The What Is CIM And Explain How Does It Work? That Wins Customers

16/01/2023 Seektronics


What exactly is CIM, what is the use of CIM, the difficulty of implementation and the market situation.

The brain that controls wafer production

 

Traditional solutions are often isolated or loosely connected and difficult to scale for additional requirements, while CIM can integrate it all. IFRS 2007 states that semiconductor wafer integration is divided into five parts: fab operations, production equipment, material handling, fab information, control systems and facilities, and CIM-driven fab operations will be the driving force behind the operation of the other parts.

 

The results of a 1986 Toshiba study showed that using IC-CIM technology to produce 256kbyte DRAM memory circuits improved four manufacturing metrics.

 

Results of a Toshiba study in 1986

 

According to other companies, one year after CIM was put into use, equipment downtime was reduced by 45%, equipment setup time was reduced by 38%, equipment utilization was increased by 30%, cycle time was reduced by 23%, scrap was reduced by 22.5%, product yield was increased by 15%, production costs were reduced by 34%, and net profit increased by nearly 60%.

 

The study found that the earlier CIM is put into use, the better the results. The entire life cycle of fab production is S-curved, and for fabs costing more than $20 billion, there will always be hundreds of millions or even billions of dollars short of the production target without the use of CIM systems, which means that this part of the payback period will be extended, and the earlier CIM is used, the sooner this part of the money can be recovered.

 

 

 

The "S-curve", shows the production target (green line) and actual production (orange line), as well as the various obstacles to achieving capacity and yield targets (gray oval)

 

The concept of CIM was introduced by Joseph Harrington in the book "Computer Integrated Manufacturing" as early as 1973, and it is not only needed in the semiconductor industry, but also in any scenario that requires intelligent manufacturing. Any scenario that requires intelligent manufacturing, it exists, such as pharmaceuticals, food and beverage, medical devices, aerospace, defense and biotechnology, etc., and it has also led the semiconductor landscape change.

 

In the early 1980s, the U.S. economic crisis affected all of society, and electronic products are no exception. Although the U.S. semiconductor in the field of technology development is still strong, but the share of their own products is decreasing, Sony and Panasonic and other Japanese companies began to dominate the storage market and the microprocessor as the next development goal.

 

The time came to the mid-1990s, just a dozen years, the United States and regained the lost market. Politics and strategy aside, SEMATECH (Semiconductor Manufacturing Technology Consortium) was undoubtedly the key point that triggered the transmutation, which officially began operations in 1988 and consisted of the federal government and 14 large semiconductor companies, including IBM, Intel, Motorola, Texas Instruments and other industry giants. With the addition of $200 million per year, manufacturing science and semiconductor process technology began to converge in the United States, with CIM being one of the most critical parts of the development at that time.

 

It is important to understand that at that time, the total cost of a typical high-volume production-ready advanced manufacturing facility exceeded $1 million (equivalent to billions or even tens of billions of dollars today), and even more difficult was the risk of losing yield at each step of hundreds of consecutive processes when IC manufacturing process yields could be as low as 20% to 80%.

 

In 1991, SEMATECH initiated the CIM framework project, which has transformed the U.S. semiconductor manufacturing industry since then, with the support of CIM, chip yields have been effectively improved and product cycle times have been shortened, ensuring product quality and performance. supplier CIM system environment in the semiconductor industry.

 

Looking back at history, just over a decade ago, fab operations still had to rely on workers pushing carts, pushing start buttons in person, and tracking products through spreadsheets, while now wafer production has the ability to integrate data from equipment and automate material handling [16], CIM is undoubtedly the key to bringing smart manufacturing to a new level.

 

Monopolized by two giants for nearly 40 years

 

CIM has a high barrier to entry and is known as the high ground in the industry for industrial software.

 

CIM as industrial software covering all aspects of wafer production, not only requires the developer to have excellent software strength, but also to know each production process, and seamlessly integrate the two together. What is more difficult is that there are many technical secrets (Know-how) in the field of semiconductor manufacturing, and it is difficult to enter the field without senior industry professionals.

 

In addition, CIM is not simply a software overlay, but an organic combination, through a high degree of customization with different manufacturers, different wafer fabs, the original independent operation of multiple unit systems to form a new system that works together, more powerful. At the same time, customer-acceptable fault tolerance is low and software stability needs to be 99.9999%.

 

Even if a CIM system can be made, replacing the old system is not an easy task. To use an analogy, if the round-the-clock semiconductor manufacturing plant as a high-speed aircraft, CIM is the core engine that drives the aircraft to continue to fly, to replace the new CIM system is like driving the aircraft to change the engine. It is clear that the CIM field is very difficult. 

 

In recent years, the rise of 12-inch wafer fabs, driven by the large-scale application of CIM. With the wafer size from 4 inches to 6 inches, 8 inches, 12 inches, not only the amount of investment skyrocketed, but manufacturing equipment, processes, and processes have also become more complex, if this case CIM failure, will be a considerable loss, therefore, the market began to put forward higher requirements for CIM.

 

 

But it is such a difficult industry, the global market is not to say very large. According to Technavio data, the potential market growth share of the entire CIM market (including photovoltaic manufacturing, pharmaceutical, semiconductor manufacturing, etc.) from 2021 to 2026 is $8.72 billion; according to another IDC report, the overall market share of MES in China in 2021 is about RMB 3.81 billion, in this case, the market share broken down to semiconductors may be even less.

 

More embarrassing is that the core of the CIM MES system accounts for only 1% of the total investment in the fab, compared to tens of billions of wafer fabs, it is difficult to trigger the industry's attention, upstream manufacturers prefer to use mature solutions to deal with various problems in production.

 

 

Fortunately, the global new wafer capacity is gradually increasing, the specific customer demand for CIM increased. According to SEMI's World Fab Forecast Report, the global semiconductor industry is expected to invest more than $500 billion in 84 large-scale chip manufacturing fabs that will begin construction between 2021 and 2023.

 

The current semiconductor CIM landscape is highly concentrated, Applied Materials (Applied Materials), IBM and known as the semiconductor CIM duo, the two companies monopolize the market for nearly 40 years. The technology time span of the two companies is also long in terms of development history.

 

Applied Materials and IBM both face the world's most advanced wafer fab or IDM customers, but the focus of their development is not the same.

 

Applied Materials is not only the absolute leader in CIM, but also a leading semiconductor equipment company. After a large number of acquisitions of advanced CIM companies, the company adopts the form of "software + hardware" bundled sales to occupy one side of the market.

 

IBM, on the other hand, is more inclined to AI cloud network combination, through the continuous acquisition of related companies to supplement the technical capabilities, while IBM also has its own fab, can accumulate experience in its own fab trial and error.