Design and construction professionals are showing unprecedented levels of collaboration through the G450C.
BY ADRIAN MAYNES and FRANK ROBERTSON, G450C, Albany, NY
Hard data and strong collaboration are proving effective in solving the challenges inherent in building 450mm semiconductor fabs. In the past year, the Facilities 450mm Consortium (F450C) — the facilities-focused off-shoot of the Global 450mm Consortium — has provided the unified forum to test and analyze utility requirements, overhead conveyance systems and energy-efficiency strategies. The Global 450mm Consortium (G450C), a New York-based public/private program with leadership from GLOBALFOUNDRIES, IBM, Intel, Samsung, TSMC and the College of Nanoscale Science and Engineering is housed on SUNY’s University at Albany campus and maintains focus on 450mm process and equipment development (FIGURE 1). Combined, preliminary results from the seconsortiaare building the framework for this next-generation fab.
The F450C came into existence in 2013 facing a series of significant technical hurdles. The purpose of the 450mm fab is to manufacture more advanced integrated circuits at lower cost with a lighter environmental footprint. However, the initial 450mm tool guidelines point to a greater cost per square foot (meter) of cleanroom space, much heavier structural loads and significantly larger tool sizes, which can detract from the manufacturing flexibility the industry seeks. The potential for competing visions among industry leaders and the cyclicality of semiconductor demand in the marketplace add head winds on the path toward widespread 450mm adoption. These factors pose challenges that will need to be managed to ensure 450mm program objectives are achieved.
- All key players are coming together
- CNSE is providing a uniquely neutral and technologically advanced home for critical research
- Work of the G450C is being guided by a strict application of an inside-out design approach
- Key advances have been made in utility requirements, overhead conveyance systems and energy efficient strategies
Data are pointing to promising advantages of the 450mm model, which may support broad industry adoption.
A collaborative approach
When the complexity of the semiconductor manufac- turing process adds the scale of a 450mm wafer, the facility requirements can seem immense. The current scale of today’s 300mm factories indicates that managing the size and complexity is key to attaining the efficiencies needed for 450mm adoption. Driving to GLOBALFOUNDRIES’ massive facility in upstate New York, for example, feels more like approaching a coliseum than a manufacturing fab (FIGURE 2).
Naturally, the industry is closely examining impacts to the facility infrastructure along with the increase in wafer size, since merely scaling the manufacturing process is not practical. The size of the 450mm fab and its associated utility consumption projections would exceed affordability and exacerbate sustainability concerns without close focus on potential efficiencies.
The facility experts involved in establishing and implementing 450mm infrastructure requirements are facing a similar degree of challenges as the integrated circuit and equipment manufacturers.
Design and construction professionals are showing unprecedented levels of collaboration through the G450C to deconstruct semiconductor facility matter associated with 450mm adoption. This consortium formed in the summer of 2013 to unite collective industry expertise to tackle pressing 450mm facility and infrastructure issues. With a special focus on safety, cost, schedule, sustainability and environmental footprint, this group aims to: reduce production cost; increase manufacturing productivity; and reduce the environmental load associated with each chip manufactured.
Everything starts at the process level
Early development of 450mm silicon and infrastructure began before the turn of the decade, yet it wasn’t until 2013 that 450mm and 300mm process tools began to progress synchronously through technology development at CNSE. The function, operation and shape of a semiconductor facility are driven by the process technology and its corresponding manufacturing requirements. Design progresses from the inside out, starting with the process, in this sequence:
1. Process: the early development of silicon and infrastructure defines the utility process requirements
2. Equipment & Automation: process requirements dictate the process equipment/ tools
3. Production Environment: process equipment defines the manufacturing environment and critical process systems (power, the periphery of the tools, capable of multiple configura- tions for somewhat heavier payloads water, vacuum, chemicals and gases)
4. Site Infrastructure: the facility and its corresponding site requirements are developed
Equipment suppliers have begun providing tools so that G450C can create a 450mm baseline. Industry guidance is for some 450mm tools to maintain 300mm footprint normalized to throughput. Through working with original equipment manufacturers (OEMs) and suppliers, it becomes clear that some tools required to manufacture wafers are bigger and many components are heavier than their 300mm counterparts. This led members of Semiconductor Equipment and Materials International (SEMI) to prioritize the topic of “cranes and hoists” in a survey of potential 450mm standardization focus areas in late 2012. Members of the G450C agreed that industry alignment was needed to deal with the handling of components for 450mm equipment that might be larger and heavier than those currently lifted manually.
In March 2014, a multifaceted work group consisting of IC makers, OEMs, and facility systems suppliers published an initial set of component lifting systems for 450mm fab equipment. The Component Lift working group addressed a number of IC maker concerns around interference with overhead track automation systems, ceiling loading and fab layout flexibility. This led to almost immediate determination that ceiling-mounted cranes generally would not work for wafer fabs. Three broad classes of component lifting were then identified:
1. Generally light-duty custom fixtures integrated with and often mounted on the tools for specific component handling operations
2. Aisle-based mobile lift mechanisms operating from
3. Gantry-like structures addressing the full span of large (e.g., cluster) tools, capable of heavy lifting and conveying payloads beyond the tool periphery
In a 10-month span, the Component Lift working group published a cost of ownership model, safety imperatives, productivity suggestions and other key considerations in their March 2014 guidelines. While more work awaits, this effort is indicative of a new era in which industry collaboration lays the foundation for 450mm success.
