Field Testing of an Affordable Zero-Liquid-Discharge Arsenic-Removal Technology for a Small-Community Drinking Water System in Rural California

Abstract

Arsenic contamination in groundwater threatens public health, particularly in small, low-income communities lacking affordable treatment solutions. This study investigated the field implementation of novel air cathode assisted iron electrocoagulation (ACAIE) technology for arsenic removal in Allensworth, California, where groundwater arsenic concentrations exceeded 250 µg/L. Over four months, a pilot-scale ACAIE system, operating at 600 L/h, consistently reduced arsenic levels to below the EPA’s maximum contaminant level of 10 µg/L. Laboratory experiments informed the optimization of charge dosage and flow rates, which were validated during field testing of the ACAIE 600 L/h system. The in-situ generation of hydrogen peroxide at the cathode speeded up the reaction kinetics, ensuring high arsenic removal efficiency while allowing high throughput, even with a compact reactor size. An economic analysis demonstrated a treatment cost of USD 0.02/L excluding labor, highlighting the system’s affordability compared to conventional methods. Adding labor costs increased the treatment cost to USD 0.09/L. The regeneration of air cathodes extended their operational life, addressing a key maintenance challenge, thus reducing the costs slightly. Intermittent challenges were encountered with filtration and secondary contaminant removal; these issues highlight opportunities for further operational improvements. Despite these challenges, ACAIE’s low operational complexity, scalability, and cost-effectiveness make it a promising solution for underserved small communities. These findings provide critical insights into deploying sustainable arsenic remediation technologies that are tailored to the needs of rural, low-resource communities.

1. Introduction

Globally, more than 50% of the population relies on groundwater for drinking water [1]. Climate change, a growing population, and surface water pollution have increased this reliance on groundwater. However, naturally occurring geogenic contaminants such as arsenic pose a significant threat to groundwater quality and availability [2,3]. More than 200 million people consume groundwater with arsenic concentrations greater than 10 µg/L, the maximum contaminant level (MCL) for arsenic in drinking water set by the World Health Organization, the US Environmental Protection Agency (EPA), and in other countries [4,5,6]. Chronic arsenic exposure is associated with numerous health complications in humans, including skin lesions, diabetes, cardiovascular and neurological diseases, cognitive impairment, and an increased risk of cancer [4,7].

In the United States, arsenic contamination in drinking water disproportionately affects small, low-income communities and communities of color. These communities often lack the technical, managerial, and financial resources to afford commercially available arsenic removal technologies [8,9,10,11]. Managers of, and consultants to, community water systems impacted by arsenic in their drinking water source refer to a list of the best available technologies, compiled by the EPA. Many approved technologies for arsenic removal require substantial capital investment, ongoing maintenance, and operation by trained, licensed personnel. Moreover, these systems produce significant waste, posing additional disposal challenges [12,13,14,15,16]. Furthermore, the total yearly costs (annualized capital costs and operation and maintenance costs) on a per capita basis for each of these best available technologies rise steeply and can become economically infeasible for small community (<10,000 people) water systems. Therefore, novel technological inventions need to be brought into the market through technology maturation, as argued by Hering et al. (2016) [17].

Current technologies for removing arsenic are unaffordable for small towns. Community water systems that violate the arsenic MCL usually consider two approaches before choosing a treatment method to satisfy compliance requirements with the arsenic MCL rule: (1) consolidating with a nearby unimpacted water system and (2) drilling new wells to access an aquifer with lower levels of arsenic contamination. The consolidation approach is considered when there is a larger water system within 3 miles and can cost several million dollars in capital costs.

Very small community water system companies (serving a population of fewer than 500) often have few options to regain compliance after notice of arsenic violations is issued by the regulatory agencies. Remote locations make the consolidation approach infeasible, and they have too few ratepayers to cover the annualized capital costs of drilling new wells or installing costly treatment systems. For instance, the Park Royal Mutual Water Company in Sonoma County (CA), which serves 27 homes with a total population of 75, received a compliance order from the CA Water Boards in 2021 due to arsenic contamination in their community well [18]. They considered either a consolidation option with Santa Rosa or installation of a new arsenic treatment system consisting of green sand filtration. This filtration system’s initial quoted capital cost was USD 200,000. In addition, the safe disposal of liquid waste would cost USD 5000 per month (Senior officer of Park Royal Mutual Water Company, personal communication with L.S., 10 August 2022). According to our estimates, the amortized cost per household would be USD 256 per month for this water treatment, which is too high per the Federal guideline threshold that says water should cost below 2.5% of the median household income (MHI) and the California guideline threshold (below 1.5% of MHI) using MHI values for severely disadvantaged communities [19,20,21]. This treatment option is untenable because of its high cost and the additional demands of meeting the regulatory paperwork requirements with volunteer hours. Although the per-cubic-meter cost of treatment is USD 8.11, the technology cannot be affordably scaled down to provide only potable water.

