Currently, the global artificial intelligence industry is entering a critical phase of large-scale implementation and coordinated development across the entire value chain. From the iterative development of generative AI large models to the intelligent transformation of industries across all sectors, AI has become a new form of productive force driving the deep integration of the digital economy and the real economy. In this technological revolution, AI chips serve as the core carriers of computing power, and the completeness and sophistication of their supply chain significantly determine the upper limits of industry development. As the fundamental backbone of semiconductor manufacturing, high-performance new materials play an indispensable role in the precision production processes of chips.
I. What Are AI Chips?
AI chips are computational units designed to process AI operations. Unlike traditional general-purpose CPUs, their key advantages lie in their strong parallel computing capabilities, efficient matrix operations, and low power consumption. They are capable of efficiently performing critical AI tasks such as machine learning, deep learning, data inference, and image recognition. As the primary hardware platform for delivering computing power and enabling AI functionality, AI chips are a key factor in the competition within the AI industry.
II. Structure of the AI Industry Chain
The AI industry chain is a comprehensive ecosystem spanning technology R&D, manufacturing, and application scenarios. It is broadly divided into three major segments: the upstream foundational layer, the midstream manufacturing layer, and the downstream application layer.
(1) Upstream: Foundational Support
The upstream foundational layer serves as the bedrock of the AI industry, providing technology R&D and key raw materials. It can be roughly divided into two segments: first, hardware infrastructure, which includes lithography machines, silicon wafers, and high-performance computing servers; Second, data services—such as data collection and filtering—which serve as the “fuel” for subsequent large-scale models.
(2) Midstream: Technology and Manufacturing
The midstream manufacturing layer is the production hub of the AI industry chain and serves as a vital link between the upstream and downstream sectors. It can be divided into two major segments: algorithms and models, and chip design and manufacturing.
1. Algorithms and Models
This field covers a wide range of topics, including visual algorithms, speech processing algorithms, and machine learning methods. The goal is to provide AI with a methodological framework for processing data. Models, on the other hand, are the specific results obtained when algorithms learn from specific datasets. The current major trend is to focus on large-scale models, endowing them with the ability to plan, remember, and use tools so that they can autonomously complete complex tasks.
2. Chip Design and Manufacturing
Design aims to ensure that chips effectively integrate the three key areas of architectural definition, hardware implementation, and software coordination, while achieving an optimal balance between performance, power consumption, and cost.
Manufacturing can be further divided into two stages: wafer fabrication and packaging and testing:
(1) Wafer Manufacturing: This is the process of transforming high-purity silicon wafers into bare wafers with complete circuit structures through dozens of nanoscale precision processes, including photolithography, etching, thin-film deposition, ion implantation, cleaning, and polishing. AI chips demand extremely high manufacturing standards. Mainstream high-end products utilize advanced processes of 7 nm and below, while next-generation products are gradually advancing toward 3 nm and 2 nm. This places stringent requirements on the production environment, process precision, and material compatibility: production facilities must meet Class 10 to Class 100 cleanroom standards to prevent contamination of wafers by microscopic dust and impurities; process tolerances must be controlled at the atomic level to prevent circuit defects; simultaneously, the production process involves high-temperature, high-pressure, and highly corrosive conditions, placing extremely high demands on the weather resistance and cleanliness of auxiliary carriers, protective materials, and production facilities.
(2) Packaging and Testing: The packaging process primarily involves dicing, thinning, bonding, molding, and lead soldering of wafers to provide bare chips with a protective casing, fulfilling three key functions: physical protection, circuit connectivity, and efficient heat dissipation. The testing phase spans the entire process—from post-wafer fabrication through packaging to post-packaging—and includes wafer probe testing, chip performance testing, reliability testing, and power consumption testing. Professional equipment is used to screen out non-conforming products, ensuring that chips meeting quality standards are shipped. The testing process for AI chips is more complex and demands higher precision; the wear resistance, insulation properties, and accuracy of test fixtures and carrier components directly impact testing efficiency and the accuracy of results.
3.Downstream: Application Deployment
The downstream application layer serves as the “value outlet” of the AI industry, encompassing a full range of scenarios such as intelligent computing centers, industrial intelligence, autonomous driving, smart cities, smart healthcare, and fintech. By integrating AI chips, it drives the intelligent transformation of various industries. From training large models in the cloud to inference on edge devices, the demand for computing power is growing exponentially, further driving capacity expansion and technological upgrades in the midstream wafer manufacturing and packaging and testing segments.
III. Applications of Plastic and Carbon Fiber Products in AI Chip Manufacturing
The extremely harsh operating conditions in wafer fabrication and packaging/testing require supporting auxiliary materials to meet key criteria such as high-temperature resistance, high insulation, corrosion resistance, low deformation, high purity, no impurity leaching, and dimensional stability. Conventional materials often fail to meet these demands; Taisheng provides high-performance plastics and carbon fiber products that are suitable for these production standards.