Correctly sizing utilities
Vital to the 450mm program’s success is the ability to create an efficient manufacturing facility that builds upon industry know-how and lessons learned from prior technologies. Yesterday’s approach of examining singular utility systems and then searching for ways to improve their individual efficiency is not sufficient. Facility designers must consider how process equipment and facility systems interact as a whole.
In the fall of 2013, the F450C conducted a survey with all aforementioned 450mm consortium member companies to build a roadmap based on the industry’s priority focus areas. A plurality of polled members identified 450mm factory utility right-sizing as the top priority for the F450C. Simply stated, we want smarter utility consumption data to enable more efficient 450mm factories, not singular systems, but as a whole factory.
A “utility right-sizing” focus group was created and has been working since the fall of 2013 to characterize true 450mm utility consumption and requirements. Using the inside-out design process, the first step is to conduct real-time measurements for all critical 450mm process tools.
A critical step in the semiconductor manufacturing process is that of wet process, in which liquid chemicals remove materials from a wafer. This step is well known for its high usage of power, water, drains and chemicals. The G450C and F450C members are installing a series of monitoring devices on CNSE’s first 450mm wet process tool.
Haws Corporation, a F450C member company, has donated a series of effluent monitoring systems that enable utility characterization for acid waste neutralization drains, HF drains and exhaust. By characterizing drain effluents, the industry can more intelligently identify opportunities for reuse, reclaim and recycling. M+W Group, CH2M Hill and the G450C are collaborating to install power and flow meters on this same tool. In the second quarter of 2014, the group will conduct real-time utility measurements at idle, operational and peak modes, and combine the data with 300mm benchmarks to assemble the most comprehensive factory utility model in the history of our industry.
To illustrate why this initiative is a priority, consider that a 10 percent decrease in process equipment power consumption would reduce facility construction capex by ~2 percent. For a $1 billion dollar fab, this equates to an impact of approximately $20 million. A 10 percent decrease in process equipment ultra-pure water (UPW) consumption would reduce UPW system cost by approximately 7 percent and facility construction CAPEX by 0.3 percent.
Wet process is the start. This monitoring methodology will be repeated for all critical tool sets within the 450mm process in the coming years.
Just the beginning
Designing a safe way to install and work on 450mm process tools, combined with measuring and characterizing utilities, are two vital priorities for improving facility sustainability and efficiency. But it is just the beginning.
Edwards Vacuum, an original member of the F450C, is also providing its expertise to advance a “Green Mode” project, focusing on resource reduction and energy savings. About half of the process tools in a semiconductor facility use vacuum pumps and abatement systems to treat exhaust gases. The abatement systems were designed to run at full process capacity, even when the tool is not processing wafers. These “sub-fab” systems have significant impact on utility usage (power, process cooling water, acid wastewater treatment and nitrogen). Putting sub-fab systems into idle or green mode when there are no wafers processed can represent substantial reductions in operating and infrastructure costs. The Green Mode project is a multi-equipment supplier collaboration with the following goals:
1. Identify the impacts of vacuum operation changes at the process chamber, i.e., Green Mode
2. Consider other monitoring parameters that could provide information on vacuum system “health” and performance
3. Collect data to better understand correlation between process conditions and vacuum performance
Results from this project will drive SEMI standards that minimize process risk while adding to utility consumption savings.
From suppliers to scientists, the path to realizing 450mm affordability will continue to require unparalleled industry collaboration. In addition to the facility and infrastructure challenges featured in this article, leaders of the semiconductor industry continue to prioritize roadmap items that will further reduce capex for future 450mm fabs.
The G450C and F450C are also working with global airborne molecular contamination organizations and forums to examine how facility-related design (automated material handling system/stocker evaluation, parts- cleaning design, etc.) can reduce defectivity and improve product yield.
And as the roadmap for 450mm is further defined, there will be more facility and infrastructure projects to come.
The G450C and F450C will continue to coordinate globally to ensure a cost-effective facility transition from 300mm to 450mm. In uncertain economic times, it is electrifying to witness unparalleled collaboration where companies that normally compete in the free market have joined together to deliver a safe, cost-efficient and sustainable fab of the future.
Ever since we (the first group components divition people of IBM East Fishkill facilities) started with semiconductor manufacturing inside one large box type of facility, the semiconductor people had follow it for many decades. In view of the large footprint the 450mm fab is going to have (if still follow the safe and traditional box approach) I think it is time for us to think out side of the box. A modular approach of many small boxes sharing a common facility network and a ring of independent fast transportation system would be more efficient and cost effactive not to mention its flexibility to meet future technology development