While the above technology was too costly from an operating expenditure perspective, the only feasible option available, and therefore comparable, to new technologies for obtaining potable drinking water would be based on driving to and then purchasing water at local stores. Based on realistic assumptions, purchasing potable water from a local convenience store for indoor use would cost around USD 0.66 per L (USD 2.50 per gallon), including transport (Supplementary Materials Section S1). Failures such as this highlight the key requirements for technologies to be viable at this small scale through low O&M costs, minimal operator intervention or oversight, and the reliability of treatment effectiveness in removing the target contaminant.

Recent advancements in modular electrochemical technologies present promising alternatives for small water systems. These systems are typically more affordable and easier to operate and maintain, making them well-suited for small communities [22]. Iron electrocoagulation has been consistently demonstrated as an effective technology for reducing arsenic levels below the EPA-MCL threshold in both synthetic and groundwater matrices [23,24,25,26,27,28,29]. While iron electrocoagulation has been successfully implemented to provide arsenic-safe water to rural communities in India, its use in the U.S. is less practical due to high labor costs, the requirement for aeration to ensure efficient arsenic removal, which demands a large footprint for the desired flow rates, and its reliance on a full-time operator, further increasing the operational expenses [14,30]. A notable development is air cathode assisted iron electrocoagulation (ACAIE) technology, recognized for its potential to address the limitations of iron electrocoagulation in low-income communities in the U.S. In ACAIE systems, a low-carbon steel plate functions as the anode, while a catalyst-coated, electrically conductive, air-permeable carbon paper serves as an air-diffusion cathode. The anode and cathode are flat surfaces positioned parallel to each other. Upon the application of low-voltage DC current, Fe(II) ions are generated at the anode, and hydrogen peroxide is produced at the cathode. The anodic and cathodic reactions are shown in Equations (1) and (2), respectively.

$$\mathrm{A}\mathrm{n}\mathrm{o}\mathrm{d}\mathrm{e}:\mathrm{F}\mathrm{e}\left(0\right)\to {\mathrm{F}\mathrm{e}}^{2+}+{2\mathrm{e}}^{-}$$

(1)

$$\mathrm{C}\mathrm{a}\mathrm{t}\mathrm{h}\mathrm{o}\mathrm{d}\mathrm{e}:{{\mathrm{O}}_{2\left(\mathrm{A}\mathrm{i}\mathrm{r}\right)}+{\mathrm{H}}_2\mathrm{O}+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2\mathrm{O}}_2+2\mathrm{O}{\mathrm{H}}^{-}$$

(2)

In the bulk solution, Fe(II) undergoes further oxidation by H₂O₂ to form insoluble Fe (oxyhydr)oxides, which have a high adsorption affinity for As(V). The reactive intermediates generated (e.g., Fe(IV) or OH radicals) during the oxidation of Fe(II) by H₂O₂ oxidize any arsenic that is present as As(III) to As(V), enabling efficient arsenic removal at a wide range of current densities and flow rates. This approach leads to zero liquid discharge and improves upon the shortcomings of other technologies, such as high operational costs and the high disposal costs of liquid toxic waste [31,32].

The present study was conducted in Allensworth, CA, an unincorporated and resource-poor community in Tulare County with historically high arsenic levels in their groundwater. The pilot plant featured a scaled-up ACAIE system capable of treating 10 L/min (600 L per hour), which was operated daily for four months to treat arsenic-contaminated groundwater. Before this study, ACAIE had never been demonstrated in a field operation for multiple months. Systematic analysis of wet chemical parameters (e.g., As, Fe, Al, pH) at various unit operations during the pilot testing showed that the ACAIE system consistently reduced arsenic concentrations from 250 µg/L to below the EPA-MCL of 10 µg/L, demonstrating its effectiveness. Furthermore, an economic analysis (presented below) suggests its practicality.

Field testing is critical to identify weaknesses that are not apparent in laboratory studies. Addressing scaling-up challenges, evaluating field performance, and conducting a preliminary economic analysis were essential steps in assessing this novel technology’s potential to serve rural communities in the U.S.

2. Materials and Methods

2.1. Laboratory Scale ACAIE Batch Experiments

Bench-scale ACAIE experiments were performed in a custom-built batch reactor with an active electrolyte volume of 0.5 L. A detailed description and picture of this reactor were provided by Bandaru et al. in 2020 [32]. Briefly, this reactor consists of a low-carbon steel plate, acting as the sacrificial anode, and an air diffusion cathode (fabrication details are provided in the Supplementary Materials). The submerged surface area of the anode was 45 cm² (7 × 6.5 cm², 1006–1026 steel grade; McMaster-Carr, Los Angeles, CA, USA), and the air cathode was 64 cm² (8 cm × 8 cm). Allensworth groundwater, without any pH adjustment or pretreatment, was used as the electrolyte. The time between the collection of raw groundwater and its use in the experiments was less than 24 h.

