According to eeNews, Flexciton, a British start-up, is using machine learning to optimize the operation of semiconductor wafer plants to reduce the use of chemicals and energy, so as to make them more sustainable without reducing production.

This is because Samsung has proved the sustainability of its wafer factory process, and the industry organization Semihas established a sustainable development alliance for wafer factory operators. In October 2021, the company raised US $15 million for the production of real-time optimized software semiconductors, and cooperated with global chip companies including Reza Electronics.

"With energy conservation becoming one of the first tasks of factories, by working more intelligently and re-evaluating their production processes, the company is absolutely likely to increase production and output, and improve energy efficiency," Jamie Potter, chief executive of Flexciton, told eeNews Europe.

"Wafer factories usually use heuristic software, in which engineers write rules based on historical data to run production, but there is no built-in intelligence in the software," he said. "The wisdom comes from skilled industrial engineers who use years of experience to adjust and fine-tune scheduling rules to adapt to changing conditions and production requirements."

"The KPI of the wafer factory is mainly based on cycle time, throughput and output, which means that energy consumption has been a secondary consideration in history. However, with the rise of energy costs and the need to reduce energy use due to its impact on climate change, this is now becoming an important KPI, especially because energy consumption can account for 30% of the operating costs of the wafer factory," Potter said.

"To properly optimize the working mode of the wafer factory, we must first accurately understand the status of the entire WIP in real time. By mapping the current status of the operation of the wafer factory, we can determine the location of bottlenecks due to suboptimal scheduling," he said.

Flexciton software already knows how to operate the wafer factory, so it does not need to install the time-consuming rule writing stage for the new wafer factory or replace the existing scheduling software. It has intelligence and pre-programming knowledge, can view data from all tools, and calculate how to run them effectively and efficiently in real time.

Sustainable semiconductor manufacturing R&D plan - IMEC solves the data gap in sustainable semiconductor production

"According to our experience in cooperation with different wafer factories, the tools that often appear in the queue involve the most energy-intensive stages in the production process - for example, lithography, diffusion furnace and dust-free room."

AI optimization of plant data

"Using AI-based optimization software to reduce bottlenecks by improving the way wafers move in energy-intensive tools can meet the main KPI of the wafer factory and reduce the energy consumption of these tools. For example, using fewer tools to do more actions in the photo stage means that some tools may be idle. Or doing the same actions, but reducing the number of batches in the furnace stage means that fewer energy-intensive furnaces run."

Intelligent optimization technology can also be used to directly control the energy consumption of less busy tools. As long as those areas prone to bottlenecks operate effectively and all major KPIs are met, tools in other areas can be optimized specifically for energy conservation.