Semiconductor Insiders - Podcast EP364: The Broad Impact of Advanced Mask Solutions, Today and Tomorrow with Dr. Germain Fenger

Episode Date: September 4, 2026

Daniel is joined by Dr. Germain Fenger, Senior Director of Product Management at Synopsys where he leads advanced mask solutions and computational lithography technologies. With a background spanning semiconductor manufacturing, lithography research, engineering, and product leadership, Germain has held roles at imec,… Read More

Hello, my name is Daniel Nenny, founder of Semaywiki, the Open Forum for Semiconductor Professionals. Welcome to the Semiconductor Insiders podcast series. My guest today is Dr. Germain Fenger, Senior Director of Product Management at Synopsis, where he leads advanced mask solutions and computational lithography technologies. With a background spanning semiconductor manufacturing, lithography research, engineering, and product leadership, Germain has held roles at IMEC, Global Foundries, Mentor Graphics, Siemens, EDA, and Synopsis. He holds a PhD in Microsystems. He has authored more than 80 publications and holds

multiple patents in the field. Welcome to the podcast, Jermaine. It's a pleasure to meet you. Thanks for having me, Daniel. Yeah, I do remember meeting you. I believe it was maybe 10 years ago, though, so it's been a while. But just to catch us up, you recently joined Synopsis. What brought you to synopsis? Well, I've had a long career, and it started in semiconductor manufacturing, and then move to nanolithography and then eventually RET OPC technologies. And, you know, I really enjoyed my time at my previous companies. And there's a lot of, you know, really nice achievements that the team made there. But it got to a point where I wanted to, you know, be exposed to more new challenges.

in Synopsis had this opening and I took it from there. And I'm really enjoying the time I'm having now at Synopsis. Yeah, I mean, synopsis is very active in this space. That makes sense. So let's just get into it. What problems are advanced mask solutions trying to solve? So at the highest level, advanced mask solutions help semiconductor manufacturers print increasingly complex designs on silicon with the required

accuracy, yield, and economics. So as features become smaller and more complex, the mask is no longer just a passive stencil. It becomes an active part of the imaging solution. So technologies like inverse lithography, curve linear masks in machines. and machine learning help bridge gaps between what designers want to build and really what the manufacturing process can reliably print. Yeah, well stated. So what has changed in the industry that makes topics like ILT, curved linear masks and machine learning especially relevant now? So several things have really converged. One is the compute power has improved dramatically.

So this is especially around GPUs. The multi-beam mask writer has made really more complex mass shapes, more manufacturable. And then at the same time, advanced nodes have much tighter process window. Constrainsen really higher sensitivity to the small errors. So the industry now has both the need and the capability to do more sophisticated mask optimization methods. methods. Interesting.

So where can machine learning create the most value in mass synthesis and verification? So machine learning really is being deployed in a few different areas in mass synthesis. One is the application inside of the models that run inside of ILT or OPC. So as I mentioned, the requirements for the process window and EPE errors are now very small. So having a predictive model is really needed and machine learning helps bridge the gap in the current compact models to be able to achieve the needed accuracy for the N2 node and below. Other areas where machine learning is being deployed and that's how to speed up the whole computational process. So if you can use machine learning to have a first good guess at what the OPC or ILT shapes would look like, you can save a lot in the computational cost of the

whole process. Okay, so how do you distinguish between useful AI and ML applications and hype in this space? So useful AI solves a real customer problem with measurable value. In manufacturing, that means better accuracy, faster turnaround time, improved process margin, lower cost, or more robust sign-off. A hype is when AI is discussed as a generic replacement for engineering judgment. And so the best solutions combined AI with domain knowledge, physics-based modeling, and rigorous validation. Yeah, I definitely agree with that. So what are customers most interested in today? How do they measure the return on investment? And what does success really look like in production? So customers are really focused on technologies that solve, you know, real manufacturing bottlenecks. And so improving process window, reducing hotspots, enabling advanced nodes, reducing cycle time, and managing costs.