The experiments were conducted at two different charge dosage rates, 30 C/L/min and 300 C/L/min, while keeping the total charge dosage constant at 600 C/L (3.1 mM Fe). Samples of approximately 5 mL were collected at various charge dosages (100 C/L, 200 C/L, 300 C/L, and 600 C/L), then immediately filtered with a 0.22 µm syringe filter and acidified with 0.5% HCl and 1% HNO₃ for ICP-MS analysis. Additionally, filtered samples (2 mL before acidification) were collected at various charge dosages to determine the dissolved arsenic, iron, and H₂O₂ that remained. At the end of the electrolysis process, 10 mL of unfiltered samples were collected and acidified immediately with 0.5% HCl and 1% HNO₃ for ICP-OES analysis to determine the Faradaic efficiency of the total iron produced. All experiments were performed in triplicate. The average and standard deviation of the measurements are reported.

2.2. Pilot Scale ACAIE Treatment System

2.2.1. Source of Groundwater

The ACAIE treatment system (described below) in this pilot study was used to treat groundwater that is naturally contaminated with arsenic (~250 µg/L) and pumped from an agricultural well (720 ft depth, 600 GPM flow, and 50 hp motor) located on a private farm in Allensworth, CA, USA. This groundwater was first pumped and filtered through two industrial-sized sand filters (LAKOS Filtration Solutions) to remove sediments, then stored in 2500-gallon (9464 L) tanks. High concentrations of sodium, chloride, and sulfate ions make this water saline and non-potable (Table S2). However, these ions have a minimal impact on the arsenic removal pathways in an iron-based electrocoagulation process [24,33,34]. Besides these species, the groundwater composition is representative of other arsenic-contaminated wells in California [35]. The average initial turbidity of this groundwater was less than 1 NTU and was used as the raw water feed into the ACAIE treatment system (Table S2).

2.2.2. Unit Processes

The ACAIE treatment system was housed in a 3.0 m × 3.6 m shed (10′ × 12′ TUFF SHED) built on a concrete pad. Grid electricity (110 VAC) was provided to the shed via a 7.5 kVA transformer. The 600 L per hour ACAIE treatment system comprised five dual-cathode ACAIE reactors, followed by traditional particle separation processes (coagulation/flocculation, settling, and filtration). The system layout is shown in Figure 1.

(1)

Electrolysis

Five ACAIE electrochemical reactors were used in parallel. Each reactor consisted of two air diffusion cathodes with an active submerged surface area of 522.6 cm² on either side. A low-carbon steel plate (30.5 cm × 26 cm, 595.2 cm² active area) was used as the anode, allowing both surfaces to participate in anodic dissolution. Raw groundwater entered each reactor from below, flowed upward through the reactor, and exited at the top through ports on either side. The flow of influent groundwater was controlled by manual ball valves and rotameters (19 L/min capacity) and split into five streams, directed evenly into each reactor with a manifold. Each reactor had its own rotameter to ensure a steady flow of 2 L/min. This modular approach facilitated easy scale-up, with each reactor treating up to 120 L/h for a total flow of 600 L/h. Five power supplies (BK Precision, 1902B, Yorba Linda, CA, USA) were operated in constant-current mode at 10 A, supplying a charge dosage of 300 C/L (86.8 mg/L Fe) for the rated water flow of 2 L/min. Voltage fluctuations were observed as the power supplies compensated for changes in groundwater conductivity and electrode fouling. Water exiting the reactors was collected in a 130-L tank (sampling point SP2) and then pumped out with a 0.2 hp submersible pump.

(2)

Coagulation/Flocculation

Alum stock solutions (10,000 mg/L) were prepared by dissolving aluminum sulfate (GEO^® Specialty Chemicals, Little Rock, AR, USA) in distilled water. The stock solution was dosed into the electrolyzed groundwater containing arsenic-laden iron(III)(oxyhydroxides) suspensions using a peristaltic pump set to a feed rate of 10 mL/min, ensuring a coagulant dose of 10 mg/L Al. An in-line static mixer enhanced the mixing process. Flocculation occurred in a 230-L tank (sampling point SP3) with a hydraulic retention time of 20 min, continuously mixed with an impeller powered by a DC motor (8.3 V at 0.1 A, McMaster Carr) at an average speed of 20 rpm. The velocity gradient in the tank during flocculation was approximately 65 s⁻¹.

(3)

Particle Separation

Effluent from the flocculation tanks flowed by gravity into a conical settling tank with a capacity of 416 L (sampling point SP4). The effluent entered the settling tank near the bottom. The hydraulic retention time within the settling tank was approximately 42 min. The sludge accumulated at the bottom of the tank and was pumped to a sand bed for dewatering and storage (Figure S9). Clarified water rose to the top of the tank and exited into a 120-L holding tank (sampling point SP5), where a float-controlled submersible pump (0.5 hp) automatically transferred the clarified water to the sand filters.