1. Plastic Products
(1) Cleanrooms: Throughout the production process—from monocrystalline silicon production to integrated circuit manufacturing and packaging—all operations are conducted in a clean environment. Cleanroom panels typically use flame-retardant materials and materials that do not easily generate static electricity, while window materials must also be transparent. Suitable materials include: anti-static PVC/PP;
(2) CMP Retaining Rings: Chemical mechanical polishing (CMP) is a critical process in wafer manufacturing. The CMP retaining rings used to secure silicon wafers are particularly important components that must exhibit excellent wear and corrosion resistance to prevent damage to the wafers. Suitable materials include PPS, PEEK, and others;
(3) Wafer Carriers: Common wafer carriers include wafer boats and transport boxes. The stability of the environment during wafer transportation and storage significantly impacts wafer quality. Therefore, wafer carriers must possess properties such as temperature resistance, antistatic properties, and low outgassing. Suitable materials include PP, PEEK, PC, PEI, etc.;
(4) Components such as bearings and guide rails: Components of semiconductor processing equipment, such as bearings and guide rails, must be capable of continuous operation across a wide temperature range (from low to high temperatures), exhibit low wear and low friction, and maintain dimensional stability. Commonly used materials include polyimide (PI), etc.
2. Carbon Fiber
During the wafer manufacturing process, wafers must be transferred between different workstations, necessitating the use of wafer forks. Carbon fiber is an excellent material choice for these forks. Carbon Fiber employs an impregnation and pressing process, resulting in more stable performance. It offers a tensile strength of up to 6,000 MPa, a material modulus exceeding 780 GPa, vibration damping that can be controlled within 4 seconds, and excellent weather resistance.
The high-quality development of the artificial intelligence industry relies on coordinated efforts across the entire industrial chain, and the midstream wafer manufacturing and packaging and testing segments are among the key areas for the industry’s large-scale implementation. HONY PLASTIC focuses on high-performance plastic and carbon fiber products, providing the semiconductor industry with suitable components that meet its evolving needs.
The 5 Major Applications of Plastics in the Wafer Production Cycle
When discussing semiconductors, the topic of wafers—the foundation for manufacturing various computer chips—always comes up. As semiconductor technology continues to advance toward smaller line widths, higher integration, and more complex structures, the quality requirements for wafers—the “foundation” of the process—are constantly increasing. Against this backdrop, plastic materials, with their excellent packaging and transport capabilities, have become essential for connecting various process steps, reducing contamination and mechanical damage, improving cleanliness, and boosting overall yield. Let’s take a look at some common applications of plastics in semiconductor manufacturing.
1. CMP Retaining Rings
Chemical mechanical polishing (CMP) is a critical process in wafer manufacturing used to achieve global planarization of the wafer surface. During this process, the silicon wafer must be securely held in place by a retaining ring to ensure uniform polishing and prevent displacement, thereby avoiding scratches or contamination on the wafer surface. Therefore, the material selected for this component must possess wear resistance, high dimensional stability, good chemical resistance, and machinability.
In the past, polyphenylene sulfide (PPS) was commonly used to manufacture clamping rings; however, polyetheretherketone (PEEK) and chlorinated polyvinyl chloride (CPVC) are increasingly being adopted by manufacturers due to their higher mechanical strength, excellent dimensional stability, and superior chemical and wear resistance.
2. Wafer Carriers
Wafer carriers are used to hold, store, and transport wafers during the manufacturing process. Common types include front-opening wafer carriers (FOUPs), wafer transport boxes (FOSBs), and wafer boats. Storage accounts for a significant portion of the wafer production cycle. Therefore, material selection is critical, as the cleanliness and antistatic properties of the carriers directly impact the quality of the finished wafers.
Materials for wafer carriers must meet requirements such as high-temperature resistance, high mechanical strength, low moisture absorption, antistatic properties, low outgassing, and low leaching. Polyetheretherketone (PEEK), perfluoroalkoxy resin (PFA), polypropylene (PP), polyethersulfone (PES), polycarbonate (PC), and polyetherimide (PEI) are all common materials that meet these requirements.
3. Photomask Cassettes
A photomask serves as the pattern master in the photolithography process, typically consisting of a quartz glass substrate with a chrome-plated pattern to block light. Any particles or scratches on its surface can cause defects in the photolithographic pattern. To accurately transfer the circuit pattern from the photomask onto a wafer coated with photoresist, maintaining the cleanliness of the photomask is critical.
As a storage and transport container, a photomask box must possess properties such as antistatic properties, low outgassing, high rigidity, and abrasion resistance. Polyetheretherketone (PEEK), due to its high hardness, low particle generation, high cleanliness, and antistatic properties, is an excellent choice for photomask boxes. It effectively prevents damage to the photomask caused by fogging, friction, or vibration during storage and transportation, while providing a clean environment with low outgassing and low ionic contamination. Antistatic polycarbonate (PC) is also used, but its overall performance is slightly inferior to that of PEEK.
4. Wafer Handling Tools
During the manufacturing process of wafers or silicon wafers, tools such as wafer holders and chucks are used for gripping or moving the wafers. Since these tools come into direct contact with the wafer surface, it is essential to prevent scratches or residue from forming, as these can adversely affect device performance and yield.
Polyetheretherketone (PEEK), perfluoroalkoxy resin (PFA), and polypropylene (PP) are widely used in the manufacture of wafer handling tools due to their high heat resistance, excellent wear resistance, good dimensional stability, low outgassing rates, and extremely low moisture absorption. These materials minimize surface friction and particle residue, significantly improving wafer surface cleanliness and integrity.
5. IC Packaging Test Sockets
Test sockets connect chips to test equipment and are used to verify the functionality of integrated circuits. Different types of integrated circuits require test sockets with corresponding specifications. Material requirements include high dimensional stability, good mechanical strength, low burr generation, long service life, a wide temperature tolerance range, and good processability.
Engineering plastics such as PEEK, PPS, polyamide imide (PAI), polyimide (PI), and polyether imide (PEI) are widely used in this field.