Turn-key collaborative programs for next-generation materials
John Doering, Lou Blanchard, ATMI, Tempe, AZ USA
For many years, semiconductor manufacturers have been striving to control escalating R&D and manufacturing costs.
The driving force enabling materials R&D efficiency is collaboration between semiconductor manufacturers, tool OEMs, and materials suppliers. Before this model was widely accepted, semiconductor manufacturers would typically engage several chemical suppliers by circulating limited information about a materials challenge and seeing what samples each would provide back as a potential solution. This process was not only lengthy, it was also very inefficient. Whichever chemical initially appeared to best meet the needs would be selected for continued development by the supplier providing it. However, many chemicals that were initially promising did not prove optimal, but were merely sufficient. The unsuccessful suppliers may have been able to provide a better solution, but were never given the opportunity. In today's end user devices, where functionality and performance are key differentiators, solutions that simply 'will do' are not acceptable - every element has to be optimized. As the materials requirements demanded by end device manufacturers became more complex, and the number of elements from the periodic table being considered for 'design-in' widened, semiconductor manufacturers better understood the need to share more information about their device challenges with materials experts and adopt a more in-depth, collaborative approach to R&D. In this model, the work starts with all parties understanding the problems that need to be solved.
As the competitive stakes increase, fabs are committing earlier to pairing up with a key supplier or a core group of suppliers to develop more tailored solutions for very specific needs, in less time, and with leaner investments. There are sound reasons for doing this. First, as more materials, and in some cases very exotic materials, are considered for applications in microelectronics, processes needed to develop and scale-up these materials can be very specialized. To develop this expertise 'in house' would not only place a significant financial and resource burden on fabs, but would also take significant time to put in place. Given the highly competitive nature of the end devices markets, it has proven faster and more effective to partner with materials providers who have specific expertise. This collaboration helps the fabs to solve, qualify and scale up their latest materials to manufacturing volume on next-generation devices in a much shorter timeframe. This commitment to a given partner on a given project through a highly collaborative and comprehensive effort is one of the main factors driving speed and success in developing and delivering materials solutions today.
Increased levels of collaboration throughout the supply chain are seeing partners share both risk and reward in the drive towards innovative solutions in R&D. This has resulted in cost of ownership (COO) savings at several points in the new materials work process ??? from process handling to delivery systems to scale-up and final delivery to customers. COO plays an increasingly important role throughout the electronics industry and will only intensify in the future. In this new paradigm, opportunities abound where focused collaborations and materials are the key enablers continuing to drive innovation and enable tangible fab efficiencies.
Another significant factor in reducing the time and costs of traditional R&D is the broader acceptance and proven capability of combinatorial testing methods, which entails highly automated testing of many different materials. Until the late 2000s, combinatorial science, though very well understood and relied upon for decades in other industries, was not widely applied in the semiconductor industry. By the end of 2008, however, it was being applied in a handful of locations in the U.S. and Asia. And by the end of 2010, virtually all of the top-tier chip makers in the industry had experienced the technology in one project or another. The rise in complexity of devices has had a significant effect on the design of experiment (DoE) process in that, as the number of factors considered in the design and functionality increases, the number of experiments rises exponentially. Consequently, the industry has moved from needing to limit DoEs to partial factorial sets to combinatorial techniques, which allows the collection of more comprehensive data with more fully populated DoE matrices. With more variables to consider, more data results in superior decision making and optimized solution creation.
The rise of optimized design
Continuing to follow the trajectory of Moore's Law, the scaling of semiconductor devices continues to move downwards at an increasingly fast pace. According to Intel's May 2011 Investor Meeting, the 14nm node is expected to be introduced to high-volume manufacturing (HVM) in 2014, bringing with it a whole set of new challenges for materials suppliers, tool manufacturers and fabs. Achieving improved device functionality at these nodes is perhaps the major challenge facing materials suppliers. For most process or design engineers, material change can either be seen as a significant differentiator that can improve functionality, or as a detractor from it if certain materials don't support device electrical performance requirements. The sharing of specialist knowledge is essential if the semiconductor industry is to continue to move forward and solve the challenging complexity of manufacturing devices that are increasingly being custom tailored for integration with specific applications. It's very much a moving target that is probably best exemplified in the design of the current crop of portable, multifunctional consumer electronics that operate using low power consumption, delivering significantly longer battery life than previous generations of devices, yet do not compromise desired functionality.
