The usage of big data is a key factor for better productivity in semiconductor industry.

Throughout the past two centuries, the industry has witnessed an unprecedented evolution. From the industrial revolution to “Industry 4.0”, the key to the evolution circled around automation, beginning with machine fabrication and continuously giving more focus on making the machine smarter. Nowadays, the key to industrial evolution is to make use of huge amount of data, which gives birth to the Industrial Internet of Things (IIoT).

Big data is the science of computationally analyzing huge amounts of data that is otherwise too complex to be analyzed manually or with traditional methods. A good big data system usually uses data that is large in volume, comes from a variety of sources with complex relations, analyzed in real-time with high velocity and has enough value and relevance to justify the engineering effort to collect it, which summarizes the 4Vs of big data. Such a system is able to use the collected data to identify correlations between different sources that may lead to discovering and fixing production issues. In addition, monitoring the data in real-time can help prevents escapes and therefore enhances the shipped products quality, and the gained info could also be helpful to give insights to engineers and managers towards improving productivity.

Big Data Challenges in Semiconductor Industry

While the usage of big data systems proves to be imperative in semiconductor industry, it also proves to be quite challenging.

The daunting tasks begin with finding and capturing the right data; since it is not uncommon to gather as much data as possible, most of which is not relevant to the production or to enhancing productivity. This, in turn, requires asking the specific target questions that the collected relevant data needs to be processed in order to answer.

Moreover, the amount of data processed and analyzed everyday could easily exceed hundreds of gigabytes, which requires the use of large databases and storage systems. The organization and sorting of such amounts of data are difficult and very important, and the analysis of complex relations between different sources of data, while it could result in very helpful correlations, is also difficult and requires very intelligent algorithms.

Yet, the biggest challenge facing big data systems is that the huge amounts of captured data need to be processed in real-time. This means that the data needs to be available immediately in the database, processed quickly and automatically by the system, and then purged so that there is enough room to receive new data whenever available. Furthermore, the system needs to be connected to the ongoing business activities so that all actions performed by the system becomes timely effective in the supply chain.

Finally, big data processing requires specialist data analytics and/or special algorithms, and complex methods such as machine learning could be put at play to process the huge amount of data and learn the answer to the target questions.

Big Data in yieldWerx Yield Management System

yieldWerx offers a Yield Management System (YMS) that incorporates huge amount of data within semiconductor production process for yield analysis. Using yieldWerx YMS, it is possible to track every product in all its manufacturing and testing stages, storing its history and conditions such as surrounding pressure, temperature, humidity, location on the wafer mapping etc. With this amount of semiconductor data, it is possible to not only pinpoint escapes and defective products and outliers that are likely to cause future reliability issues but also to trace back the root causes of a product defect. This, in turn, allows necessary measures to be taken in order to eliminate these root causes, which enhances the yield and increases profitability.

YMS also uses huge amount of data to provide very useful information about quality and performance, and it produces organized reports for engineers and managers to reach crucial production decisions that lead to a better production experience and enhanced productivity.

Take a 15 days free trial or schedule a demo to uncover the potential of your data when processed by yieldWerx.