Active learning for advanced biodegradable film design

Developing biodegradable alternatives to traditional plastics is crucial for sustainable packaging. However, carboxymethyl cellulose (CMC), a promising eco-friendly material, suffers from brittleness and high hydrophilicity. This study integrates materials science, machine learning, and optimization...

Full description

Saved in:
Bibliographic Details
Main Authors: Yi Gao, Chengcheng Liu, Yang Zhao, Peng Zhao, Feihui Chen, Qinghai Li, Yanguo Zhang, Bin Yang, Hui Zhou
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Nexus
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2950160125000178
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850153957664489472
author Yi Gao
Chengcheng Liu
Yang Zhao
Peng Zhao
Feihui Chen
Qinghai Li
Yanguo Zhang
Bin Yang
Hui Zhou
author_facet Yi Gao
Chengcheng Liu
Yang Zhao
Peng Zhao
Feihui Chen
Qinghai Li
Yanguo Zhang
Bin Yang
Hui Zhou
author_sort Yi Gao
collection DOAJ
description Developing biodegradable alternatives to traditional plastics is crucial for sustainable packaging. However, carboxymethyl cellulose (CMC), a promising eco-friendly material, suffers from brittleness and high hydrophilicity. This study integrates materials science, machine learning, and optimization techniques to enhance CMC films by improving mechanical performance, hydrophobicity, and optical properties using only low-cost, readily available solvents. Through six iterations with 186 data points for three-target optimization and seven iterations with 166 data points for two-target modeling, a data-driven platform was established to accurately predict CMC film properties based on composition and autonomously identify optimal formulations for specific performance targets. The best-performing CMC film achieved a water contact angle of 113.7°, tensile strength of 37.7 MPa, and elongation at break of 31%, outperforming the films enhanced with costly or nano additives. Harnessing the predictive power of the model facilitated the development of high-performance CMC films with tunable contact angles and optical properties while maintaining mechanical integrity. Structural analysis further confirmed a dense surface with excellent thermal stability and durability. Additionally, the integrated software platform, ALA Designer, enhances user-machine interaction and data management, extending its applicability beyond CMC films. This approach enables the multi-objective, multifunctional design of CMC films, optimizing mechanical properties essential for packaging while allowing tunable hydrophobicity to control release properties for pharmaceutical and cosmetic applications. By combining machine learning and materials science, this work accelerates the development of sustainable materials with properties similar to synthetic alternatives but with lower environmental impact and production costs, making a significant real-world contribution to eco-friendly materials in industry. Broader context: The global reliance on plastics has led to severe environmental challenges, with annual production exceeding 400 million tons. Packaging alone accounts for nearly 40% of this total, contributing significantly to pollution and long-term ecological risks. Reducing plastic dependency and developing biodegradable alternatives from renewable resources is critical for sustainability. Biopolymers such as cellulose have emerged as promising candidates due to their biodegradability and renewability. Among them, carboxymethyl cellulose (CMC) offers excellent film-forming properties but suffers from brittleness and high hydrophilicity, limiting its widespread adoption.Traditional methods to improve biopolymer films often depend on costly additives and labor-intensive optimization processes, which can enhance certain properties at the expense of others. This study integrates materials science, machine learning (ML), and optimization techniques to streamline CMC film design, achieving superior performance using only low-cost, widely available solvents. By identifying predictive relationships in experimental data, ML enables efficient and cost-effective formulation development without expensive additives. This approach aligns with green chemistry principles and offers a scalable solution for industries such as packaging, pharmaceuticals, and cosmetics. Ultimately, it provides an economically viable and environmentally friendly pathway toward reducing plastic waste while enhancing the functionality and commercial viability of biodegradable films.
format Article
id doaj-art-de8ceef0919e415da8fffd43cd4f0a03
institution OA Journals
issn 2950-1601
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Nexus
