A Team Effort
1 March 2008 Joe Draina
The costs of building and running a semiconductor fab continually on the rise. Joe Draina, the associate director at International SEMATECH Manufacturing Initiative (ISMI), explains how common approaches and solutions to cost-cutting methodologies and increased capital equipment productivity are now needed more than ever.
Building and running a semiconductor fab is becoming increasingly costly because many economic and productivity solutions are tied to the supply chain. Collaborative approaches appear to be the most promising.
Advanced fabs of the near future – with 2010 as a good point of reference – will include higher levels of automation that make extensive use of such standards as equipment data acquisition (EDA) and recipe and parameters management (RaP) to increase operational efficiency through data-driven decision making.
A sophisticated enterprise system, linked to the automated factory for more efficient ordering and tracking, can also be expected.
Greater energy efficiency to counter rising power costs, along with the increased commitment to environmental protection demanded by regulators and the public, will also shape the future microchip factory.
FUTURE FAB REQUIREMENTS
Inside the fab, process equipment will need to vector constantly towards increased productivity. Metrology equipment will have to keep pace with manufacturability expectations and stay ahead of the process technology curve, to allow for measurement capabilities in advanced development. And, as we move towards the introduction of larger wafers, an affordable plan for transitioning from 300mm to 450mm must be implemented.
These future requirements are already being addressed by the International SEMATECH Manufacturing Initiative (ISMI) on behalf of 14 member companies, which together account for nearly 60% of the world's semiconductor silicon starts.
The need for cost-cutting methodology and benchmark sharing is pervasive. Also imperative is further evolving automation through e-manufacturing while improving equipment productivity. ISMI is targeting all of these areas in current or planned projects.
As shown in Figure 1, the industry has begun to experience a slowdown in the rate of decline in average transistor price. ISMI refers to this conceptually as a decrease in year-to-year productivity gains, which we attribute mostly to a slowing in device technology implementations due to increasing complexity.
When we model this phenomenon, it accounts for approximately $150bn in aggregate productivity losses for the semiconductor industry between now and 2013.
Much of ISMI's work is focused on trying to recapture these projected losses by working to improve factory and equipment productivity, identifying new cost-cutting measures through shared methodologies and benchmarking, as well as working with suppliers to develop standards and solutions to common problems.
IMPROVING EQUIPMENT PRODUCTIVITY
Given its vital importance to future factory performance, ISMI has made equipment productivity improvement one of its top priorities. Our projects in this arena are targeted at defect reduction, installation protocol methodology, equipment chamber matching methodology (see Figure 2) and workshop forums aimed at making specific process tools more productive. Suppliers are important contributors to each of these projects, which in the past year tallied the following successes:
ISMI supplier workshop forums identified solutions to approximately 25 specific equipment issues common to our members, while recording more than 250 individual improvement methods valuable to one or more members.
Global equipment improvement workshops were held by ISMI in maintenance technician productivity, predictive maintenance, second-source spares, equipment monitoring, implanter productivity, etch equipment productivity, backside contamination, fab energy conservation and technician training and certification. All of these meetings documented numerous benchmark comparison metrics and established benchmarks for cost-cutting and efficiency improvement.
ISMI developed a unique methodology for chamber-to-chamber matching in process equipment. Member companies using the methodology have calculated significant savings resulting from improved yield and process parameters (Cpk).
We can expect future fabs to standardise or automate this methodology to achieve greater output and consistency.
A GREENER FAB
Eco-friendly factories that strive to minimise their environmental footprint while maximising employee and community safety are standard in developed nations; within the next decade they are likely to become a requirement in developing areas.
Recognising this trend, ISMI is working to establish green guidelines for semiconductor fabs that incorporate several energy conservation efforts, along with a pioneering drive to establish a green fab building standard. Specifics include the following:
- A cleanroom energy simulation tool that allows our members to calculate the energy cost of ownership. This has been particularly useful in analysing a fab's energy consumption, as well as understanding energy use sensitivity of components in fab upgrades and new fab designs.
- A supplier application guide for energy conservation that documents a common method that equipment manufacturers can use to measure and report the energy usage of their tools.
- Energy conservation forums and workshops on key tools and topics, including equipment chillers, LCD monitors, pumps, heat trace, cooling water, process exhaust, ultrapure hot water, boiler efficiency and adiabatic humidification. All have succeeded in identifying opportunities for lowering consumption and cutting costs.
