Advantest Corporation is offering a new yield-improvement solution that leverages artificial intelligence (AI) to expedite identifying the root causes of yield loss and increasing the efficiency of analyzing test results. The innovative and scalable Advantest Cloud Solutions Engineering AI Studio for Yield Improvement (ACS EASY™) can increase the productivity of both device engineering and production operations for a wide range of users, from chip designers to outsourced semiconductor assembly and test (OSAT) companies. Device yields are key performance indicators in semiconductor production, requiring continuous engineering efforts to debug and fine-tune.

The unique ACS EASY application uses AI to automatically monitor test conditions and inferences to isolate and analyze the causes of yield degradation. This resolves production issues in a timely fashion, slashes troubleshooting time and dramatically reduces test workloads for data analytics. ACS EASY is capable of handling huge volumes of data to compare new lots' test results with those of previous lots to quickly identify abnormal bin trends.

The solution's GUI facilitates the online sharing of test results, eliminating the need to create separate reports. Moreover, ACS EASY is intuitive to use and does not require operators to have familiarity with AI, machine learning, data analysis or statistics. A leading Advantest customer evaluated ACS EASY's ability to automatically analyze data from its test floor, a task that used to consume 100 engineering hours per day.

The new system reduced the time needed to identify the causes of yield impact by 80% and successfully detected multiple yield problems and causes that had been previously overlooked. ACS EASY utilizes prescriptive self-learning to categorize new yield-related issues for future monitoring and analysis. This further extends the system's stored knowledge base, allowing inferencing applications to present bigger values.