spelling doaj-art-de8ceef0919e415da8fffd43cd4f0a032025-08-20T02:25:35ZengElsevierNexus2950-16012025-06-012210007010.1016/j.ynexs.2025.100070Active learning for advanced biodegradable film designYi Gao0Chengcheng Liu1Yang Zhao2Peng Zhao3Feihui Chen4Qinghai Li5Yanguo Zhang6Bin Yang7Hui Zhou8Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, ChinaKey Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, ChinaPrismlab China Limited, Shanghai 201615, ChinaKey Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, ChinaKey Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, ChinaKey Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, ChinaKey Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, ChinaKey Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, China; Corresponding authorKey Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, China; Corresponding authorDeveloping biodegradable alternatives to traditional plastics is crucial for sustainable packaging. However, carboxymethyl cellulose (CMC), a promising eco-friendly material, suffers from brittleness and high hydrophilicity. This study integrates materials science, machine learning, and optimization techniques to enhance CMC films by improving mechanical performance, hydrophobicity, and optical properties using only low-cost, readily available solvents. Through six iterations with 186 data points for three-target optimization and seven iterations with 166 data points for two-target modeling, a data-driven platform was established to accurately predict CMC film properties based on composition and autonomously identify optimal formulations for specific performance targets. The best-performing CMC film achieved a water contact angle of 113.7°, tensile strength of 37.7 MPa, and elongation at break of 31%, outperforming the films enhanced with costly or nano additives. Harnessing the predictive power of the model facilitated the development of high-performance CMC films with tunable contact angles and optical properties while maintaining mechanical integrity. Structural analysis further confirmed a dense surface with excellent thermal stability and durability. Additionally, the integrated software platform, ALA Designer, enhances user-machine interaction and data management, extending its applicability beyond CMC films. This approach enables the multi-objective, multifunctional design of CMC films, optimizing mechanical properties essential for packaging while allowing tunable hydrophobicity to control release properties for pharmaceutical and cosmetic applications. By combining machine learning and materials science, this work accelerates the development of sustainable materials with properties similar to synthetic alternatives but with lower environmental impact and production costs, making a significant real-world contribution to eco-friendly materials in industry. Broader context: The global reliance on plastics has led to severe environmental challenges, with annual production exceeding 400 million tons. Packaging alone accounts for nearly 40% of this total, contributing significantly to pollution and long-term ecological risks. Reducing plastic dependency and developing biodegradable alternatives from renewable resources is critical for sustainability. Biopolymers such as cellulose have emerged as promising candidates due to their biodegradability and renewability. Among them, carboxymethyl cellulose (CMC) offers excellent film-forming properties but suffers from brittleness and high hydrophilicity, limiting its widespread adoption.Traditional methods to improve biopolymer films often depend on costly additives and labor-intensive optimization processes, which can enhance certain properties at the expense of others. This study integrates materials science, machine learning (ML), and optimization techniques to streamline CMC film design, achieving superior performance using only low-cost, widely available solvents. By identifying predictive relationships in experimental data, ML enables efficient and cost-effective formulation development without expensive additives. This approach aligns with green chemistry principles and offers a scalable solution for industries such as packaging, pharmaceuticals, and cosmetics. Ultimately, it provides an economically viable and environmentally friendly pathway toward reducing plastic waste while enhancing the functionality and commercial viability of biodegradable films.http://www.sciencedirect.com/science/article/pii/S2950160125000178carboxymethyl cellulosefilmmaterials designactive learningmulti-objective optimization
spellingShingle Yi Gao
Chengcheng Liu
Yang Zhao
Peng Zhao
Feihui Chen
Qinghai Li
Yanguo Zhang
Bin Yang
Hui Zhou
Active learning for advanced biodegradable film design
Nexus
carboxymethyl cellulose
film
materials design
active learning
multi-objective optimization
title Active learning for advanced biodegradable film design
title_full Active learning for advanced biodegradable film design
title_fullStr Active learning for advanced biodegradable film design
title_full_unstemmed Active learning for advanced biodegradable film design
title_short Active learning for advanced biodegradable film design
title_sort active learning for advanced biodegradable film design
topic carboxymethyl cellulose
film
materials design
active learning
multi-objective optimization
url http://www.sciencedirect.com/science/article/pii/S2950160125000178
work_keys_str_mv AT yigao activelearningforadvancedbiodegradablefilmdesign
AT chengchengliu activelearningforadvancedbiodegradablefilmdesign
AT yangzhao activelearningforadvancedbiodegradablefilmdesign
AT pengzhao activelearningforadvancedbiodegradablefilmdesign
AT feihuichen activelearningforadvancedbiodegradablefilmdesign
AT qinghaili activelearningforadvancedbiodegradablefilmdesign
AT yanguozhang activelearningforadvancedbiodegradablefilmdesign
AT binyang activelearningforadvancedbiodegradablefilmdesign
AT huizhou activelearningforadvancedbiodegradablefilmdesign