Fabs & fabless
Control processes and improve yield
Total cost: a better balance
Use more data while reducing overall cost.
Without PollenWith Pollen
- Faster root cause analysis
- Better control of process windows
- Lower losses and improved yield
Collaborative AI · Process control softwareFrom industrial data
From industrial data
to process understanding.
Smart3 connects your data, expertise and AI, from R&D to production.
Substrate
Deposition
Patterning
Etching
Smart3 connects your data, from R&D to production.
↗ On-premise deployment↗ Your data, your infrastructure↗ Your algorithms, your intellectual property↗ R&D → Production
Smart3 Software Platform
One platform. Your expertise. All your data.
Smart3 brings together the data involved in manufacturing innovative materials, organized into six broad categories.
ImagesPoint cloudsDesignsSpectraTime seriesTabular data
On-premise installation. Data confidentiality for every customer.
Smart3 runs on your infrastructure: you retain control of your data and know-how, from R&D to production.
Smart3Collaborative AI Platform
01
Connect
Multi-source data
02
Analyze
Workflows · AI · Simulation
03
Decide
Measurements · Defects · Processes
Your gains with Smart3Improve yield.
Improve yield.
Accelerate equipment innovation.
Two industrial challenges, one platform to turn data into value.
Equipment makers
Create more value around your equipment
More value, less shared software development
Create your own IP on Smart3 and retain ownership of your developments.
Without PollenBuild everything
Software foundation
Data & connectors
Business applications
Maintenance
With PollenDifferentiate
Your IPYour expertiseApplications & algorithmsIntellectual property created and retained by you.
Smart3 · Foundation maintained by Pollen
- Faster time to market
- Less shared software development and maintenance
- Resources focused on your added value
Conceptual comparisons: gains depend on data, processes, integration scope and use cases.
Smart3 · CapabilitiesOne software platform.
One software platform.
Capabilities that work together.
Smart3 combines your data and analytical capabilities in one environment: metrology, inspection, simulation and optimization work together to understand your materials and processes.
Metrology & characterization
Extract measurements and characterize structures using multi-source data.
Inspection & defect analysis
Detect, segment and classify defects within your industrial context.
Simulation & process optimization
Connect experimental data, manufacturing parameters and physics-based models to inform decisions.
Smart3 · Natural-language customizationTurn your expertise into a workflow.
Turn your expertise into a workflow.
Without writing code.
Describe your needs in natural language. With an LLM and documentation context supplied through RAG, create data readers and combine your plugins in Smart3.
Describe → Create → Compose
Watch the AFM & EBSD demo ↓Your request
LLM
RAGDocumentation context
MCP
Smart3Plugins & workflows
Towards Agentic AI
Smart3 · Simulation & experimental dataSimulation meets real-world data.
Simulation meets real-world data.
A continuous learning loop.
Simulation becomes more useful when tested against reality. Smart3 connects simulation tools with characterization data to compare results, refine models and guide the next experiments.
Illustrated example: plasma etching — from simulation to SEM/TEM cross-sections and measurements in Smart3. The same loop also applies to other processes, substrates, components and materials.
Smart3Data · Models · DecisionsWafers · Substrates · Materials
01DOE & recipes
02Process
03Characterization
04Results & analysis
05Simulation & optimization
- 01
DOE · Design of experiments
Define objectives, factors and experimental conditions.
- 02
Process
Run experiments using the selected recipes and parameters.
- 03
Characterization
Measure structures, materials and defects using SEM, TEM, AFM and other sources.
- 04
Results & analysis
Connect measurements with parameters, compare performance and investigate discrepancies.
- 05
Simulation & decisions
Compare models with reality, explore scenarios and guide the next DOE.
Back to DOE: each cycle informs the models and helps select the next experiments. Smart3 connects data and tools throughout this loop.
Improve yields
Connect measurements with parameters to identify performance drivers.
Accelerate development
Focus experiments and connect simulation with experimental data.
Reduce manufacturing costs
Automate analysis and make better use of data to guide decisions.
From your first use case to deployment.
Train your teams and explore your data.
Assess a focused use case and its feasibility.
Develop and validate your application.
Integrate and deploy in production.
Expand applications and continuously improve.
Working with Pollen
A transparent model, tailored to your needs.
Start with a concrete application, then expand Smart3 as your needs evolve. Your proposal clearly separates licenses and services, with scope and costs agreed together.
Licenses sized for your deployment
Based on the number of licenses and how they are used: R&D teams, production sites or integration into your equipment.
The level of service you need
Training, support and custom development: we tailor our involvement to your resources and your goals for autonomy.
