Planet-friendly plastics

June 2023
Filed under: Los Alamos
A Los Alamos team applies machine learning to find environmentally benign plastics.

Codes lines from BioManIAC. (Image: Los Alamos National Laboratory.)

The United States’ efforts to cut greenhouse gas emissions by switching from fossil fuels – natural gas, oil, coal – to renewable energy produce another significant impact on everyday life: reducing inexpensive plastics production.

Most of the plastics we use in bottles, grocery bags, food containers and more are made from petroleum. Although the raw materials to make these and other products comprise only a quarter-barrel of oil, these goods are almost as valuable as the 75% of the contents used to make fuels.

“You can’t think of fuel in isolation of these other products,” says Babetta Marrone, a senior scientist and fellow at Los Alamos National Laboratory (LANL). Plastics are vital to oil and gas companies’ profits.

Plastics outlive many manmade materials and have irked environmentalists because they pollute water and soil and ensnare animals. A 2017 study found that about 79% of the 8.3 billion metric tons of plastics produced to date has accumulated in landfills or the natural environment. The market for conventional plastics will continue to rise, with global waste likely to triple by 2060, the Organisation for Economic Co-operation and Development reports. In other words, plastics are unlikely to disappear entirely.

Researchers believe bioplastics, made from renewable sources, can be a fully biodegradable alternative. LANL’s Biofuels and Bioproducts program is one of many efforts to identify, research and develop these products. Marrone’s project, BioManIAC (Bio-Manufacturing with Intelligent Adaptive Control) uses machine learning and artificial intelligence to accelerate deployment of materials to replace conventional plastics. The name honors LANL’s MANIAC, the first fully operational electronic computer when it came online in 1952.

“We developed machine-learning models to try and understand the relationships between the chemical structure of the (plastic) polymer and the material properties it could code,” Marrone says. The team’s computational tool won an R&D 100 award as a top new technology for 2022, recognizing its potential to revolutionize biofuel and bioplastic design and optimization.

The LANL researchers have focused on one class of chemicals, polyhydroxyalkanoates (PHAs), that are naturally produced by several species of photosynthetic bacteria called cyanobacteria, or blue-green algae. Using computational tools and a large-scale approach, the team studied PHAs’ chemical structures and how they relate to function.

“There are a lot of  PHA variations, and you can chemically tune them to have different lengths or different side chains,” Marrone says. In other words, the compounds can be readily modified, depending on how researchers want to use the chemicals in plastics . PHAs are an ideal bioplastics candidate, Marrone says, because they’re already biocompatible, or harmless to human tissue. Although the Food and Drug Administration hasn’t yet approved their use in pharmaceuticals, researchers also are exploring PHAs as a nanoparticle material for drug delivery.

Using BioManIAC, the team developed machine-learning models to study the chemicals’ structure and they’d degrade in the environment.

In a December 2022 Nature Communications Materials paper, the LANL BioMAnIAC team used a deep neural network machine-learning algorithm to identify 14 PHA-based bioplastics that it believes can replace seven key petroleum-derived plastics. The neural network was trained on a search space of 540 known PHAs plus 13 conventional polymers and their mechanical, thermal and gas permeability properties – a total of 22,731 data points. The algorithm sifted a possible 1.4 million bioplastics that would perform similarly to conventional, petroleum-based plastics to identify the candidate materials.

The team also is pursuing machine learning tools to study how these PHA candidates degrade and to identify the most promising cyanobacteria species – and their genes – to help develop bioplastics at large scale.

The research excites Marrone and her colleagues. “Conventional plastics can last decades or more in the environment. We really want to minimize” their environmental damage. “Our work creates a vital step in that direction.

(Visited 137 times, 1 visits today)

About the Author

Wudan Yan is a Seattle-based freelance science writer.

Comments are closed.