Additional particle removal was achieved with a rapid sand filter (PureWaterProducts, Denton, TX, USA, Zeolite Backwashing Filter, filtration rate 44.5 m³/m²/h), followed by two pleated sediment filters in series (Aquaboon 5 Micron, Oceanside, NY, USA). The filtration rate through the rapid sand filter was approximately 11.8 m³/m²/h. Periodically, the rapid sand filter was backwashed when the turbidity leaving the rapid sand filter approached 1 NTU.

(4)

Sampling Plan and Wet Chemical Measurements

Samples were collected at six locations (labeled SP1 through SP6, Figure 1) on-site in 250 mL beakers. Water quality parameters such as pH, DO, and conductivity were measured using a portable meter (ThermoScientific, Orion Star A Series, Waltham, MA, USA), with the probes calibrated daily. Turbidity was measured with a portable turbidimeter (ThermoScientific, Orion AQ4500), calibrated daily to EPA standards.

Two sets of samples were acidified immediately after collection with 0.5% HCl and 1% HNO₃ for later analysis. One set of samples, unfiltered, was used to determine the total concentrations of arsenic, iron, and aluminum at specific locations during the treatment process. The second set of samples were syringe-filtered through a 0.2 µm filter prior to acidification. This set was used to measure the concentrations of species remaining in the solution. ICP-OES (PerkinElmer 5300 DV, Springfield, IL, USA) and ICP-MS (Agilent 7700, Santa Clara, CA, USA) instruments were used to analyze the unfiltered and filtered samples, respectively. During operation, total dissolved arsenic was estimated at SP1 and SP6 using a portable rapid-test kit (Wagtech Arsenator, Palintest, Gateshead, UK). The Arsenator data are not reported here; their purpose was to verify, in real time, the adequate removal of arsenic.

(5)

Faradaic Efficiency of Air Cathodes in H₂O₂ Production

In ACAIE, H₂O₂ generation by the air cathodes is critical for the rapid oxidation of Fe(II) and rapid arsenic removal. Faradaic efficiency describes the conversion efficiency of charge (electrons) consumed at the air cathode/electrolyte interface into a 2e⁻ reduction of O_{2 (air)} for producing H₂O₂ (Equation (2). The Faradaic efficiency of freshly prepared air cathodes was evaluated by measuring H₂O₂ production in the catholyte compartment of a dual-chamber electrochemical cell. In this experimental setup, a mixed metal oxide anode (25.4 cm × 25.4 cm) was used against the freshly prepared air cathode. A cation exchange membrane (CMI-7000, Membrane International, Ringwood, NJ, USA), placed between the anode and air cathode, prevented H₂O₂ oxidation at the anode. The active electrochemical surface area of each electrode was 506.25 cm².

A simple electrolyte solution consisting of 100 mM Na₂SO₄, 10 mM NaCl, and 10 mM borate buffer (pH 8.0) was used as both the anolyte (2.25 L) and catholyte (2.25 L). Experiments were conducted at a charge dosage of 300 C/L and a charge dosage rate of 69.3 C/L/min. Immediately after electrolysis, bulk H₂O₂ in the catholyte was measured using a DR6000 UV-vis spectrophotometer (HACH, Loveland, Colorado, USA) with the titanium sulfate method at 405 nm [36]. The Faradaic efficiency of the air cathodes was calculated, based on the measured H₂O₂ concentration and the theoretical H₂O₂ concentration predicted by Faraday’s law.

The Faradaic efficiency of the air cathodes gradually decreased over time, necessitating regeneration. To restore performance, the cathodes from four ACAIE reactors were regenerated by circulating a 1% ascorbic acid solution. The reactors were disassembled, and all gaskets and HDPE components were thoroughly rinsed until visibly clean. Subsequently, the reactors were reassembled using the air cathodes without iron anodes. Approximately 100 L of the 1% ascorbic acid solution was recirculated for 30 h to remove the precipitates that had accumulated on the cathode surfaces. Samples were collected during the process to assess the composition of the regenerated solution. After completing the regeneration cycle, the reactors were rinsed with tap water and disassembled. The regenerated air cathodes were mounted into a dual-chamber electrochemical cell for Faradaic efficiency measurements.

3. Results and Discussions

This section is divided into three important sections: laboratory results, field results, and economic analysis.

3.1. Results from the Laboratory Testing of an ACAIE System with Allensworth Groundwater

3.1.1. Minimum Charge Dosage for Arsenic Removal Below Safe Levels

In iron-based electrocoagulation, charge dosage (C/L) strongly influences arsenic removal and must be optimized for a given groundwater composition [25,27]. Unlike in conventional iron electrocoagulation systems, the charge dosage rate (C/L/min) or current density (mA/cm²) in ACAIE has a minimal effect on arsenic removal due to rapid reaction kinetics [32].