A gradual, controlled expansion
From your first project to broader deployment, you have visibility into the costs of each stage.
Local partners
Support closer to you.
Japan
Keisoku Engineering System Co., Ltd.
Your local partner in Japan to discuss Smart3 and your applications.
Visit the KESCO website ↗Website currently under maintenance.South Korea
Woowon Technology Co., Ltd.
Your local distributor in South Korea to discuss Smart3 and your applications.
Visit the Woowon website ↗Intelligence for your industrial environment.
Semiconductors
From process development to production monitoring: make use of metrology and inspection data.
Advanced packaging
Analyze structures, interconnects and defects. Connect characterization data to manufacturing parameters.
Displays & photonics
Characterize microLEDs and optical structures. Bring measurements and simulation together.
Advanced materials
Build analysis workflows to understand the relationships between structure, properties and process.
Smart3
In your fab
Connect data, analyze results and guide process decisions.
Smart3 Data · AI · Workflows · SDK
Smart3
In your equipment
Integrate your algorithms and enhance your tools’ capabilities.
Smart3 · OEM & SDK
One platform. Two integration paths.
Save years of software development.
Fabs, fabless companies and equipment suppliers: build your domain applications on Smart3, an open, customizable and unified software platform.
Build on ten years of core technology development to integrate your data, algorithms and workflows while retaining your know-how and differentiation.
One platform, from R&D to production.
Harmonize tools and methods across teams, instruments and sites.
Focus resources on your added value.
Concentrate on data science, processes and the applications that set you apart.
Reduce software infrastructure costs.
Limit cross-cutting development and proprietary platform maintenance with a common software foundation maintained and enhanced by Pollen.
On-premise deployment, in your environment.
Whether deployed in a fab or integrated into equipment, Smart3 runs on your infrastructure. Data confidentiality remains central to every deployment.
Savings extend beyond time: shared platform development, maintenance, testing and software infrastructure help reduce your total cost over time.
About PollenData science.
Data science.
Industrial reality.
Founded in Grenoble in 2014 and based in Moirans since 2015, Pollen grew out of fifteen years of research by its founder, Johann Foucher, at CNRS and CEA in plasma physics for semiconductor etching and advanced metrology.
One conviction guides Smart3: understanding processes requires connecting data and making analysis customizable. Our open software platform combines experimental data, domain expertise and simulation within a holistic view of processes.
From a scientific vision to collective expertise.
From 2015 to 2024, ten years of development built the Smart3 core and a multidisciplinary team spanning software, AI, signal processing, semiconductors, materials characterization, physics and chemistry.
Today, this collective expertise translates into practical applications for industrial customers and equipment suppliers in the United States, Europe and Asia, from R&D to production.
Our management team
Johann Foucher (PhD)
Founder & CEO
LinkedIn ↗
Ali Hallal (PhD)
Team Leader Application & Support
LinkedIn ↗
Julien Baderot (PhD)
Team Leader Data Science
LinkedIn ↗
Hervé Ozdoba
Team Leader Software Development
LinkedIn ↗See Smart3 in action.
YouTube ↗No-code · AFM & EBSD plugins and workflows with LLM + RAG
Watch on YouTube ↗
YouTube ↗Advanced waveguides · 3D reconstruction & process control
Watch on YouTube ↗
YouTube ↗From LLMs to Agentic AI · Towards lights-out fabs
Watch on YouTube ↗
YouTube ↗Plasma simulation · Connecting simulation to reality
Watch on YouTube ↗
YouTube ↗MicroLED · AI analysis and process control
Watch on YouTube ↗Pollen highlights
Grenoble · AlpexpoCareers
forum14–15 · 10 · 2026
forum14–15 · 10 · 2026
Pollen at the Grenoble careers forum
On 14–15 October 2026 at Alpexpo, Pollen showcased its careers, Smart3 and recruitment and internship opportunities at OPCO Atlas booth N3.
View on LinkedIn ↗
From packaging data to operational decisions
Pollen at the 3rd National Packaging Days, held on 13–14 October 2026 at CEA-Leti in Grenoble. Richard Fontanges presented how Smart3 connects images, measurements, process data and defects to guide engineering decisions.
View on LinkedIn ↗
Pollen · Booth B00Pollen at MicroLED Connect 2026 — booth B00
Pollen exhibited at MicroLED Connect and AR/VR Connect on 16–17 September 2026 in Eindhoven, the Netherlands, at booth B00.
View on LinkedIn ↗Event website · image source ↗
CICERIF working group: advancing sustainable electronics
A look back at the CICERIF working session at Valeo, focused in particular on FALAISE — Failure AI for Sustainable Electronics. A collective initiative of the French Electronics Federation around innovation and sustainable electronics.
