Delivering Real-Time Analytics To Semiconductor Test

As defined by Moore’s Law, the semiconductor field has been growing at a steady pace since the 1960s. Concurrent with this progression, semiconductors are becoming more complex, densely integrated, and expensive to produce. While such advancements pose new challenges to semiconductor manufacturing, we can extend Moore’s Law well into the future by reimagining the way we approach the semiconductor manufacturing process.

Fortunately, we have the technologies we need to change our approach. The strategic integration of data analytics, machine learning and artificial intelligence (AI) into semiconductor manufacturing processes will fuel our ability to keep up with the expected pace of the market for many years to come.

Complex semiconductors = Complex manufacturing

Before we get to the solution, let’s examine the roots of the manufacturing challenges confronting the semiconductor industry today.

The forces of Moore’s Law have brought the industry to a space where single chips can consist of trillions of transistors, and process nodes are measured on the order of angstroms. Facilitating these technological leaps has required revolutionary shifts in manufacturing and packaging technologies. A prominent example is 3D packaging technology, which features the integration of multiple semiconductor devices, in the form of chiplets, which are vertically stacked in a single package. More advanced semiconductor nodes, which feature smaller geometries and intricate device physics, coupled with packing higher transistor density per wafer in 3D packaged devices places a greater burden on semiconductor test technology, simply because we’re testing infinitely more complex devices.

Trajectory of Moore’s Law

Complex manufacturing = Complex test

With shorter time to market windows, higher device quality requirements and more complex chip technologies, improving yield while maintaining or reducing costs will keep the economics of high volume manufacturing in check. But these factors increase the test challenges significantly.

The integration of semiconductors into more complex systems has pushed quality standards increasingly higher. In the automotive sector, for example, safety standards such as ISO 26262 have become necessarily rigorous. With the growing adoption of advanced driver assistance systems (ADAS) spanning automatic braking, front- and rear-collision avoidance, lane departure warnings, automated parking and more, semiconductors have become the building blocks for automotive safety. In a future where fully autonomous vehicles are ubiquitous and a semiconductor’s performance may be tied to a matter of life or death, manufacturers can’t afford to sacrifice device reliability.

Today data analytics are playing a role in addressing these challenges. They are being used during the verification and validation process to make better decisions more quickly, eliminating errors before they occur and fixing errors when they happen, positively impacting test times, throughput and device quality.

Archimedes analytics solution

Teradyne Archimedes analytics solution is an open development environment that enables real-time analytics with the flexibility to deliver both out-of-the-box and custom solutions that are easy to deploy.

Our natively secure environment reduces exposure to security risks present with cloud-based solutions. In addition, it provides a structured data stream that delivers information from the tester in a format that’s easy to analyze quickly and efficiently—and that supports the highest level of performance.

The bi-directional capability of this real-time, structured data stream ensures that the feedback loop can be closed by pushing learnings back to the tester for immediate test optimizations during validation and high-volume test. Ultimately, this approach delivers higher device quality and improved business results.

In addition, Teradyne Archimedes allows our customers to deploy the data analytics solution of their choice. Teradyne has partnered with leading analytics providers and the Archimedes platform is tightly integrated with these solutions, ensuring that the focus is on optimizing the test flow instead of the connection between the tester and analytics platform.

Another important component of the solution is the UltraEdge2000, a high-performance and secure parallel-compute platform that enables data consumption and heavy computational processes to be performed in real-time at the edge of the network. This edge component ensures real-time optimizations can be implemented securely by utilizing a zero-trust model and modern software container packaging to protect IP. Containers are deployed to the UltraEdge2000 through secure sessions, and multi-client support enables our customers and their analytics providers to deploy analytic rules and recipes without modifying test programs.

Designed for use with Teradyne’s testers, Teradyne Archimedes offers a unique set of benefits for testing complex semiconductors, including:

• support for a real-time, bi-directional data stream—which creates a closed feedback loop through which data can flow from the tester to the analytics solution, and back to the tester. This real-time data stream allows our customers to utilize their analytics solutions to identify and implement process optimizations immediately, during both validation and high-volume manufacturing, resulting in improved device quality and yield.
• an open architecture—which gives customers the flexibility to choose the analytics solutions of their choice and customize them to best suit their requirements. Teradyne’s tight integration with leading analytics providers ensures quick and efficient deployment. At the same time, Teradyne Archimedes can also integrate with custom solutions.
• secure platform at the edge of the network—with Teradyne UltraEdge2000, customers can realize real-time analytics in a secure environment that protects test data. UltraEdge2000 is a secure platform that executes real-time analytics optimizations with millisecond latency.

Advancing the next generation of chips

Propelled by escalating demand for smarter, faster and more sophisticated semiconductor devices, the forces of progress will continue to push the semiconductor industry forward. AI and machine learning analytics will play a pivotal role in ensuring device quality remains high as time-to-market compresses.