To determine the minimum charge dosage required to reduce arsenic levels in Allensworth groundwater to below the EPA-MCL requirement of 10 µg/L, 0.5 L batch-scale experiments were conducted at two charge dosage rates: 30 C/L/min and 300 C/L/min (Figure 2). As expected, the dissolved arsenic level progressively decreased from 250 µg/L to 0.1 µg/L with an increase in charge dosage from 100 C/L to 600 C/L. This is because more adsorption sites are available for As(V) adsorption with an increase in iron dosage [37,38]. These findings are consistent with prior published literature on iron electrocoagulation with external H₂O₂ addition for Fenton-reaction-driven pollutant removal [14,31,39,40,41,42,43]. The minimum dosage required to achieve arsenic levels below 10 µg/L was 100 C/L at both charge dosage rates. The minimum dosage required to achieve arsenic levels below 10 µg/L was 100 C/L at both charge dosage rates.

The behavior of dissolved arsenic followed a similar trend at dosage rates of 30 C/L/min (3.9 mA/cm²) and 300 C/L/min (39 mA/cm²). The effect of charge dosage rate or current density on arsenic removal was minimal, due to the rapid oxidation of Fe(II) by H₂O₂ (k_{app_ H2O2} = 104.5 M⁻¹ s⁻¹) to form insoluble iron (oxyhydr)oxides, which have a very high adsorption affinity for As(V) [37,38,40,41,42,44,45].

Tables S1 and S2 summarize the wet chemical parameters (pH, DO, conductivity) from the bench-scale ACAIE experiments and the average initial composition of the Allensworth groundwater (column “SP1”), respectively. The charge dosage rate did not affect the solution’s pH, with the final pH values remaining close to the initial levels. In contrast, the dissolved oxygen nearly doubled by the end of electrolysis at both 30 C/L/min and 300 C/L/min. These findings are consistent with previously published results on iron electrocoagulation systems [32,42,43].

3.1.2. Residual Hydrogen Peroxide

In ACAIE, Fe(II) and H₂O₂ are generated in an equimolar ratio at the electrodes. In the bulk solution, short-lived strong oxidants (e.g., hydroxyl radical, Fe(IV), and superoxide radical) are produced during the oxidation of Fe(II) species by H₂O₂ or dissolved oxygen. These strong oxidants can further oxidize freshly generated Fe(II), leaving some unreacted, residual H₂O₂.

Figure 3 shows the concentration and fraction of residual H₂O₂ at charge dosage rates of 30 and 300 C/L/min. The charge dosage rate strongly influenced the residual H₂O₂ concentration. At 30 C/L/min, residual H₂O₂ steadily increased from 0.05 mM to 0.17 mM as the charge dosage rose from 100 C/L to 600 C/L. Conversely, at 300 C/L/min, residual H₂O₂ increased more rapidly from 0.05 mM to 1.01 mM over the same charge dosage range. At a charge dosage of 300 C/L, approximately 0.09 mM and 0.4 mM of H₂O₂ remained as unreacted, corresponding to 6% and 26% of the theoretical H₂O₂ expected at this dosage.

This behavior is likely attributable to differences in the rates of H₂O₂ decomposition and the generation of strong oxidants in the presence of Fe(III)(oxyhydr)oxides formed at 30 C/L/min and 300 C/L/min. Previous studies have indicated that the crystallinity, surface area, and structure of iron oxides significantly impact H₂O₂ decomposition and the yields of strong oxidants [32,46,47,48,49]. A longer electrolysis duration of 20 min at 30 C/L/min allows more time for H₂O₂ to undergo self-decomposition than the shorter 2-minute electrolysis period at 300 C/L/min.

3.2. Results from Field Testing of Pilot Scale ACAIE System in Allensworth

Based on the results of the bench-scale experiments discussed in Section 3.1.1 (Figure 2), a charge dosage of 100 C/L was sufficient to remove enough arsenic to reach values below 10 µg/L. At 200 C/L, a value of about 1 µg/L was achieved. For field testing of the pilot-scale ACAIE system, we selected a somewhat higher value for the charge dosage (300 C/L) to include an engineering safety factor for non-ideal conditions (e.g., non-uniform mixing, hydraulic short circuits, etc.).

