Real-time Cloud Analytics in Semiconductor Manufacturing

Ever since technologies like Artificial Intelligence, Machine Learning, the Internet of Things (IoT), etc have been introduced into industries, the manufacturing sector has undergone a revolution in the mass production of various systems and devices.

Ever since technologies like Artificial Intelligence, Machine Learning, the Internet of Things (IoT), etc have been introduced into industries, the manufacturing sector has undergone a revolution in the mass production of various systems and devices. Manufacturers can achieve this by becoming more efficient, and increasing output while simultaneously reducing additional costs. An essential part of semiconductor production is fault detection and classification (FDC), which enables real-time monitoring and analysis of sensor data, protecting against yield loss and process variations while enhancing product quality and profitability.

Fig 1: Overview of DFD-based systems.

Dynamic Fault Detection (DFD) is a new and next-generation fault detection tool that analyses and evaluates big data using machine learning. However, manufacturers may encounter a number of difficulties in terms of expertise, capital investments, etc. while deploying DFD systems onto a device. The cloud-based DFD system with its database can be seen in Fig 1.

Advantages of using Cloud-based Systems

Cloud solutions come with various advantages that manufacturers can make use of during mass production to reduce cost as well as process high-quality outputs. Some of the features and advantages of Cloud solutions are as follows: -

● Cutting down Hardware/ Software costs: In a general FDC solution, there are several hardware and software implementations needed to be installed onto a new application.  In cloud-based applications, the technology and the related requirements can be stored and executed virtually by cloud vendors. Depending on the size of the system, manufacturers have to pay a subscription cost typically much lesser than traditional hardware or a software-based system.

● Reduction in cost of Resources: If a system's design is sophisticated and extensive, more resources will be required to complete a task during installation or maintenance. In such cases, manufacturers may lack the technical skills to operate a technology based on big data systems in new applications. But, if systems are deployed in the cloud, the cloud vendors can manage and control the supporting big data applications. By doing so, they can reduce the system rollout time and also cut down on hiring costs for new resources.

● System Flexibility: When a solution is deployed on the cloud, it not only saves cost but also makes the system future-proof. Cloud vendors can provide almost unlimited on-demand virtual performance and bandwidth, where platforms of different sizes can be deployed. When a manufacturer needs any changes to be made to the system, the cloud vendor just has to make online reconfigurations to change the capabilities of the system. If the system had been implemented on-site, significantly more work would have been required for different hardware and software upgrades, system modifications, and testing, among other things. This would result in an increase in both the cost and the amount of time needed to complete a given task.

Deploying Cloud-based Systems

The steps for setting up a cloud-based system are similar to those for on-site system deployments. There are five main phases in cloud deployments: 1) Designing, 2) Planning, 3) Installation, 4) Testing & Validation, and 5)Deployment. In order to improve system performance and data security, more time and effort are put into planning and constructing the system firewall as data is continuously analyzed and evaluated in cloud systems. The primary distinction between cloud-based systems is that testing and review of data only occurs within the premises in traditional on-site installations.

If systems are installed on the cloud, deployment time can be reduced drastically by 50 percent when compared to on-site systems. For example, if a first-of-a-kind system was to be introduced on-site, it would have taken at least close to a year, compared to the six-month cloud deployment of the same system.

Various factors to be considered

Even though there are numerous benefits to using Cloud-based systems, there are some concerns about data security and system performance, which may reduce efficiency. When the right tools and techniques are used in combination with a well-planned and built cloud system, the system can deliver much higher and superior performance as well as higher-grade data security. There are several places that cloud architects must closely examine to avoid any compromise on security or performance. Some of them are as follows:

● Data security:  Data security is an important aspect of cloud-based systems, and several steps must be taken to ensure that sensitive data is completely protected in an environment outside the manufacturer's IT domain. The cloud partner must collaborate closely with the client, putting in extra hours to design and implement security measures to protect various types of data. A cloud-based system has four major security components. They are as follows:

Fig 2: Data Security

A.  Authentication: Authentication is a key requirement to gate all/any access requests from any party/user. In order to boost the security of a facility's data, centralized authentication must be implemented across different applications and technologies.

B. Data Encryption: Data encryption must be a standard measure taken to secure communications between two endpoints. A multi-VPN connection can be established for further protection from unauthorized users.

C. Logging/Auditing: Logging of all actions and activities of employees and non-employees must be precisely stored and searchable in a safe repository which allows complete transparency for auditing.

D. Compliance: A trusted cloud provider with global compliance is essential before starting any operations. Cloud providers that have vast resources with high-quality standards over evolving requirements will help and improve the system on a regular basis. Cloud providers must ensure proper due diligence to design a cloud system with all the necessary security and protection regardless of the application provided.

● System performance: Manufacturing applications in any industry must be delivered on time and hence, data throughput and system performance are key elements for a productive and commercialized business. To minimize risk and latency. applications such as FDC systems are deployed on-premise to achieve the best performance.

Fig 3: Latency

Fig 3 shows the pictorial depiction of latency. For example, if a cloud-based DFD system is receiving tool data from manufacturing tools continuously, the data must be analyzed to detect issues and sent back to the system in real time. For the system to respond smoothly, and at the same time avoid any damage to wafers, latency must be shorter than one cycle of a wafer. This implies that an alarm must be delivered before any process starts on the next wafer.

But on the other hand, in a cloud-based FDC system tests have shown that the average latency is just around 5 seconds, which is way lesser than a DFD system measured in minutes. This is hence, a crucial aspect that provides the system sufficient time to respond accordingly when an alarm is issued.

Conclusion

By making use of cloud-based solutions, manufacturers in different industries can attain various technical as well as commercial benefits in increasing the productiveness of their business. There can be a drastic reduction in system deployment time while maintaining high levels of system performance. This also assures a future-proof system with higher flexibility to changing production demands. Modern technologies such as AI and IoT will always be important components in increasing smart manufacturing.