Accelerating 2D Materials Exploration with Autonomous AFM
Sungkyunkwan University combines AI with Park FX40 to automate 2D flake identification, measurement, and property analysis.
- Customer: Sungkyunkwan University
- Research Focus: Autonomous 2D materials exploration combining AI with atomic force microscopy; automated flake identification, measurement, and property analysis
- Solution: Park FX40
- Key Value: Accelerates 2D flake characterization through AI-driven automation; enables large-scale dataset collection with minimal manual intervention
The Challenge
Characterizing exfoliated 2D materials with conventional scanning probe microscopy can be slow and labor-intensive. Researchers must locate randomly distributed flakes, identify regions of interest, perform repeated high-resolution measurements, and interpret complex imaging data.
As the number of samples increases, these manual processes limit measurement throughput and make it difficult to collect large datasets for systematic and objective analysis of structure–property relationships.
“Conventional SPM imaging requires manual identification of regions of interest and heavily depends on human intuition for image interpretation, which severely limits the ability to collect large datasets and conduct objective analysis.”
— Prof. Yunseok Kim, Sungkyunkwan University
The Solution
The research team developed an autonomous measurement and analysis workflow combining deep-learning image segmentation with Park FX40. Through SmartRemote, external Python scripts communicate with the AFM to connect AI-based feature recognition with automated measurement and analysis.
- AI-Based Flake Identification: Deep learning identifies and segments individual 2D flakes from AFM error signal images, including irregular and complex flake geometries.
- Autonomous AFM Measurement: SmartRemote connects Python scripts with Park FX40, enabling autonomous probe positioning and targeted high-resolution measurements.
- Multimodal Property Characterization: Multimodal KPFM simultaneously captures topography and surface potential, enabling correlated analysis of flake thickness and work function.
- Automated Data Analysis: Automatically flattens, segments, and analyzes AFM data to extract thickness and electrical properties with minimal user intervention.
The Results
The research team combined deep learning with Park FX40 and SmartRemote to create an autonomous workflow for 2D flake identification, measurement, and analysis.
“With Park FX40 and AI-driven automation, we can move beyond conventional manual AFM workflows and explore nanoscale material properties more efficiently and systematically.”
— Prof. Yunseok Kim
Key Outcomes
- Autonomous identification and characterization of 106 MoWTe₂ flakes
- Over 90% mAP in AI-Based Flake Segmentation
- Correlated analysis of flake thickness and work function
- Successful extension of the AI-driven workflow to MoS₂
Related Publication
This capability is further demonstrated through peer-reviewed research from Sungkyunkwan University and KAIST, presenting an AI-driven autonomous SPM workflow that uses AFM error signals to identify and segment 2D material flakes for automated characterization. Published in Small Structures. (DOI: 10.1002/sstr.202500379)
About Park FX40
Park FX40 is a fully automated AFM platform designed for advanced nanoscale research and automated measurement workflows. Capabilities include automated probe exchange, auto laser alignment and tip detection, image-guided sample navigation, and external control through SmartRemote — enabling reliable, reproducible, and efficient nanoscale characterization.
Learn more: www.parksystems.com/fx40
This case study is based on an interview with Prof. Yunseok Kim at Sungkyunkwan University. By integrating Park FX40 with AI-based image recognition and SmartRemote, the research team developed an autonomous workflow for 2D flake identification, targeted AFM measurement, and multimodal property analysis, enabling more efficient and systematic exploration of 2D materials.