- Energy leadership imperatives where equipment suppliers work with device makers to establish roadmaps for energy reduction.
- Industry-oriented meetings such as the green fab workshop hosted by ISMI in March 2007, where representatives from universities, fab designers and construction companies joined ISMI members to develop a green fab building standard consistent with the leadership in energy and engineering design (LEED) rating system established by the US Green Building Council.
A cornerstone of ISMI is their sponsorship of industry and member forums where experts can share best-known methods, benchmarking data, industry strategies and success stories. These collaborative councils have proven invaluable to members in developing strategies for effective manufacturing in the next decade and beyond.
In the future, leading fabs will be fully automated, with EDA-compliant equipment and extensive implementations of automated equipment control and automated process control (AEC / APC).
ISMI has been a long-term leader in driving these attributes, with recent achievements that include providing the industry with an equipment data acquisition client emulator to serve as a standard reference for EDA; establishing requirements for data time synchronisation, including the use of network time protocol (NTP); completing the SEMI E139 recipe and parameter management (RaP) and demonstrating RaP systems in member factories.
ISMI's ongoing efforts in e-manufacturing will encompass areas that they believe are essential to future factory productivity, these include guidelines and standards for data quality; methodology and approach to predictive and preventive maintenance of equipment; applications of equipment data such as virtual metrology, cycle time improvements with small lot sizes; varied product mix emphasis and a seamless linking of the enterprise to the factory applications and controls including adaptive scheduling and dispatching
OUTLOOK FOR LARGER WAFERS
Transitioning to the next wafer size (most likely 450mm) will play a central role in mitigating the industry's slowdown in productivity as much as it will characterise leading fabs in 2010.
ISMI's membership recently adopted improvement metrics of 30% for productivity and 50% for cycle time. A key economic analysis underway at ISMI will evaluate the benefits from potential 300mm design or performance improvements (a concept referred to as 300mm prime) and compare them to the improvement metric for the next wafer size. The results will establish a direction for ISMI on whether to begin a full-scale pursuit of efforts to prepare its members' factories for 450mm insertion.
Key areas being examined and modelled for 300mm prime are the effects of equipment availability improvements, first wafer delay improvements, lot size and automated material-handling systems layout with a buffering strategy.
The structure of 300mm prime draws on lessons from our industry's conversion from 200mm, recognising the importance of collaboration, consensus-building and compromise among chip-makers and equipment suppliers.
Whether a transitional phase or an end in itself, the 300mm prime is expected to continue playing a key role in ISMI's project portfolio and influence the advanced factory of 2010.
MEASURING THE NANOSCALE
Manufacturing-capable metrology will be a prime focus for next decade fabs, as we race to develop new technologies to measure our industry's penetration into the nanoscale. Gaps between advanced manufacturing capability and metrology constitute an important concern at ISMI, resulting in our extensive evaluations of commercially available equipment with a focus on potential for extensibility.
Over the past year, ISMI has evaluated more than 30 metrology tools using wafer metrology samples that represent current manufacturing needs and anticipated process development requirements. These wafer metrology samples cover a range of technologies including films, overlay, and defect metrology.
For example, an important metrology sample in this effort is our intentional defect array (IDA) wafer, which incorporates structures down to 15nm to enable the evaluation of sub-45nm tools.
Metrology accuracy also needs careful consideration throughout the process of establishing capable metrology. ISMI is exploring various reference candidates for the future such as atomic force microscopy (AFM), optical beam, ion beam, and X-ray techniques. As an example, in collaboration with NIST, ISMI maintains an AFM reference system with a total measurement uncertainty less then one nanometre.
To operate profitably, wafer fabs in 2010 and beyond will be characterised by extensive automation and integrated enterprise systems, highly productive process equipment driven by norming methods and highly sensitive metrology tools that can provide defect detection down to a few nanometers.
Larger wafers may be unavoidable for fabs in the next decade, although the industry, through 300mm Prime, will retain flexibility on timing for 450mm. Eco-friendly manufacturing will be paramount, with strong pressure to implement energy reduction strategies both to control costs and respond to community expectations. And collaboration aimed at sharing best-known practices and market data will be essential for owners of next-generation factories.
ISMI recognises its responsibility to facilitate these transformations in ways that help its members and the industry avoid a decline in their historical productivity curve. The resulting benefits and revenue will help sustain all of us for decades to come.