bonder technology will finally run out of steam, to be replaced with one variety or another of flip chip.
Increasingly, use of BGA-based packages will be die specific, and package size will change with die shrinks.
The challenge of chip-scale packages that most of the assembly will be done in wafer form.
Known good die will finally become a mainstream technology, "obliterating the traditional idea of assembly."
The industry will learn that testing twice, at probe and at final test, has a certain amount of redundancy.
Built-in semiconductor test techniques will be increasingly implemented. "No one but Intel will be able to afford exhaustive testing technology."
For the future of test-and-assembly manufacturing technology, "One size of factory integration and automation won`t fit all," said Kulicke. He noted that with emerging technology, some factories will partially automate; others will integrate but not automate; and others will continue to make mature products for which you can`t justify the capital investment for automation technology.
Conference authors Harold Trammell of Amkor Electronics, and Emmett Hughlett and Richard Burda of Kulicke & Soffa, elaborated that with each semiconductor manufacturer having a unique set of business imperatives related to its competitive advantage in manufacturing, no single factory integration or automation approach can adequately address the entire semiconductor industry`s requirements. "Therefore, simulation modeling is a vital tool to quantify the benefits of various scenarios and arrive at optimum solutions." The early use of factory simulation technology for emerging assembly and test facilities was evident at SEMI TAP.
Kulicke said, "Factories making leading-edge parts, parts characterized by very high transistor counts, high transmission speeds, and pads with very small feature sizes, will have to change. These factories will more than anything be characterized by a high degree of intellectual integration. They will be one big network with constant programming, lot tracking, scheduling, and data collection."
In Kulicke`s view, the automation-and-integration technology applicable for assembly isn`t the same as the lights-out factory concept. He explained, "Even though automated materials transfer technology exists, the economics of using such is situational - materials transfer makes sense in high value, short cycle situations that are favorable in commodity factories."
Kulicke highlighted some of the challenges that stand in the way of needed automation and integration technologies:
The industry needs a good open architecture system with well defined interface standards, so factory managers can plug and play as they mix and match different networks, systems, and material handling solutions.
Test program file downloading time, which is measured in tens of minutes, challenges central recipe management, which by contrast is relatively easy in assembly.
The robustness of individual systems is crucial before automation is successful.
Kulicke closed, addressing the philosophical barrier of TAP automation technologies. "Few factory managers are clear about the primary criteria used to judge success in automation processing. If you automate for cycle time, you get a different solution than if you automate for yield. If I could pick a single reason for the lack of progress in back-end automation, it would be the lack of managerial clarity about what constitutes success." - P.B.