<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=574268839740120&amp;ev=PageView&amp;noscript=1">

Smart Backend Assembly Factory for Industry 4.0 – Key Technologies, Challenges and Applications

I recently spoke with Chan Pin CHONG, Executive Vice President and General Manager of Products and Solutions at Kulicke & Soffa, about how smart manufacturing is driving new production efficiencies in the semiconductor industry. During our conversation, he also provided practical steps for factory operators to follow in evaluating their smart manufacturing needs in order to ensure successful implementation and discussed the potential payoffs. Based in Singapore, Kulicke & Soffa is a leading global provider of ball bonding, advanced packaging, wedge bonding, and electronic assembly equipment for the semiconductor, power and automotive industries.

Ng: Industry 4.0 and smart manufacturing are critical to the growth of the semiconductor industry. What does the smart manufacturing movement mean to you or Kulicke & Soffa?

KS Chong 2Chong: The future of smart manufacturing is the vision of building a digital connected factory to drive new manufacturing efficiencies by combining physical and cyber technologies. Industry 4.0 integrates discrete systems and harnesses the power of large volumes of data to move towards greater automation.

At K&S, we define smart manufacturing across the following four key areas embedded in our roadmap for all K&S products, from wire bonders and advance placement tools to pick and place machines:

  • Interoperability – This is about machines, devices and sensors connecting to each other. In fact, the very basis of smart manufacturing is that all devices are connected.
  • Information transparency – Through simulation, various artificial intelligence (AI) tools use contextual information to emulate the actual world.
  • Technical assistance – Robots or machines support humans in making decisions or solving problems.
  • Autonomous decision-making – This is our vision that robots or machines can learn from machines to make decisions on their own.

Ng: Please elaborate on some of these areas and how they’re the relevant to smart manufacturing.

Chong: The need for machines, devices and people to communicate with each other forms the basis of connectivity, the idea of all machines communicating with each other or a host. Connectivity protocols now in place for machine-to-machine connectivity include SEMA, SECS/GEM, SEMI-ELS and IPC-CFX. Machine technology uses various sensing technologies. For example, for a pick and place machine such as SMT platform on K&S Hybrid, the algorithm to recognize and align processes is part of the sensor needed in each machine before to can process and add thousands of components to the substrate or panel. In a wire bonder, the ultrasonics or EFO signal can provide some form of sensing technology for a machine to detect process conditions. Importantly, these sensing technologies can be used to collect feedback for process improvements.

One example of how K&S machines are connected to the host is our use of an intermediate server or host named KNeXt to connects to all assembly equipment in the fab. The equipment can then, in turn, connect to an external secured cloud or K&S Global Cloud.

KS image

Ng: What are the objectives for smart manufacturing?

Chong: The ultimate goal is to achieve higher factory productivity or better OEE (Overall Equipment Effectiveness) by improving machine yields, productivity and efficiency. The key is to leverage AI, 5G, the Internet of Things (Iot) and other industry 4.0 technologies to drive automation and process improvements. Ultimately, each factory must meet productivity, yield and cost goals. Smart manufacturing enables factory operators to meet these goals. That is the focus of smart manufacturing.

Ng: What is the biggest potential benefit of smart manufacturing?

Chong: Smart manufacturing uses data to predict outcomes of a process step or machine operation. Once data is available in the global cloud, analytics can start to build data sets to run statistical modelling and examine factory operation trends. We can also use the data to identify past machine behaviors in order anticipate outcomes, including undesirable ones that we can then prevent.

In the SMT example, if we can systematically examine days or weeks of historical performance, we can plot some statistical variations in the process specifically to a pick or placer or a robot and anticipate or avoid problems. However, all sensors must be in place in the bond head or the robot so that we can detect changes or variations in the robot’s movements.

KS sm facilityKulicke & Soffa smart manufacturing facility


Ng: What are some recent factory improvements smart manufacturing has enabled?

Chong: Kulicke & Soffa has contributed to the hierarchical architecture of the smart factory and key technologies. COVID-19 is driving demand for greater factory connectivity, and K&S offers solutions that are key to remote management and full control of smart equipment from a central control and embedding Internet of Things (IoT), big data, cloud computing and sensors in manufacturing. Using these technologies, a small smart factory can be remotely operated and managed.

With COVID-19 limiting air travel around the world and access to support engineers, the need has grown for remote machine access to reduce the downtime per machine. Remote factory access enables off-site engineers to remotely identify and diagnose machine problems.

Ng: What are barriers to faster adoption of smart factories?

Chong: While most smart factories are capable of network connectivity and data collection, a key challenge is the lack of a business model for smart factories and smart equipment. Most factories must justify major capital investments by demonstrating ROI (Return of investments) potential. Capital improvements for every factory usually take several years to implement and are based on a complex business model.KS PQ Factory connectivity requires substantial investments and years to implement. The same is true of the cloud infrastructure buildouts necessary to generate big data and meaningful analytics. The executive mandate for factory management to install capability usually calls for specific business targets in the planning stage.

Another longstanding barrier to entry is the lack of compatibility of existing tools with new factory protocols, raising the question of whether the cost of replacing legacy tools justifies the need for a smart factory. If new factory investment is required for the latest tools to support the production of new products, the ROI will be much easier to justify.

Ng: How is AI is important in smart manufacturing?

Chong: AI interprets and learn from data to perform tasks and meet specific goals. Good examples of AI implementations include Amazon’s Siri and Alexa voice-command devices and self-driving cars being developed by Google and Tesla.

KS AIAt K&S, over the years we’ve implemented AI in our smart wire bonders to reduce human intervention in our ProCu-7, PSP-2, ProCu Loop 2, Pro Bump and overhang processes.

Thanks to AI, with senses of signals from the bonder, we can reduce the amount of parameters that an engineer or technician have to do trial and error. With on bonder metrology, PBI, loop height, wire sway features, AI allows us to measure process efficiency and provide feedback.

Over several years of AI development, we have leveraged the technology to monitor machines and provide real-time performance feedback in order to provide better closed loop control such as short tail recovery in our bonder process. We can also use the data to predict machine behavior, monitor its health and track maintenance. Ultimately, AI enables fabs to improve manufacturing efficiency, productivity, yields and device quality.

Ng: What’s an example of how AI has solved your manufacturing equipment problems?

Chong: We’ve used AI to set RPM (real time monitor) limits, identify defective P-parts and monitor various conditions such as wire size and capillaries. These types of cases can arise in any manufacturing environment as humans make process mistakes or use the wrong part for a machine. With AI, we can prevent these problems and reduce the risk of further material lost from the wire bonding process.

Ng: What advice do you have for factories looking to implement smart manufacturing systems?

Chong: To build a smart factory, start by focusing on a clear set of business objectives and how smart manufacturing will help minimize or eliminate current factory inefficiencies. In other words, start with the end in mind – the problems that needs to be solved and the business goals – and identify the information you need to demonstrate ROI. Do you need to resolve, automate or improve processes or just to be more efficient? Before investing millions or billions of dollars to build a smart factory, identify those clear goals upfront. Then map out the particulars of implementation to avoid major problems around standards, protocols and connectivity.

Bee Bee Ng is president of SEMI Southeast Asia.

Topics: AI , Artificial Intelligence , microelectronics , semiconductor manufacturing , Smart Manufacturing , Industry 4.0 , Southeast Asia , semiconductor industry , semiconductors , Singapore , return on investment , ROI , Kulicke & Soffa , cloud infrastructure , factory connectivity , machine behavior




Subscribe to SEMI Blog