3.2.1. Characteristics of the Treated Water

(1)

Total arsenic

Figure 4 shows the total arsenic content in the treated water from the ACAIE 600 Liter per hour (LPH) pilot plant, as measured from 24 May 2022 to 5 November 2022. The average initial arsenic in the raw groundwater was approximately 252 ± 55 µg/L. The total arsenic content in the treated water always remained well below the EPA-MCL, staying under 5 µg/L throughout the field testing. A slight, steady increase in total arsenic content was observed near the end of the field-testing period (September 2022 samples, 0908 to 0927), likely due to the fouling of the air cathodes and poor particle separation. After the air cathodes were regenerated, the total arsenic content in the treated water followed a steady or decreasing trend and remained below 1 µg/L. The average arsenic concentration in the treated water was 1.2 ± 0.6 µg/L. The maximum contaminant level goal is defined as the level of a contaminant in drinking water below which there is no known or expected risk to health. The U.S. EPA set the maximum contaminant level goal for arsenic in drinking water to 0 µg/L, indicating that no amount of arsenic is considered safe for consumption. Based on technological and economic constraints, the regulators established the enforceable MCL at 10 µg/L [49]. Because of arsenic’s acute toxicity, even at low concentrations, some countries are revising their MCLs to below 1 µg/L [50,51].

(2)

Total iron and total aluminum contents

In the ACAIE treatment process, aluminum sulfate was added post-electrolysis to increase the settling velocities of the arsenic-laden Fe(III)(oxyhydr)oxide particles formed during electrolysis. It is critical to achieve efficient particle separation, such that the total iron and total aluminum concentrations in the final samples of treated water remain below their respective EPA secondary MCLs.

Figure 5 shows the total iron (A) and aluminum (B) concentrations in the treated water. The average total iron concentration in the treated water was around 0.2 mg/L, and the average total aluminum concentration in the treated water was around 0.1 mg/L. The total iron concentrations remained below the EPA secondary MCL of 0.3 mg/L for most of the duration of the trial. Similarly, the total aluminum concentration remained below the EPA secondary MCL of 0.2 mg/L.

Turbidity measurements in the treated water followed a trend similar to those for the total iron and total aluminum concentrations (Figure S5). The total iron and total aluminum concentrations remained below their respective secondary MCLs, whenever the turbidity was below 1 NTU. The strong linear correlation between turbidity measurements and total iron concentrations supports the hypothesis that inadequate particle removal contributed to their high values (Figure S14A). The inadequate performance of the filtration system, particularly the cartridge filters, caused both occasional high levels of iron and aluminum (Figure 5) and the corresponding turbidity spikes (Figure S5) observed in the later part of the field test (after 7 August 2022). Operators routinely backwashed the rapid sand filter once every two weeks, whereas the cartridge filters were only backwashed when the effluent turbidity exceeded 1 NTU. Backwashing the 5-micron cartridge filters reduced the effluent total iron, total aluminum, and turbidity values to below the U.S. EPA-recommended limits for drinking water.

The solids loading rate of the filters determines the frequency of backwashing. Factors such as the influent water composition, the addition of chemicals (alum, in this case), and the amount of iron (oxy)hydroxides generated during electrolysis can be expected to influence the frequency of backwashing. Future real-world installations of ACAIE will need to monitor for filter breakthroughs and develop guidelines for backwashing frequency from site to site, as is similar to standard practice in municipal water treatment plants [52].

(3)

Bulk solution pH and dissolved oxygen

The bulk solution pH in the final samples of treated water remained below the initial pH of the raw groundwater during the testing period (Figure S3). The average pH decreased from 7.9 ± 0.3 in the raw groundwater to 7.2 ± 0.2 in the final samples of treated water (SP6, Table S3). During electrolysis, the solution pH remained at the initial value of 7.9 ± 0.3, due to the buffering effect of the hydroxide ions generated at the cathode [53,54]. However, the solution pH dropped to 7.1 ± 0.2 after adding aluminum sulfate as a coagulant. The hydrolysis of Al³⁺ species releases protons [55]; hence, a slight drop in pH was observed. The alkalinity of this groundwater (1.93 mM) prevented significant pH changes during coagulation (Table S2) [55].

In contrast to pH, the dissolved oxygen (DO) in the final samples of treated water followed a steady decreasing trend (Figure S4), likely due to the steady fouling of the air cathodes. However, the DO increased slightly from 4.9 ± 1.2 mg/L at the beginning of electrolysis to 7.1 ± 0.7 mg/L in the final samples of treated water. The aeration has likely caused a steady increase in DO at each unit process (Table S3).

During the anodic dissolution of Fe(0) plates, impurities (e.g., Mn) in the steel plate are released into the solution. Typically, these impurities are present in trace quantities and are often adsorbed or co-precipitated with Fe(III)(oxyhydr)oxides. However, changes to the solution pH at various unit processes can influence the removal of these contaminants. The total Mn concentration in the post-electrolyzed water was 155.4 ± 45.6 µg/L (Table S2) in the unfiltered samples. However, the dissolved manganese concentration in the post-electrolyzed water (Sampling Point SP2, Figure 1) remained below the EPA secondary MCL value of 50 µg/L (Figure S6) in all but three samples. This indicates that Mn adsorption or co-precipitation by the Fe(III)(oxyhydr)oxides removed a significant fraction of the Mn released by Fe(0) plates [56,57]. However, Mn concentrations in the final samples of treated water (Sampling Point SP6, Figure 1) were well above the EPA secondary MCL value; the average Mn concentration in the treated water was 126 ± 54 µg/L (Figure S7). The pH decreased after the addition of a coagulant, which likely caused the desorption or re-release of Mn into the solution, as is consistent with the published literature [56,57,58].

