Semiconductor testing challenges and advantages of smart manufacturing solutions in the digital era
The ASC white paper explores the significant challenges facing semiconductor testing in the digital age and highlights the benefits of smart manufacturing solutions. Artificial intelligence (AI) and data analytics enable semiconductor manufacturers to extract valuable insights from the vast amounts of data generated throughout the silicon lifecycle. By leveraging AI algorithms, semiconductor manufacturers can optimize silicon design, assembly and test processes. By analyzing huge data sets, AI can identify patterns, predict failures, and optimize quality to improve yields, reduce production costs, and speed time to market.
Semiconductor manufacturers are actively developing data analytics applications to establish fully integrated workflows across the entire semiconductor test ecosystem. State-of-the-art data analytics solutions include critical capabilities such as streaming data collection and control, data feed-forward, low latency, intelligent decision-making powered by ML-driven algorithms, and robust data security and integrity capabilities. This enables end-to-end exploitation to be analyzed throughout production and testing.
This white paper explores the key challenges facing semiconductor testing in the digital age and highlights the benefits of smart manufacturing solutions that increase efficiency and optimize production processes for semiconductor customers. These solutions include ACS Real-Time Data Infrastructure (RTDI) and machine learning-driven analytics solutions from Advantest and its ecosystem partners, creating a digital highway for seamless integration of all test data. By leveraging low-latency edge computing and analytics solutions, real-time monitoring of semiconductor test operations becomes possible, enabling rapid corrective action. This improves quality and yield and shortens time to market for customers.
The current state of the semiconductor industry in 2023
The semiconductor industry is fundamentally a thrilling, high-stakes game of cat and mouse.
This is a world of cutting-edge technology, where the lines between science and science fiction are often blurred. In this industry, the largest players hold nearly all the cards, but new players are constantly emerging to disrupt everything.
In the semiconductor world, the line between success and failure is razor-thin, and the future is always uncertain.
With the development of new technologies such as 5G, Internet of Things (IoT) and artificial intelligence, the demand for more advanced semiconductor devices and solutions is also growing.
Integrated circuits reflect a relatively new world order in which trade tensions between countries and supply chain disruptions are ongoing threats, and the fallout from the pandemic continues to impact the industry. Despite these challenges, the integrated circuit industry continues to move forward, driven by a relentless pursuit of innovation and advancement.
This industry constantly challenges the boundaries between the possible and the impossible. This is a world of innovation and endless possibility, where the brightest minds and boldest entrepreneurs can change the course of history.
The risks are high, but so are the rewards.
The future is always full of variables, and the only sure thing is that nothing is certain. In the end, only the strongest and smartest will survive.
Industry Forecast: 2023 and Beyond
With the development of big data and advanced technology, people have already felt the impact of the rapidly developing digital highway.
According to one forecast, the semiconductor market size in 2030 will be approximately US$1.1 trillion, with a compound annual growth rate of 9.03% from 2020 to 2030, and a compound annual growth rate of only 3.96% from 2030 to 2030.
The increase in software content in electronic products will also have a positive impact on the growth of the semiconductor market.
This rapid market growth will have a significant impact on testing, including the explosive volume of test data and methods for handling this data volume:
"The volume of data generated per day in 2019 was approximately 2 terabytes," a report said, adding: "Test data volume increased dramatically from 10nm to 7nm, and will increase again with each new node. .The report attributes this growth in part to more chips per wafer, but also more transistors per chip, increasing design complexity, and increasing process corners. , new processes, testing, diagnostics and assembly are also increasingly complex.
Part of the reason for the data explosion is the increasing number of test vectors required to test the ever-increasing number of transistors on each chip (Figure 2). The industry is paying more and more attention to data analysis. Increasingly, the data generated during the data testing process will focus more on data analysis to gain insights and improve decision-making. Complexity of Semiconductor DevicesContinuously increasing, this will drive the application of artificial intelligence and artificial intelligence in semiconductor testing. Artificial intelligence and ML can enable more advanced testing methods and equipment, helping to improve efficiency, accuracy and scalability.
Data feedforward and data feedbackThe use will become more and more common, helping to improve the accuracy and efficiency of semiconductor testing.
Data and edge computingSemiconductor testing will be enhanced, making test methods more flexible and scalable.
market growthAlso added will be test stress application areas, packaging strategies, environmental issues related to edge applications, and network security:
For autonomous vehicles, IoT and 5G devicesGrowing demand will drive the need for more advanced semiconductor testing methods and equipment.
heterogeneous integrationThe increasing use of integrating multiple technologies and materials in semiconductor products has created a growing need for test methods that can handle this integration.
Test EquipmentIt must be flexible and able to handle a variety of environmental conditions, and test data should be provided in a distributed network rather than being restricted to a central database.
Devices under test will contain more and more sensors, the data generated by these sensors needs to be harnessed to improve product quality, reliability and efficacy. In addition, test equipment will have more embedded sensors that will be used to monitor and control test operations in real time.
As semiconductor devices become more complexthe connection is getting closer and closer, and we will need to pay more attention to network security to prevent potential network threats and ensure the integrity of intellectual property during the testing process.
Finally, new business models are likely to emerge in the semiconductor test industry, such as pay-per-use and subscription-based services, and advances in automation and robotics will continue to be more widely used in semiconductor test, helping to increase efficiency and reduce human error. .
Key challenges facing the industry
In addition to challenges related to heterogeneous integration, 2.5D and 3D packaging, test data volumes and service-based business models, the complexity of semiconductor devices continues to increase, requiring faster test times and higher throughput to keep up. Rapidly growing demand for semiconductor devices; the need for more accurate and precise testing as semiconductor devices become denser and more complex with tighter tolerances; and flexible and open solutions to test various devices using different source technologies. device.
