Explore chapters and articles related to this topic
Semiconductor Fabrication
Published in Nassir H. Sabah, Electronics, 2017
The top and bottom parts of the boule are cut off, and the boule is ground to a near-final diameter. Flat regions may be ground along the length of a boule to identify crystal orientation and semiconductor type. The boule is then dipped in a chemical etching solution to remove the damage caused by mechanical grinding. The boule is sliced into circular wafers 0.6–1.3 mm thick using a wire impregnated with diamond particles. Wafer thickness generally increases with diameter, for more rigidity. The wafers are polished on one surface by a chemical mechanical process (CMP) in which the wafer is rotated over a polishing pad that is continually supplied with an abrasive chemical solution, or slurry, of progressively finer grain. Polishing is due to both fine mechanical abrasion as well as chemical action. As a final step, the wafers are chemically cleaned to give a mirrorlike finish that is virtually free from imperfections.
Chemical-Mechanical Polishing
Published in Robert Doering, Yoshio Nishi, Handbook of Semiconductor Manufacturing Technology, 2017
Gregory B. Shinn, Vincent Korthuis, Gautum Grover, Simon Fang, Duane S. Boning
Given that CMP slurries contain many chemical additives, safety handling and environmental issues become important. They could have a significant impact on the CoO of a process; with a corrosive chemical, for example, different components of the system could be affected over time, requiring more frequent hardware replacement and higher downtime. Certain corrosive chemistries even affect platens on polishers by corroding them over long periods of exposure. To conclude, while the polishing performance of the slurry is important in determining the CoO of a CMP step, many of the other factors discussed here can have a large effect on final cost. Using a slurry that is easy to implement in manufacturing may lead to higher uptime, lower problems, and eventually a smaller CoO even though the removal rates may be lower and the actual polish times per wafer longer. These manufacturability factors are often overlooked during evaluations, but become more important, as the process moves to high volume manufacturing.
Overview of CMP Technology
Published in Ungyu Paik, Jea-Gun Park, Nanoparticle Engineering for Chemical-Mechanical Planarization, 2019
The CMP process has a higher possibility of defects than other processes because it uses abrasive in slurry. It especially causes scratches; therefore, controlling the defects is important. To repress scratches attributed to slurry, filter is generally placed at the supply system, circulation loop, and point of use (POU). These factors can be mixed diversely according to the polishing machine’s structure or processing condition selection. However, other materials are also influenced because of the correlation when a factor changes. Therefore, each CMP process should be controlled appropriately because polishing target film and processing can be changed.
Evaluation of chemical mechanical planarization slurry dispersion using a combined scanning mobility particle sizer-optical particle sizer system
Published in Aerosol Science and Technology, 2023
Donggeon Kwak, Juhwan Kim, Taesung Kim
With the Fourth Industrial Revolution (Industry 4.0), various fields, such as the Internet of Things (IoT), artificial intelligence, 5 G, and big data, are developing in novel ways. Accordingly, shrinkage of semiconductors for data processing and storage is required (Ghahramani et al. 2020). Among various processes used for manufacturing semiconductors, chemical mechanical planarization (CMP) is used for the required polishing of wafer surfaces during semiconductor manufacturing, and the number of CMP processes is increasing with decreasing device sizes (Li 2007; Steigerwald, Murarka, and Gutmann 1997). Various consumables, such as the pad, conditioner, and slurry, are used in the CMP process for smooth planarization. Among them, the slurry particularly can affect the performance of the CMP process since it contains abrasives that directly contact the wafer. The slurry contains chemicals to improve process performance and nanoparticles to polish the surface. Dispersion of the CMP slurry is important for ensuring consistency of CMP performance. The dispersion stability of the slurry can be disrupted due to not only gravity, but also the interaction between additives and nanoparticles in the slurry. A slurry with disrupted dispersion stability can cause various problems, such as consistency issues and concentration change due to sedimentation and large particle formation. Therefore, precise evaluation and analysis of dispersion stability are important in slurry preparation.
Reinforcement learning for process control with application in semiconductor manufacturing
Published in IISE Transactions, 2023
Yanrong Li, Juan Du, Wei Jiang
Chemical mechanical Planarization (CMP) is a crucial process in the semiconductor manufacturing process. Virtual metrology systems are often applied in the CMP process to remove the non-planar parts of the films. To simulate the CMP process, we first accept the simulation model in Ning et al. (1996) and Chang et al. (2006), which is defined as: where is the output vector that represents the removal rate and non-uniformity, respectively. is the control vector, that denotes the platen speed, back pressure, polish head downforce, and profile, respectively. According to Ning et al. (1996), the parameter matrices in (9) are assigned as
Stability conditions and robustness analysis of a general MMSE run-to-run controller
Published in IISE Transactions, 2019
Sheng-Tsaing Tseng, Fugee Tsung, Jo-Hua Wu
CMP is an important step in the manufacture of semiconductors. In this section, we illustrate how to use the proposed dynamic qMMSE controller to improve the control performance of a CMP process. Let and denote the removal rate (in angstroms/minute) and the platen speed (the original scale is revolutions per minute (rpm) and is normalized to the range of (-3,3)) of the CMP process for run respectively. Assume that By considering the second-order carry-over effects on both input and output variables and slightly modifying the model proposed by Fan et al. (2002) and Tseng and Chen (2017), we adopt the following dynamic model as the plant I-O model: where is an series with parameters and That is, where We simulated 100 observations from Equation (26), which can serve as the offline stage data. The results are displayed in Figure 2.