An aluminum anode electrocoagulation method could be used instead of aluminum sulfate chemical addition to deliver the desired dose of coagulant. This method would minimize the pH changes caused by Al³⁺ hydrolysis while still efficiently removing potential secondary contaminants like Mn.

3.2.2. Air Cathode Performance in the Field

The average Faradaic efficiency of H₂O₂ generation with new (#10), used (#8), and regenerated (#8) cathodes is shown in Figure 6A. New air cathodes show a very high Faradaic efficiency of 86%, which is consistent with previous measurements [32] (Figure S1 and Table S4). Over time, the Faradaic efficiency of air cathodes tends to decrease due to the accumulation of iron oxides on the electrode surface. During field testing, the average Faradaic efficiency of the 10 air cathodes decreased to 42% after a cumulative operation time of 135 h (27 September 2022; see Figure 4). Each of the 10 cathodes used in the five ACAIE reactors shows a similar decline in Faradaic efficiency, indicating uniform fouling of the cathodes (Figure S2). This also indicates a uniform anodic dissolution of the Fe(0) anodes. The average total iron concentrations measured at the end of electrolysis were closer to the expected Fe at 300 C/L (Figure S8).

The regeneration of used air cathodes (Section 2.2.2) restored their Faradaic efficiency to about 80%. Over 30 h of exposure to 1% ascorbic acid removed the surface layers that had accumulated on the air cathodes (Figure 6B–D).

The elemental composition of the regenerated solution revealed that the surface layers consist primarily of Ca (67.5%), Fe (23.8%), Mg (7.2%), and Si (1.1%) (Table S6). The fraction of arsenic attached to the air cathodes was negligible, suggesting the primary removal pathway for arsenic is by its adsorption onto Fe(III)(oxyhydr)oxides.

3.3. Economic Analysis of the ACAIE Community-Scale Plant

To assess the economic feasibility of the ACAIE treatment process, a comprehensive cost analysis was conducted over the operational period of this small community-scale plant. This period includes the phases of (1) planning, (2) purchasing, (3) construction, and (4) two months of operation. Our economic evaluation considers two primary factors: capital expenditures and operating expenditures (both fixed and variable), excluding researcher labor costs, land acquisition costs, and costs outside the fence line.

Costs typically categorized under engineering and construction, contingency, and working capital were excluded from this financial analysis for the following four reasons: (1) UC Berkeley performed the engineering design work using academic funding resources, (2) the Allensworth community contributed voluntary labor to some of the construction work, (3) no working capital was needed, and (4) no contingency funds were allowed.

Additionally, the costs associated with financial management and oversight, the procurement of equipment and supplies, expenditure control in relation to invoices, and project management were aggregated by the external parties overseeing the project funds. These charges are not detailed individually, making it difficult to pinpoint specific costs. However, using UC Berkeley’s overhead rate of 60% on all costs allows for a rough estimation. While larger, established organizations often execute routine tasks more efficiently, this 60% markup for administrative and overhead costs has been factored into the estimate. Before accounting for management overheads, the total capital expenditure for the 600 LPH ACAIE pilot plant was calculated at USD 28,500 (reflecting 2021–2022 prices). If adjusted to December 2024 prices, this would be USD 31,920, as discussed in Supplementary Materials Section S1.2. Within the fence line, costs amounted to USD 14,100 (as shown in

Table 1. Costs inside the fence line of the ACAIE 600 LPH system.

Category	Item	Total (in USD)
Electrical	Power Supplies	3010.00
Electrical	Power Strips, Electrical Wires, and Air Conditioning Units	745.00
Plumbing	Hoses, CPVC Piping and Fittings, Flow Meters, and Check Valves	1368.64
Plumbing	Submersible Pumps	392.00
Structure	Supports for ACAIE Reactors, Flocculation, and Storage Tanks	110.00
Treatment	Treated Water Storage Tank	4000.00
Treatment	Flocculation Tank, Holding Tank, Post-electrolysis Tank, and Settling Tanks	657.00
Treatment	In-line Static Mixer and Dosing Pump	456.52
Treatment	Sand Filter, Micron Filters, and Filter Housing	2937.22
Treatment	ACAIE Reactors	425.00
		USD 14,101.38

), while costs outside the fence line, which include the transformer, concrete pad, and treatment shed, totaled USD 14,400. In this context, the fence line refers to the physical boundary of the arsenic removal facility.

Adding the 60% overhead rate increases the total capital expenditure estimate to USD 45,600. The annualized capital cost of the ACAIE plant, assuming a 10-year lifespan, zero value at the end of 10 years, and 5% interest, is, therefore, USD 5804.