In addition, the test system must be able to test multiple technologies on the same platform, such as digital and analog technologies; be able to handle more device types and package sizes; be able to test at different temperatures and voltages; be able to test new materials and new structures , such as: 2.5D and 3D packaging. Finally, in the era of big data and data computing, strategies must be developed to address the challenges of data privacy, security, and compliance.
The challenges of big data are daunting. It is estimated that 80% or more of the data collected throughout the semiconductor supply chain, from design to manufacturing to the field, is never viewed. Contractual obligations, particularly in the automotive and aerospace industries, may require data to be archived for 10 or 15 years, but engineers often only view the data needed to solve a specific problem. Additionally, some stored data lacks traceability and context, making it useless. This is not to say that data has no value.
However, the tools and infrastructure to effectively analyze data and extract maximum value from it currently fall short. Everyone is desperate to mine their own data, but they simply can’t.
Another problem is that current testing is typically done with file-based methods, where data is only available at end-of-wafer or end-of-batch, and only then can the data be analyzed offline, often at a different location for analysis. Even as time passes, it often takes hours or days to perform analysis and corrective actions.
With the rapid development of the testing industry, this approach is no longer feasible. As mentioned previously, chips are becoming increasingly complex, and in many cases (including medical and automotive device testing), zero defects is the goal. In other cases, even tiny flaws that were acceptable in the past now render more sensitive, complex chips useless. As a result, it has become a challenge to test these more advanced chips while meeting tighter volume production time targets and volume production time targets that ensure integrated circuit manufacturers remain competitive.
One company reported success using online infrastructure methods to reduce latency from hours or days to milliseconds or microseconds. This approach enables instant ML scoring and decision-making while maintaining secure communication.
The method applies neural networks to parameter predictions, taking into account the correlation of temperature, power and other conditions, thereby significantly reducing test times with zero impact.
ADVANTEST ACS: Solutions to Semiconductor Test Challenges
Leveraging advanced analytics, including ML capabilities, and future-proof, real-time automated production control, the ecosystem addresses test challenges by integrating all data sources across the integrated circuit manufacturing supply chain.the ecosystem
Enable customers to implement intelligent data-driven workflows and help customers achieve better yields, faster time to market, faster time to volume production, and higher quality and reliability. As part of Advantest's Big Design strategy, ACS is enhancing edge and data infrastructure services, data analytics and AI/ML solutions to help customers achieve data-driven workflows.
Advantest's data-based products and services are based on a single scalable real-time data platform, enabling customers to develop or source market-leading solutions from Advantest and its partners. ACS uses real-time ML analytics technology to enable customers to quickly turn insights into production-worthy actions. These rules and procedures are easy to use and accessible throughout the semiconductor value chain.
Advantages of advantest ACS
Given that current testing methods are costly, cumbersome and time-consuming, the industry needs a new model. To meet this need, Advantest created the industry's first Real-Time Data Infrastructure (RTDI)—an innovative, revolutionary approach to delivering analysis, corrective analysis, and action in an ever-changing test environment (Figure 4). Online edge computing analytics offers the following advantages compared to file-based analytics:
Executed in a secure True Zero Trust environment, detecting issues and taking corrective action takes just milliseconds. In this way, ACS RTDI provides real-time, adaptive decision-making and its solution is easily integrated into any test program and enables customers to maximize the Advantest value chain to improve throughput, quality, time-to-market and volume Production time.
ACS RTDI serves as the communications backplane for the test platform, facilitating fast, secure and accurate information exchange. It enables online edge computing/analytics to enable real-time adaptive decision-making within and between touchdowns.
Real-time data monitoring with intelligent data extraction and control enables corrective action to be initiated in milliseconds.
ACS RTDI includes several specific key capabilities and technologies
ACS Container Hub Using container technology, automatic software distribution of containerized applications is achieved. The product ensures easy, secure, and reliable deployment of applications into test fleets while ensuring seamless integration into the ACS ecosystem.
ACS Edge Server is a high-performance, highly secure edge computing and analytics solution that delivers millisecond latency to facilitate real-time adaptive decision-making during test execution.
ACS Nexus Provides a standard interface for real-time test unit data flow and online device control between all Advantest platform devices and external clients, enabling online analysis functions, real-time control and across different test stages. It enables applications to maximize yield, optimize throughput and ensure test quality without increasing the risk of failures on the production line.
ACS Unified Server supports scalable, redundant compute and storage resources to enhance workload management and minimize the risk of downtime, while providing True Zero Trust security for the test platform. It also supports data feed forward/ feedback, and provides transaction-level security, cross-platform support, multi-party sharing and data management, while supporting packet inspection and logging.
Together, these solutions help customers turn data into real-time production controls, delivering superior results and higher ROI for their products.
Review Editor: Huang Fei
#Semiconductor #testing #challenges #advantages #smart #manufacturing #solutions #digital #era
- Microchip further expands its mSiC solutions with the introduction of 3.3 kV XIFM plug-and-play mSiC gate drivers,
- What is NAND type Flash memory?
- Data leaks can sink machine learning models
- What does the AC voltage regulator indicate when it is overvoltage?
- Adjustable voltage stabilized power supply circuit diagram based on LM317
- Demystifying Qualcomm Domain Controller Level 1 Power Design: Power Supply Design and Computation
- EU consumers challenge Meta paid service as privacy ‘smokescreen’
- Qualcomm Snapdragon
- Perform complex motor drive and interactive control testing with the MDA8108HD Motor Drive Analyzer
- ‘Iron Man’ pilots race in jet suits against a backdrop of Dubai skyscrapers