The purpose of this project was to field-validate the use of a portable electrochemical sensor technology for rapid assessment of per-and polyfluoroalkyl substances (PFAS). Comparatively low regulatory screening levels for PFAS have created a demand for fast, reliable, and cost-effective measurement. The intended application of this sensor is to use downstream of PFAS treatment systems to measure PFAS concentrations with high selectivity and sensitivity. The overarching goal of this project was to field-validate the use of a portable electrochemical sensor technology for rapid assessment of PFAS. Originally, objectives for this project included evaluating the sensor performance with six PFAS in synthetic samples, studying sample interference, and field-testing the sensor downstream of a PFAS treatment system.

The electrochemical sensor was developed collaboratively at the Pacific Northwest National Laboratory (PNNL) and the New Jersey Institute of Technology (NJIT). The flow through electrochemical sensor uses a nanoporous and capacitive electrode technology based on a nonplanar interdigitated microelectrode array (NP-IDμE). The NP-IDμE device consists of three layers: a top and bottom microelectrode (IDμE) and a middle layer of adsorptive probes (consisting of metalorganic framework, zeolites, covalent organic frameworks, or hierarchical porous carbons) to capture PFAS. The benefits of using these materials for PFAS capture are the separation of perfluoroalkyl chains from other organics and the tunable affinities toward PFAS functional groups to differentiate among them. The probe is designed as a mediator to transduce the chemical interaction signal to an electrical signal, which ultimately enables PFAS quantification.

The detection of perfluorooctane sulfonic acid (PFOS) in spiked tap water solutions was demonstrated with separate sensors, one based on Cr-MIL-101 and other on Fe-MIL-101. These two metal–organic frameworks (MOFs) showed high PFOS uptake during sorption studies conducted at PNNL with Fe-MIL-101 outperforming Cr-MIL-101. The sensor housing Cr-MIL-101 was evaluated at six different PFOS concentrations (10, 25, 50, 75, 100 and 150 ng/L). The initial results were promising with the sensor detecting the lowest standard of 10 ng/L. Even though a linear relation between the PFOS concentration and normalized sensitivity was obtained, the large standard deviations among detections at each concentration overlapped with most other standards. Furthermore, statistical analyses indicated low confidence in this correlation. Tests using sensors that housed Fe-MIL-101 showed better sensitivity compared to Cr-MIL-101. Using Fe-MIL-101, detections at 1, 10 and 50 ng/L PFOS solutions were conducted. A linear relationship between concentration and normalized sensitivity was obtained, but similar to Cr-MIL-101, the standard deviations at each concentration overlapped with others.

Detection of perfluorooctanoic acid (PFOA) in spiked tap water solutions was demonstrated using separate sensors that housed Universitetet i Oslo (UIO)-66 and UIO-66-NH2 at PFOA concentrations of 10, 50, 100, 500 and 1000 ng/L. These two MOFs were selected based on their high PFOA uptake observed during sorption studies conducted at PNNL. A linear relationship between PFOA concentrations and normalized sensitivity was not obtained, possibly due to the inability to lower the particle size of UIO-66 and UIO-66-NH2, resulting in packed channels of relatively high porosity in the sensor. Loosely packed MOFs in the sensor could cause electrical charges to transfer to the external circuit through other pathways resulting in erroneous sensitivity readings. Additionally, the sensor showed high sensitivity, even in blank samples, which overlapped with sensitivities obtained for all other samples, including the highest PFOA standard of 1000 ng/L. Tests using Cr-MIL-101 based sensors for PFOA showed poor selectivity between PFOA and PFOS. The selectivity and performance of the sensor in PFAS mixtures could be improved by using MOFs with high affinity for the target PFASs and adopting multiple sensors in series and parallel configuration.

The high variation in sensitivity readings and false positives in blanks warranted further investigation. To address this, Arcadis coordinated a double-blind quality control test to assess the precision and accuracy of the PFAS sensor. Nine PFOS standards including a deionized (DI) water blank were sent to NJIT for sensor testing and to Pace Analytical for validation. Nine standards included one DI water blank, five at a PFOS concentration of 47 ng/L, two at 15 ng/L, and the remaining one standard at 75 ng/L. The sensor results were compared to the Pace Analytical results to arrive at the accuracy and precision values of the sensor. The PFAS sensor results for the 47 ng/L PFOS standard ranged from 102 to 580.7 ng/L which was 429 to 2420% of the concentration reported by Pace Analytical (24 ng/L). Precision of the sensor was at a %RSD of 83% among the five sensor readings. For the 75 ng/L PFOS standard, the PFAS sensor showed a result of 117.3 ng/L, which was 267% of the Pace Analytical result (44 ng/L) and had a standard deviation of 52 ng/L. Finally, the sensor showed a PFOS concentration of 140.7 ng/L for the DI water blank, while Pace Analytical did not detect any PFOS in it. Based on the results, the accuracy and precision of the sensor deviated substantially from the accuracy precision performance goals for the sensor.
 

In parallel, a study was conducted to evaluate the MOF variability between different batches and a method to get a uniform particle distribution across different batches was developed. However, previous tasks that evaluated the PFAS sensor used MOFs from different batches. Interestingly, the electrochemical behavior between the two batches were found to be different which made translating the calibration curve from one batch of MOF to another challenging. Another challenge was achieving uniform chip packing and variable accessibility to MOF sorption sites, because the chips are small, single use, and packed manually. Because of this, MOF sorption behavior varies significantly from chip to chip.

The precision goals could not be achieved under the manual fabrication method. To avoid chip-to-chip to variation in the future, a larger chip design with automated packing needs to be adopted. However, this will likely require complete chip redesign and different chip production equipment, which the team currently cannot achieve. This will require significant additional bench testing and troubleshooting, which are beyond the scope of this project. Hence, a no-go decision was recommended for the remaining project tasks. (Project Completion - 2023)