Modeling Semiconductor Wafer Fabrication Water Quality and Quantity to Inform Treatment Approach Development

Introduction
Semiconductor industry (SI) growth to meet continued world-wide demand for computational, data analytics, data storage, automation, and control by computer processing chips has resulted in greater needs for rapid development of WT (WT) approaches than ever before. Growth in the United States is in part due to the 2022 CHIPS and Science Act (CHIPS) signed into law in 2022 that has fueled over $395 billion in investments in semiconductors and electronics in the US1,2.

Approaches must consider ultra-pure, cooling and production water requirements balanced with water supply and discharge treatment. In areas with water supply and discharge treatment constraints greater levels of water reuse is required resulting in ZLD and brine management. Discharges are characterized as high in organics, nitrogen, sulfates, silica, oxidants, cleaning agents, and TDS, including biological inhibitory and PFAS chemicals. Figure 1 depicts a typical semiconductor fabrication facility water flow diagram with limited reuse and circularity.

The SI is balancing and considering other cultural initiatives such as Environmental Sustainability Governance (ESG), Circular Water Economy (CWE), and Water Energy Food interconnections to their manufacturing for best practices3. Figure 2 shows a typical semiconductor facility with extensive reuse/circularity.

SI requires large quantities of highly pure water and chemicals, resulting hesitancy by the SI to consider source control and recovery of chemicals required in the variety of toolsets needed to produce their chip products. These often-overlooked water pollution prevention and recovery approaches can provide reduced pretreatment and end of pipe (EOP) treatment costs. NPDES and Air Quality discharge permitting from a semiconductor fabrication facility must also be considered when evaluating WT methods. This often is an iterative process between applying and treatment process approach and evaluating the resultant discharges compared to probable water and air discharge requirements.

Opportunities for optimization of WT systems that will be provided with these new grass roots semiconductor will require water modeling tools and expertise to address these issues through the life of the treatment and manufacturing debottlenecking after steady state operation. Based on Brown and Caldwell's work on multiple Front-End-Engineering-Designs (FEEDs) projects for the SI, this paper presents a model developed for 2 projects with basis of designs (BOD), mass balances (MB), and treatment approaches developed with the model that balances a variety of the semiconductor industries objectives, needs and preferences for WT considering the tensions associated cost, permitting, operation, pollution prevention, water and chemistry circularity, energy and sustainability.

Model
While the general SI fabrication sequence of toolsets for quartz and silicon wafers used in the manufacture of electronic devices that was developed in the late 1970s and early 1980s has not changed completely, there has been several process steps, chemistry and cleaning changes added over the past thirty years to incorporate newer technologies, achieve smaller node sizes for more transistors per area, and increase the wafer size to 300 mm3.

The model considers various toolset operation, wet chemistry, water use and discharges based on information provided in literature 4,5,6,7,8, conceptual, and SI FEED projects. Potable water quality, UPW quality, and chemical use can be input as required for the toolsets. Cooling, scrubber and steam water requirements used may also be incorporated into the model. The model will produce a MB, BOD for water quality/quantities from a given fabrication facility UPW treatment, source water pretreatment, reuse/reclaim requirements, and applied chemicals associated with the toolsets including COD, TKN and total nitrogen estimating given provided use. Ion balancing and TDS estimation tools or models are used. Figure 3 shows a block flow diagram for Case Study 1 of 2. Table 1 provides water quality parameter and flows. Table 2 provides Case Study 1 of 2 detailed BOD for EOP WT alternatives analysis.

Results
The HMB information can inform conceptual engineering for treatment approach development. Model output used to prescribe or identify in SI resource recovery/pollution prevention alternative approaches (i.e., acid and alcohol recovery) that can impact end of pipe discharge treatment. It enables treatment sensitivity analyses for azoles, fluorides9, hydrogen peroxide, isopropyl alcohol, PFAS, silica, sulfates, and tetramethylammonium hydroxide (TMAH)10,11.

The MB can be used to inform EOP discharge WT approaches12 from no reclamation and reuse to ZLD13,14 streamlining decision making on COCs, inorganics and TDS treatment approaches. The model assists with decision making on water reclamation both in the fabrication facility considering UPW, process, and cooling water reclamation as well as EOP treatment water reclamation. Figures 4 and 5 show 2 of 8 alternatives developed for EOP treatment for Case Study 1.

Conclusions
Modeling SI WT alternatives with MBs is useful in optimizing WT approaches to meet reuse, circularity, ZLD and EOP WT objectives. This paper will present 2 case studies minimizing treatment requirements for organics, nitrogen, and TDS while balancing the silica, fluorides, sulfates, azoles, hydrogen peroxide, sanitizers and PFAS.