Photo courtesy of Botlagunta Preethish Nandan

Opinions expressed by Digital Journal contributors are their own.

A distinguished expert in AI-driven computational lithography and advanced metrology, Botlagunta Preethish Nandan has recently come up with a new approach to elevate the precision and efficiency of High-NA Extreme Ultraviolet (EUV) lithography and metrology by leveraging Agentic AI, deep learning, and neural networks. In his research paper “Enhancing Chip Performance through Predictive Analytics and Automated Design Verification” published in the Journal of International Crisis and Risk Communication Research, he has demonstrated how predictive analytics and data-driven modeling can fundamentally transform the chip manufacturing process.

Nandan claims that in addition to extending Moore’s law, this innovative framework has the potential to transform semiconductor manufacturing by integrating predictive analytics and generative AI into critical chip fabrication processes. His strategy looks to address critical challenges at advanced technology nodes by achieving dynamic process control, improved yield, and reduced defect rates. 

New frontier in semiconductor manufacturing

With the rapidly growing demand for faster, smaller, and more energy-efficient chips, the semiconductor industry is under severe pressure to scale down feature sizes while maintaining high yield and reliability. Despite technological advancements, traditional lithography methods are not able to meet the demands of cutting-edge nodes because of bottlenecks such as manufacturing defects and metrology limitations. 

Nandan recognizes this gap, and proposes using a next-generation framework to drive breakthroughs in High-NA EUV lithography and advanced metrology leveraging agentic AI. Apart from addressing the longstanding challenges in defect detection and resolution enhancement, this approach dynamically optimizes mask patterning and process control by introducing predictive modeling techniques. 

“The convergence of Agentic AI with High-NA EUV lithography represents a paradigm shift in semiconductor manufacturing, enabling real-time, data-driven decision-making that significantly enhances yield and performance,” says Nandan. “By integrating generative AI models and advanced metrology strategies, we are poised to push the boundaries of what’s possible in chip miniaturization and defect control.” 

Integrating AI-driven lithography with predictive analytics

Nandan informs that the core concept of his research revolves around a seamless integration of AI-powered lithography tools with predictive analytics platforms. As feature sizes approach the sub-2 nm regime, traditional lithography techniques often encounter resolution and pattern fidelity challenges. His AI-enhanced High-NA EUV lithography models address these issues by leveraging neural networks to adjust exposure parameters in real time, optimize mask patterning, and predict potential resolution failures before they manifest.

Nandan complements this dynamic and closed-loop optimization system by real-time metrology tools capable of feeding performance data back into the lithography process. His framework enables predictive process control by combining telemetry from in-chip performance monitors with advanced machine learning algorithms. 

Some key features of the framework include 

• Generative AI-based mask pattern optimization: Generation of optimal mask layouts capable of maximizing feature integrity and reducing defect probability by using transformer architectures and deep reinforcement learning. 
• Dynamic exposure control: improving critical dimension (CD) uniformity and overlay precision by making real-time adjustments to exposure parameters based on predictive analytics.
• AI-augmented metrology: To achieve sub-nanometer accuracy in measurement and defect detection, advanced metrology models trained on vast datasets of pattern deviations and process variances are used. 

High-NA EUV lithography

High-NA EUV lithography is considered to be one of the most effective solutions for overcoming the limitations of current EUV technologies. However, its implementation involves several challenges such as complex pattern corrections, resist blur, and stochastic defects. 

To address these issues, Nandan’s AI-powered framework introduces the following features. 

• Adaptive mask synthesis: Reducing stochastic variations and resist limitation by predicting and correcting mask pattern errors before they propagate by using AI models. 
• Real-time defect detection and correction: minimizing scrap rates and rework cycles by identifying and correcting potential defects during the lithography process through integration of neural network-based models. 
• Predictive yield optimization: Forecasting yield trends by utilizing historical and real-time process data. 

“High-NA EUV lithography, combined with AI-driven predictive models, is the key to unlocking the next era of semiconductor innovation,” Nandan emphasizes. “This approach not only mitigates the risks associated with advanced nodes but also accelerates time-to-market for new chip designs.” 

Looking ahead

Extending beyond technical innovation, Nandan’s research encompasses ethical and collaborative AI models that safeguard data integrity and foster interdisciplinary cooperation.

His approach stresses on 

• Interdisciplinary collaboration: co-creation of robust manufacturing solutions by bridging the gap between digital designers, chip architects, AI researchers, and metrology experts. 
• Data privacy and ethical AI: Ensuring predictive models do not compromise proprietary designs or sensitive process data.
• Continuous innovation: Scaling solutions across different process nodes and manufacturing environments with the help of transfer learning and adaptive AI. 

“The future of semiconductor manufacturing lies in intelligent, adaptive systems that seamlessly integrate AI with process control and metrology,” Nandan concludes. “Our work demonstrates that by embracing Agentic AI, we can overcome the complexities of advanced nodes and usher in a new era of chip performance and reliability.”