Annual operating costs include both fixed and variable costs. Fixed costs typically cover expenses such as rent, labor, insurance, land, permits, and overall maintenance. However, for this project, the Allensworth community partners provided land at no cost, and no rent was incurred. Maintenance and operational tasks were performed by UC Berkeley engineers, whose labor costs were not included in this analysis. As a result, the total fixed operating costs for the year are considered negligible. These costs are also location-specific and would need to be adjusted based on where the water treatment plant is located. Additional costs excluded from this analysis include: all costs related to rent, operator labor, land, permits and routine quality control, testing, equipment calibration, and maintenance. Disposal costs for dry arsenic-bearing sludge are negligibly small and are not included in the analysis; however, if the arsenic removal process leads to liquid waste discharge, the disposal cost can be exorbitant. For example, the very small community water system of Park Royal Mutual Water Company in Sonoma County, California found that they could not afford to operate an arsenic removal plant primarily because of the monthly disposal costs of arsenic-bearing liquid concentrate (Senior officer, personal communication, 10 August 2022).

Variable operating expenses include consumables such as iron anodes and air cathodes for the ACAIE reactor, electricity, and aluminum sulfate used in the coagulation/flocculation process. We envision that two additional sand filters used in series will negate the need for pleated micron filters. Hence, these costs are not included in these estimates. Under these assumptions, the variable costs for 1 year total USD 5750, as shown in

Table 2. Direct variable operating costs (at 2021–2022 prices).

Item	Cost (USD)	Unit	Count	USD/year
Iron anodes	37	USD/plate	60	2220
Air Cathodes	100	USD/cathode	20	2000
Electricity	0.20	USD/kWh	7200	1440
Aluminum sulfate	2.0	USD/kg Al2(SO4)3	46	90
Variable annual cost				5750

. The table assumes operation for 4 h per day, 365 days per year. Values in the column titled ‘Count’ (Table 2) are scaled up from the observations over the duration of the field test.

These costs also require management and support functions (e.g., invoices to be paid, control of expenses, accounting, and placing orders for and receiving consumables). Therefore, they too are subject to a reasonable overhead markup of 60%, increasing the annual variable costs to USD 9200.

A key finding of this economic analysis is that the air cathode costs are as significant as the iron anode costs; together, these are the two materials that most affect the overall variable operating cost of running the plant, as seen in Table 2. Extending the usable lifetime of air cathodes reduces labor and operating costs. Regenerating used air cathodes can restore their performance. We assume that regeneration approximately doubles the lifespan of air cathodes. In the above estimate, it was assumed that the regeneration process was used to extend the life of the air cathodes.

Average daily drinking water consumption is assumed to be 4 L per person per day (averaged over a community of adults and children). For a small town like Allensworth (population 600), drinking water consumption is approximately 2400 L per day. Treated water production at those volumes would require the use of the (600 LPH) pilot plant for 4 h per day (876,000 L per year). Using this value in the denominator, and the annual costs in the numerator, the cost per liter of water treated by the ACAIE system is estimated at (USD 5804 + USD 9200)/876 Kiloliters = USD 0.017/L, or about 2 cents per liter. Note that this estimate does not include the costs related to rent, operator labor, land, permits, routine quality control, testing, equipment calibration, and maintenance. The additional costs of marketing, public education to ensure water purchases, sales efforts, interest on borrowed working capital, and a business margin for investors in a for-profit enterprise must also be added if this system were to be operated as a business.

If labor costs were included, the cost per liter of treated water would increase substantially. Assuming one full-time operator with a labor cost of USD 30 per hour (including benefits) for 52 working weeks per year, this results in an additional annual cost of USD 62,400. This added labor cost would increase the cost of treated water from the previously mentioned 2 cents per liter to approximately 9 cents per liter. In contrast, remote monitoring and substantial automation can reduce operator hours and substantially decrease the costs of drinking water per liter. In all cases, the amount remains significantly below the baseline costs of purchasing water from a nearby town (66 cents per liter).

4. Concluding Remarks

These results demonstrate the excellent technical capability of the ACAIE system in removing arsenic from groundwater, reducing concentrations from a high level of 250 µg/L to levels below the EPA-MCL of 10 µg/L. A practical, low-cost regeneration method was demonstrated to enhance the system’s durability. Estimates presented in this work indicate that the cost of annual amortized materials and consumables, excluding operator and maintenance expenses, is modest, at around USD 6000 per year for a 600 LPH system. For a community of 600 people, each consuming an average of 4 L of drinking water per day, the cost of ACAIE-treated water is estimated to be around 2 cents per liter at 2022 prices. However, incorporating labor costs could increase the price of water to around 9 cents per liter. These findings underscore the importance of minimizing operator costs with a simple, automated, low-maintenance treatment system to make safe drinking water more affordable for low-income, rural communities.