They are also very focused on production readiness, whether the solution fits into the existing flow, it scales reliably, and can be deployed with confidence. ROI is measured through yield improvement, process window improvement, reduced rework, faster cycle time, fewer hotspots, and the ability to enable difficult layers or products. Sometimes the value is a direct cost reduction that can be measured directly. Other times it is strategic. So enabling a product that would otherwise be difficult or risky to manufacture. Success means a better way for results. Predictable runtime, manageable mass complexity, successful verification, and clear business value, with the technology becoming part of the normal manufacturing flow

rather than a special experiment. Okay, so if you had to predict one major change in mask technology by the end of the decade, what would that be? Good question. So by the end of the decade, I expect mass synthesis and verification to be much more intelligent and automated. The mask will increasingly be. optimize as part of a larger system that includes design intent, lithography conditions,

mass manufacturing constraints, and fab feedback. So that's a big shift from point tool optimizations to system level optimizations. Yeah, that makes sense. So what excites you the most about where the field is going and what should we all be listening to and what should we all remember? Yeah. So really what excites me the most is the convergence of advanced algorithms, GPU acceleration, curve linear manufacturing capability in AI. We're entering a phase where capabilities that were ones theoretical or too expensive are becoming practical, opening new ways to solve scaling challenges. The key message for listeners is that masks are becoming smarter, more optimized, and more essential to semiconductor scaling. Advanced mass solutions are not just about making better mass shapes.

They're about enabling the next generation of semiconductor products. So just from my perspective, what synopsis products are you involved with? Yeah, so I'm involved in the Proteus line of products. So that's our ILT and OPC and also litho verification products. I'm involved in S-Litho, which is our rigorous lithography simulator. And I'm involved in our MDP products like smart fracture, smart MEC in our smart MPC products. interesting so you know you mentioned GPUs GPUs are very big deal now and Nvidia is really active in this space so is part of the research you guys have done with invidia are you involved in that oh absolutely so synopsis as a whole is a big

partnership with Navidia and inside of say Proteus and as Litho we have enabled GPU acceleration on in video platforms and inside Proteus we We're actually really partnering with NVIDIA to utilize their KULitha library, which is a NVIDIA library based on KUDA that really accelerates our geometric-based operations that are used inside of the OPC technology. So there's a lot of collaboration between NVIDIA and Synopsis in this area. So is TSM also involved in that collaboration? I know they're working with NVIDIA as well, and they're certainly a huge partner with

synopsis. Oh, absolutely. Yeah. So we have a three-way partnership with TSM, NVIDIA, and us. And that's been going on for several years now. And we've, I think, have some pretty impressive press releases based on the Proteus platform at TSM in the video, where our GPU's flows have gone into production there. Yeah, it's impressive. You know, it's all about computational. power and who has more computational power than Nvidia, right? Right. So just final question. How can customers find out more? The best place to start is the Synopsis website, where you can learn more about our advanced mass

solutions, computational lithography, and related manufacturing technologies. Synopsis has a long history of innovation in this space and continues to invest in technologies needed to address the industry's most challenging manufacturing problems. From there, you can connect with synopsis experts to discuss your specific challenges and goals. I would be more than happy to reach out to anyone who is interested in learning more. So I'm guessing we met at a conference. What conferences do you normally attend? Yeah, we actually, I think, met at SPAE advanced lithography, maybe 15 or 10 years ago. That long ago?

It was a while. And yeah, besides that, I usually attend Bacchus, sometimes EMLC, three beams at some point, the litho workshops. So yeah, I've usually go to most of the major lithography and OPC conferences. Yeah, SPIE is my favorite one. You know, not only is it here in San Jose,

but just the best content, the best papers. So I'm sure we'll meet again. Yeah, I was just talking about this earlier, and I don't think I've missed the SVA in the last 15 years or so. And except for COVID, there was one year that it was offline. But besides that one, I've been in San Jose in February for the last 15 years. Yeah, I think it was virtual. You know, that's the one nice thing about being part of Semiwiki is

I get free conference passes. So I go to, I've been left for the last 15 years, I've come into all of the conferences. And I tell you, we just work with some of the smartest people in the world, especially lithography. It's a very complex issue. And I'm telling you, it's just amazing at how intelligent this whole industry is and the people are. So it's quite an honor to work here.

And it's an honor to speak with you as well. And hopefully we can chat again, maybe after the next SPIE. All right. Sounds great, Daniel. Thanks for having me. That concludes our podcast. Thank you all for listening and have a great day.