Grantee Research Project Results

Practical Management of PFAS contaminated agricultural soil using an innovative platform developed through integrated experimental research and machine learning approaches

EPA Grant Number: R840958
Title: Practical Management of PFAS contaminated agricultural soil using an innovative platform developed through integrated experimental research and machine learning approaches
Investigators: Liang, Yanna , Zhang, Huichun Judy , Bowers, Clifford R
Institution: The State University of New York at Albany , University of Florida
EPA Project Officer: Hahn, Intaek
Project Period: September 1, 2024 through August 31, 2028
Project Amount: $1,600,000
RFA: Research for Understanding PFAS Uptake and Bioaccumulation in Plant and Animals in Agricultural, Rural, and Tribal Communities Request for Applications (RFA) (2024) RFA Text |  Recipients Lists
Research Category: PFAS Detection , Urban Air Toxics , Watersheds , Endocrine Disruptors , Heavy Metal Contamination of Soil/Water , Environmental Justice , PFAS Treatment

Objective:

This project aims to test the overarching hypothesis that proper engineering control will allow continued and beneficial use of biosolids on agricultural soil and that this practice can sustain marketable and PFAS-free products from such soil. To achieve this goal, we seek to accomplish four specific objectives:

a)  Modeling the relationship between PFAS solid-water partition coefficient (Kd) and soil properties with or without powdered activated carbon (PAC).

b)  Understanding uptake of individual and total precursor PFAS by representative plants grown in biosolids-amended soil. The resulting data will be processed by machine learning (ML) to build models that can predict PFAS behavior in a given growth environment.

c)  Elucidating the binding mechanisms for and stability of the bound residues formed between PFAS, soil, biosolids with or without PAC. New sorbents whose design is guided by the mechanisms will be tested thoroughly.

d)  Performing field trials to verify the accuracy of the developed models. Results from the field trials will confirm whether the established database and predictive tools are ready to be handed to potential users.

Approach:

Task 1 focuses on developing ML models predicting Kd for 40 target PFAS and total precursors based on an extensive test of 12 different soils, three kinds of biosolids with PAC at three different doses, plus literature data. Task 2 centers on understanding uptake of 40 native PFAS, two ¹³C labeled PFAS, and total precursors by three plants in 12 different soils. The obtained data will be used to develop predictive models for forecasting PFAS uptake by different plants in various soil environments. Task 3 aims to elucidate the mechanisms underlying formation of bound residues between PFAS, soil, biosolids and PAC. The elucidated mechanisms will then lead to the design of new and better sorbents than PAC. Task 4 seeks to validate the ML models developed from Tasks 1 and 2 by conducting field trials.

Expected Results:

A thorough understanding of the fate of PFAS in the complex soil-biosolids- plant systems with or without an amendment will be attained. This understanding plus the predictive models will enable the research team to generate engineering control strategies and guidelines to assist the PFAS-affected communities in their decision making. Consequently, sustainable agriculture will be promoted while the environment and human health are protected.

Supplemental Keywords:

agriculture, bioavailability, community-based, decision making, environmental chemistry, machine learning modeling, pollution prevention.

Progress and Final Reports:

2025