Inkjet-Printed Droplet Classification & Printing Parameter Optimization Using AI

Overview

This research develops machine learning methods to classify droplet behavior and automate parameter optimization for inkjet printing fabrication. By combining high-speed video analysis with AI, we replace traditional trial-and-error tuning with a systematic approach to achieving optimal printing parameters across diverse inks and substrates.

Flexible Electronics

Flexible electronics represent a critical frontier in modern electronics manufacturing. These integrated circuits and sensors can bend or wrap around objects such as skin or clothing, opening unprecedented possibilities for wearable and bio-integrated devices.

Common flexible electronics manufacturing processes include:

• Laser patterning
• Inkjet printing
• Carbon-based printed inks
• Screen printing
• Aerosol-jet printing
• Gravure printing

Inkjet Printing (IJP)

Inkjet printing is a key technology that benefits flexible electronics by enabling the deposition of functional materials as droplets to create different patterns for electronics devices, sensors, and e-textiles.

Key Advantages:

• Rapid Prototyping: Allows rapid prototyping of any electronics devices
• Material Versatility: Works with diverse inks (metals, polymers, biologics)
• Surface Flexibility: Enables printing on many solid and flexible surfaces, including plastics, glass, photo paper, and textiles

Critical Performance Metrics:

• Line resolution and feature quality
• Device performance and reliability
• Print yield and efficiency

The Challenge:

Operators must manually adjust printing parameters such as voltage, jetting frequency, and meniscus pressure for every new ink—a process that is slow, wasteful, and not scalable. This research aims to solve this problem using artificial intelligence.

Methodology

Drop Watcher Setup and Data Collection

We employed a Fujifilm Dimatix DMP-2850 Materials Printer equipped with a Drop Watcher imaging system featuring a UV lamp and fiducial camera, enabling real-time droplet observation and classification.

Systematic Parameter Variation:

• Voltage: 18–40V (controls droplet ejection pressure)
• Jetting Frequency: 5–30 Hz (controls droplet spacing)
• Meniscus Pressure: 0.5-10 in H₂O (affects ink refill)
• Temperature: 30°C, 40°C, 50°C (impacts ink viscosity)

Dataset Characteristics:

• Total samples: 4000+ droplet experiments
• Training samples: 2000+
• Multiple ink types tested for generalization

Results & Outcomes

For the Ethaline Water 75-25 formulation at optimal meniscus pressure (0.5 in H₂O), our analysis revealed a precisely defined operating window where droplet quality is maximized and printing reliability is assured.

Conclusion & Impact

Key Contributions:

• Replace trial-and-error: Replace traditional trial-and-error tuning with AI-driven optimal printer setup
• Speed and Efficiency: Speed up inkjet-printing fabrication while providing a general framework to test new ink materials for devices
• Scalability: Enable rapid, data-driven optimization across diverse inks and substrates

Future Directions:

• Develop a future closed-loop control system for real-time inkjet printing fabrication
• Create a real-time, video-based system to monitor and label droplet behavior during printing
• Train AI models to predict optimal printing parameters and droplet outcomes automatically
• Extend outcome maps to additional ink materials and substrates

This work demonstrates that machine learning and systematic characterization can transform inkjet printing from a manual, iterative process into a data-driven, predictive discipline—enabling faster innovation in flexible electronics fabrication.