Grantee Research Project Results

Final Report: Machine-learning-assisted development of PCB-free alternatives to commodity pigments

EPA Contract Number: 68HERC23C0008
Title: Machine-learning-assisted development of PCB-free alternatives to commodity pigments
Investigators: Saikin, Semion
Small Business: Kebotix, Inc.
EPA Contact: Richards, April
Phase: II
Project Period: October 21, 2022 through October 20, 2024
Project Amount: $399,983
RFA: Small Business Innovation Research (SBIR) Phase II (2023) Recipients Lists
Research Category: SBIR - Air , SBIR - Homeland Security , SBIR - Sustainability , SBIR - Air Quality

Description:

Key Results:
• A library of non-chlorinated amine substrates suitable for producing novel azo pigments has been discovered in silico using Kebotix’s proprietary AI-pipeline, combining both high-throughput screening of molecular libraries using ML models and generative-AI models for predicting synthetic routes. The pigments were synthesized with an automated system and tested for performance in thermoplastics. The applied discovery approach is generic and can be used for a wide spectrum of colorants.
• Novel synthetic processes have been developed to produce a wide range of azo pigments from affordable reagents without the potential to generate any PCBs. These processes have been optimized for broad suitability with the production of existing structures and managed to attain results competitive with established commercial methods.
• A pipeline for AI-driven discovery of blue and green pigments to replace phthalocyanine-based colorants has been developed and a library of lead candidate molecules have been generated. Four iterations—two for each discovery pipeline—were completed, which helped us to build and calibrate the ML models. The discovered candidates will be used in the further lead optimization process.

Conclusions:

Commercialization: During the project, Kebotix built a network of commercial partners in pigment manufacturing and formulation, engaging with many under NDA. Pigment manufacturers highlighted challenges due to thin profit margins, which make them hesitant to adopt new technologies, despite complying with PCB containment standards. Requests for greener pigments were noted, but economic factors dominate decision-making. Early adopters among pigment users are seeking environmentally friendly alternatives, driven by consumer demand rather than specific concerns about inadvertent PCBs due to lack of information. Additionally, Kebotix partnered with contract manufacturing organizations (CMOs) to explore scale-up options, although CMOs lack direct sales channels and require upfront investment. The project also consulted experts on PCBs, who confirmed the environmental health impact of inadvertent PCB in products but noted low public awareness. Discussions with Washington State's Department of Ecology revealed PCB contamination concerns in the Spokane River, though regulatory influence over pigment manufacturers remains limited to local efforts.

SBIR Phase I:

Machine-Learning-Assisted Development of Alternatives to Diarylide Pigments  | Final Report