| | www.johnstark.com | |
| |
Tokyo Electron : Ansoft's Maxwell 3D Tokyo Electron is the world's second largest manufacturer of computer chip making equipment. The organization is made of 29 companies organized around seven separate business units covering diffusion, single-wafer deposition, clean track, etch, cleaning, test, and LCD systems. One of its subsidiaries, Tokyo Electron Arizona, Inc. (TAZ) was established in 1998 to focus on metal deposition through physical vapor deposition (PVD) and chemical vapor deposition (CVD) techniques. The facility is located outside Phoenix, AZ, and manufacturers the Eclipse Mark IV PVD product used for semiconductor and other thin film special applications. At TAZ, Jozef Brcka, Ph.D. works as a research and development engineer. As a member of the R&D team, he works on developing and implementing new product concepts, including experimental verification and fundamental diagnostics of new products. He says, "Semiconductor device fabrication requires many technological processes. Our products are designed to accomplish metallization of integrated circuits. There are also a large number of special applications of PVD thin films and TAZ has customers in these areas as well. However, our parent company TEL also focuses on other processes including wet cleaning, plasma etching, CVD processes, and others." Today's semiconductor processing tools are represented by complex systems, which typically consist of many subsystems. The heart of the processing tool is the processing chamber with a plasma source. Generally, supporting subsystems include vacuum systems, electrical power systems, compressed air systems, cooling systems, heating systems, gas supply systems, and others. These systems are crucial, since deposition of high quality thin films requires low contaminant levels in the processing chamber. The process gases used for PVD are typically inert, such as argon, though nitrogen is also commonly used. The gas reactivity is enhanced using the plasma, an electrical discharge condition. The plasma accomplishes efficient transfer of electrical energy to gas components (ions, electrons and radicals) that in turn produce ionized metal needed to form the interconnect structure of a future chip. To generate plasma, electromagnetic fields are employed in a wide frequency range (from hundreds of kHz to tens of MHz). Many plasma sources also include a magnetic field assembly to control the additional properties of the plasma. Brcka adds, "Our design requirements include those imposed by customer process and cost of ownership specifications, as well as safety and reliability requirements. In order to evaluate the effects of the electromagnetic fields on the overall hardware and process performance of the tool, it is very important for us to know how all these components will interact with one another. Simulation tools are of great importance in our pre-design engineering work and contribute significantly to our finding optimal design solutions." Meeting tough challenges "Our approach," notes Brcka, "is to design in satisfaction with all the requirements at the beginning of developmental work, starting with the first pre-design reviews as a part of concept and feasibility work. Redesign work is very expensive and delays the whole development process. We run extensive simulation experiments using Maxwell 3D software from Ansoft Corporation (Pittsburgh, PA) to help answer many questions that could arise from concept and feasibility studies. Our concepts are becoming virtual designs. Then we proceed with real prototypes." Then, the R&D team begins experimental testing, performing prototype diagnostics, verifying, and correcting simulation outputs. Throughout the process, Brcka and his team work closely with TAZ designers. While many university-developed "plasma software packages" are based on physical plasma models, they usually do not allow for easy exchange of geometry. Brcka says that it's not a productive use of his or his colleagues' time to develop new software to meet their specific needs. Instead, they rely on Maxwell 3D, a commercial electromagnetic field simulation software package to meet their plasma needs. Brcka continues, "We leverage our knowledge and understanding of plasma properties by using Maxwell 3D to geometrically treat complex hardware parts of plasma sources within reasonable computing time and with acceptable precision. This integrated approach provides us with powerful and fast feedback to our design work." Maxwell 3D benefits Brcka says the major benefit of using Maxwell 3D electromagnetic simulator is that it allows the R&D team to easily model 3D hardware geometry. "From my own experience," adds Brcka, "many of our simulation cases just could not be solved using 2D geometry. Or, the 2D results provided insufficient validity. Maxwell 3D allows us to quickly solve simple tasks. In addition, with reasonable computing resources, the software can offer solutions for more complex models. Most models can be solved overnight." Brcka is an advocate of multiprocessing and believes that it would be a great benefit in computing speed to implement multiprocessor performance within Maxwell 3D. "I like the built in calculator that allows users to assess needed information from the solution. That feature saves a lot of time used for post-visualization by other programs." The TAZ team agrees that using Maxwell 3D bolsters their confidence levels that the simulations closely match experimental verification tests. Real world project Brcka used Maxwell 3D to develop an antenna and deposition shields for an inductively coupled plasma source. Both components of the plasma source are significant factors in determining plasma and process uniformities. "Their geometry and internal structure play important roles in source performance, as well. The goal was to obtain a desired performance of a source at the lowest cost and complexity. Due to the complexity of the source physics, the analytical or intuitive solutions did not help in this case. So, we needed to perform an extensive feasibility study to find an optimal solution. The dynamic flexibility of the Maxwell 3D modeler simulator, as well as its solver speed and increased PC power allowed us to examine a large number of configurations to satisfy the requirements set by the process technology. We were able to perform studies on scaling hardware components and even post-optimization of the prototype using Ansoft's parametric study and optimization module called Optimetrics. Moreover, during the experimental testing of the prototype, we verified parameters obtained by simulation. I have to say they were very close, which heightened our confidence in the simulator's outputs," adds Brcka. The R&D team acknowledges that Maxwell 3D helps them save overall development time due to the ability to make the right decisions based on the results from predictive simulations and modeling. While tailoring plasma source properties to ever evolving requirements is a complex task, knowledge and intuition in the field are prerequisites for this task. Brcka says that conducting an extensive number of experiments provides the critical information on the sensitivity of process to varying hardware parameters. Yet, parametric modeling and optimization on the computer can be performed virtually before cutting metal and reduce the number of hardware experiments. Brcka believes that the use of Maxwell 3D helps save time and money. "Intense design of experiments (DOEs) through simulations including Maxwell 3D provide valuable qualitative information and provide orientation in developmental engineering work that help us save R&D money. This makes the developmental process much faster, more innovative, and definitively saves overall project costs. The technology also provides a platform for evaluating new innovative solutions." Getting up and running When Brcka began using Maxwell 3D, he experimented with it to determine how it is organized and how it operates so that he could adjust his own procedures to help eliminate time consuming trial and error occurrences. "Using Maxwell 3D, I progressed through the initial learning curve pretty fast," adds Brcka. "I've found that it is useful to use macros effectively and to organize my own DOEs utilizing simulation in such way that speeds the performing of routine tasks. Of course, in our line of work, there is always room for more learning, especially when solving new tasks." Author: Laura Carrabine |