Lam Research Corporation
08/27/2024 | Press release | Distributed by Public on 08/27/2024 17:02
What Are Digital Twins and How Does Lam Use Them
- Digital twins help increase precision, affordability, sustainability, and speed-to-solution
- Semiverse Solutions helps find the best process flow or recipe out of infinite combinations
The big picture: "Digital twins" and "digital threads" reduce the cost and speed required of working with physical assets and experiments.
- Digital twins: Virtual representation of a process or physical asset that allows designers and maintenance professionals to understand and predict how the process or asset behaves.
- Digital threads: Associated data and processes of record (such as design plans, manufacturing details, and maintenance records) for a process or physical asset that follows its journey from origination to its current state.
Why it Matters: The use of digital twins and digital threads enables designers and maintenance professionals to identify and avoid problems. Lam uses digital twins and threads to enable faster design, deployment, and maintenance of leading-edge equipment and processes-without sacrificing our customers' requirements for precision, affordability, sustainability, and speed-to-market.
Lam's customers can create digital twins using our Semiverse™ Solutions, resulting in lower process engineering costs, greater speed-to-market and better outcomes. Among our software solutions:
- VizGlow™ software performs large, complex, high-fidelity plasma-flow simulations for any semiconductor process, enabling the creation of digital twins that allow chip designers to quickly make informed changes to processes, resulting in faster, less costly design iteration.
- SEMulator3D® models complete process flows and predict downstream ramifications of process changes that would otherwise require costly and time-consuming build-and-test cycles in a physical fab.
Lam's Innovations
Lam innovations enabled by digital twins include:
Virtual builds use software renderings in the design process rather than physical materials. Virtual builds enable the experimentation and refinement needed to build high-quality products while helping reduce cost, time, and materials.
- Example: Lam's Altus® GL was manufactured using a virtual build, which allowed the design team to catch part interference and pinch points that would have necessitated redesign, accelerating the transition to manufacturing by four months.
Training and installation activities use immersive simulations of real-life environments, helping reduce costs and travel time.
- Example: During the COVID-19 pandemic, service engineers couldn't travel. Lam sent customers virtual reality (VR) glasses instead, enabling their engineers-with assistance from Lam engineers using a digital representation of the customer's equipment-to perform maintenance and service operations.
Virtual process development uses digital twins and digital threads to assist in developing chip recipes, which can significantly reduce the time, money, and resources spent.
- Example: The Samsung 92-layer 3D NAND-one of the most challenging etch applications in the industry-features circular memory holes whose depth is one hundred times deeper than width. In a study published in the journal Nature, Lam found that when skilled process engineers begin recipe development using digital twins and threads, then allow computers to take over when the high number of variables require many iterations, higher quality results are achieved quickly and at reduced cost.
Cross-sectional scanning electron microscope (SEM) micrograph of a Samsung 92-layer 3D NAND with memory hole detail (in blue), from TechInsights. (Reference the paper)
Lam's Law
Lam does not simply sell equipment to customers-we sell the equipment with the recipe that meets the customer's specs. That's no easy feat, considering that just as Moore's Law shows transistor size decreasing over time on a logarithmic scale, "Lam's Law" observes the number of possible recipes has scaled exponentially in the opposite direction.
"Lam's Law" is the observation that recipe combinations increase exponentially as chip component feature sizes decrease. (Lam Research)
Lam is ahead of the game: We already use digital twins, digital threads, and the "human first, computer last" approach to help accelerate our time-to-solution in process engineering. Mapping to a cost structure, we find meaningful cost savings:
- Real-world experiments cost on average $1,000 and take on average roughly half a day to complete, but
- Simulated experiments cost on average $0.11 and take on average roughly eight minutes to complete.
Bottom line: Through digital twins and digital threads, virtualization leverages investment in physical assets and real experiments to help save time, money, resources.