View the FdEF post ↗
One team, a shared ambition and a milestone together
A day in the mountains to celebrate a new milestone for Pollen and thank the team, customers and partners. Hiking, lunch at Altitude 2000 and time together celebrating what makes the journey possible: trust, passion and teamwork.
View on LinkedIn ↗
PESM 2026: connecting plasma simulation and real-world processing with STMicroelectronics
Presented by Julien Baderot in Grenoble, collaborative Pollen–STMicroelectronics research on FDSOI gates combines AI-based automated TEM metrology, digital twins and multi-objective plasma etching optimization to accelerate the identification of process windows.
View on LinkedIn ↗The science behind Smart3.
Explore six selected publications and an outside perspective on Pollen. Metrology, defect analysis, open platforms and process optimization: research rooted in industrial applications.
Featured · SemiWiki
A decade of development, an AI platform deployed in industry
Lauro Rizzatti explores the development of Smart3 and its role in connecting data, equipment and engineering expertise across semiconductors and advanced materials.
Read on SemiWiki ↗Our selection
Infinitesima
Large-scale AFM image stitching
Combine high-speed AFM scans into large-area maps and analyze them in Smart3, in collaboration with Infinitesima.
View publication ↗Full reference
Advancements and challenges in large-scale atomic force microscopy image stitching for semiconductor devices
- Authors
- Julien Baderot, Ali Hallal (Pollen); Matthew Tedaldi, Richard Thomas, Andrew D. L. Humphris (Infinitesima); Johann Foucher (Pollen)
- Publication / conference
- SPIE Advanced Lithography + Patterning 2026, Metrology, Inspection, and Process Control XL (Proc. SPIE)
Metrospection: metrology meets inspection
Combine measurements and defect detection, from patterns to wafers, in an open platform configured by the engineer.
View publication ↗Full reference
Metrospection: holistic AI-driven process control software platform dedicated to metrology and defectivity for patterning process at die and wafer levels
- Authors
- Julien Baderot, Ali Hallal, Martin Jacob, Sergio Martinez, Johann Foucher (Pollen)
- Publication / conference
- 39th European Mask and Lithography Conference (EMLC 2024), Proc. SPIE 13273, 132731J
Open AI platform and digital twin
Connect data cleaning, modeling and process optimization while retaining the ability to integrate proprietary algorithms.
View publication ↗Full reference
Open collaborative AI platform to support process optimization for semiconductor industry with digital twin
- Authors
- Julien Baderot, Marion Grould, Ali Hallal, Johann Foucher (Pollen)
- Publication / conference
- SPIE Advanced Lithography + Patterning 2026, DTCO and Computational Patterning V (Proc. SPIE)
STMicroelectronics
Augmented Engineer: plasma simulation and real data
A no-code AI layer to help engineers connect simulation with process data and explore optimization trade-offs, with STMicroelectronics.
View reference ↗Full reference
Augmented Engineer: A No-Code AI Application Layer for Bridging the Gap Between Plasma Simulation and Real-World Processing
- Authors
- Julien Baderot, Marion Grould (Pollen); Caroline Lacomme, Linda Parmigiani (STMicroelectronics); Ali Hallal, Hervé Ozdoba, Johann Foucher (Pollen)
- Publication / conference
- PESM 2026 (Plasma Etch and Strip in Microelectronics), June 2026, oral presentation
Meta
Adaptable metrology for AR/VR structures
Measure slanted gratings in TEM and structures in CD-SEM with workflows that evolve alongside Meta’s process development.
View publication ↗Full reference
Scalable and adaptable structural metrology for AR/VR device images based on deep learning
- Authors
- Subhei Shaar, Maclean Harned, Bo Zhao, Kundan Chaudhary, Raja Muthinti (Meta); Ali Hallal, Martin Jacob, Julien Baderot, Sergio Martinez, Johann Foucher (Pollen)
- Publication / conference
- SPIE Photomask Technology 2024, Proc. SPIE 13216, 132161H
Deep-learning metrology on SEM/TEM images
The foundations of Pollen’s AI metrology: annotation, detection, segmentation and measurements from user-provided examples.
View publication ↗Full reference
Application of deep-learning based techniques for automatic metrology on scanning and transmission electron microscopy images
- Authors
- J. Baderot, M. Grould, D. Misra, N. Clément, A. Hallal, S. Martinez, J. Foucher (Pollen)
- Publication / conference
- Journal of Vacuum Science & Technology B 40, 054003 (2022)
Build with us
Which process would you like to improve?
Tell us about your data, application and objectives. Together, we will define a practical first step.