From both a design and a manufacturing perspective, this tailored approach to device fabrication, which increasingly includes the challenges of ever denser packaging solutions for multichip modules in portable applications, in essence ends up defining ever more demanding boundary conditions for the design rules of the device or component fit for its intended use. The demands placed on modern devices and components are driving preferential choices to change the materials employed and/or the manufacturing process used to arrive at a working integrated solution.
Integration, delivery and COO
There are many hazardous materials used in semiconductor manufacturing processes, so the correct handling and storage procedures are critical for ensuring safety, quality and consistency in semiconductor manufacturing. Materials such as TMA and TiCl4 are prone to react and form particles upon exposure to air or moisture, so any breaches during the manufacturing process can cause a potentially serious incident that could compromise the end of line yield of product and the safety of personnel exposed in the fab. Collaboration in this context draws on the materials handling expertise of chemical suppliers to ensure the safe and efficient movement of hazardous materials through the supply chain.
To achieve the most cost effective processing, the most reliable delivery systems are needed. For example, the introduction of precursors to the deposition process must be performed in an exacting manner with sufficient controls to deposit layers a few atoms thick. Because of the high value of modern semiconductor fabs, a shutdown of the manufacturing process can cost millions of dollars per day in lost product revenue.
While the current economic climate has shown optimism, the last industry downturn served to amplify the level of scrutiny on manufacturing costs at every level, with each element of the supply chain considered for economic improvement and COO. This extends to the materials supplied by electronic chemicals manufacturers. As the MOCVD industry, for example, moves into high-volume manufacturing, overall COO has become an important consideration. Reducing the cost of transporting metalorganic chemicals to the process tool plays a significant part in reducing overall COO. In addition, the reliable, flexible and highly controllable dosimetry of metalorganic precursors is of paramount importance in production environments where downtime is prohibitively expensive. When considering the total COO, the delivery system is a critical element that may lead to significant increases in process tool availability. Ensuring better asset utilization at this stage in the manufacturing process, so that tools can be run at full capacity without stoppages, optimizes process efficiency and reduces operational costs without compromising ultra high-performance levels.
The future: turnkey materials solutions programs
For the past three years, we have been working with leading semiconductor manufacturers on numerous R&D engagements employing our own High Productivity Development (HPD) capabilities. Through this pooling of knowledge and sharing more of the overall burden of materials development, deeper levels of collaboration are designed to address future challenges in the face of mounting development costs. Working more closely with key end users in the semiconductor manufacturing field, the program combines combinatorial science discovery techniques with materials expertise and development processes for deployment on specific developmental projects. This approach accelerates cycles of learning, helps discern the most viable solutions quickly ??? as well as which ones to eliminate ??? and generates voluminous data that helps reduce risk.
The work conducted has illustrated dramatic reductions in time and cost for identifying new materials answers for next-generation challenges. And process improvements keep coming. One such recent challenge, which took about four months to complete using traditional methods, went through process design, testing, verification, integration, scale-up and delivery in three and a half weeks. This proved what many had hoped all along; that speedy R&D methods could be process-mapped to materials qualification and delivery to bring faster turn-key options.
Having seen proof and value of materials R&D efficiency value, and now watching it coupled with scale-up and materials delivery to form streamlined, holistic programs, several manufacturers are now receiving the solutions they need ??? and hitting roadmap targets ??? when, previously, they may have still been in R&D mode or forced to go to production with risky or expensive compromises.
John Doering is Director of Deposition Technologies at ATMI, 2151 East Broadway Rd. Suite 101, Tempe, AZ 85282; ph.: 480-736-7609; email: firstname.lastname@example.org
Lou Blanchard is Director of Central Marketing at ATMI; ph. 480-736-7600; email@example.com.