Evaluating Sustainable Development Pathways for Protein- and Peptide-Based Bioadsorbents for Phosphorus Recovery from Wastewater

Abstract

Recovering phosphate (P) from point sources such as wastewater effluent is a priority in order to alleviate the impacts of eutrophication and implement a circular economy for an increasingly limited resource. Bioadsorbents featuring P-binding proteins and peptides offer exquisite P specificity and sensitivity for achieving ultralow P concentrations, i.e., <100 μg P L^–1, a discharge limit that has been implemented in at least one treatment facility in nine U.S. states. To prioritize research objectives for P recovery in wastewater treatment, we compared the financial and environmental sustainability of protein/peptide bioadsorbents to those of LayneRT anion exchange resin. The baseline scenario (reflecting lab-demonstrated performance at a full-scale implementation) had costs that were 3 orders of magnitude higher than those for typical wastewater treatment. However, scenarios exploring bioadsorbent improvements, including increasing the P-binding capacity per unit volume by using smaller P-selective peptides and nanoparticle base materials and implementing reuse, dramatically decreased median impacts to $1.06 m^–3 and 0.001 kg CO₂ equiv m^–3; these values are in line with current wastewater treatment impacts and lower than the median LayneRT impacts of $4.04 m^–3 and 0.19 kg CO₂ equiv m^–3. While the financial viability of capturing low P concentrations is a challenge, incorporating the externalities of environmental impacts may provide a feasible path forward to motivate ultralow P capture.

Short abstract

Capturing ultralow concentrations of phosphate from wastewater is increasingly required in discharge permits. This study prioritizes research pathways to implement sustainable phosphate capture by using protein and peptide bioadsorbents.

Introduction

Alchemist Hennig Brand made history in a most unusual way—by boiling a lot of urine.¹ While Hennig did not meet his objective of turning metal to gold, he isolated something even more valuable: phosphate (P).¹ Unbeknownst at the time, P is an essential element for biological growth and, hence, a necessary component of fertilizers. P is a nonrenewable, irreplaceable resource with an uneven global distribution of reserves,² which makes securing sufficient P resources a priority for addressing life’s bottleneck of ensuring sufficient P for food security.³

Paradoxically, too much P (through its uncontrolled release into the environment) is a nuisance. In the linear food-to-fork flow of modern food production, only about 16% of the P that is applied as fertilizer ends up in human food; to varying extents, that “lost P” contributes to water pollution.⁴⁻⁶ Excess P contributes to eutrophication, hypoxia, harmful algal blooms, the loss of fish habitat, and economic repercussions.^5,7,8 Eutrophication has been described as the “biggest overall source of impairment of the nation’s rivers and streams, lakes and reservoirs, and estuaries.”⁹ Moreover, some P-limited waters are particularly susceptible to eutrophication, even at “ultralow” μg L^–1 P levels.¹⁰

To address this crisis that impacts environmental health and food security, which directly influence the progress toward the United Nations’ Sustainable Development Goals (SDGs) 6 and 2, respectively,^11,12 P removal from wastes coupled with P recovery is needed. Eighty percent of the publicly operated treatment works (POTWs) in the U.S. with P effluent permits regulate effluent at levels of ≤1 mg L^–1 (see the Supporting Information (SI), Section 1). Given P limitations in receiving waterbodies, 26% of POTWs have ultralow permits in the μg L^–1 range, with 17% of the facilities at ≤500 μg L^–1 and 2% at ≤100 μg L^–1. Conventional P treatment strategies, such as biological nutrient removal, adsorption, crystallization, and precipitation, are typically unable to reduce P below approximately 100 μg L^–1 due to thermodynamic and kinetic limitations.¹³⁻¹⁸ Thus, advanced strategies are needed to satisfy increasingly stringent P regulations.¹⁰ Moreover, P recovery is an increasingly important consideration, given the potential of capturing 3.7 Mt P year^–1 from municipal wastewater (which is approximately 20% of agricultural demands).¹⁹ Coupling the value of the recovered P with recovered energy, nitrogen, and water equates to $1.8 million per year in potential wastewater-recovered resources (for a 100 000 m³ d^–1 facility).²⁰

To address this dual need for P removal to ultralow levels and P recovery for beneficial reuse, P-specific bioadsorbents offer strong potential for wastewater treatment.²¹ Bioadsorbents featuring amino acid-based macromolecules, including immobilized, high-affinity phosphate-binding proteins (PBP) or phosphate-binding peptides (PBPep), are one option.²²⁻³⁰ PBP features extraordinary P sorption kinetics²⁷ and specificity, enabling P extraction from complex matrices containing other anions (e.g., arsenate),^28,31 even when P is present at initially low levels. Immobilized PBP is reusable, using pH to control adsorption and desorption cycles. The elevated pH desorption phase releases P in a concentrated, pure solution, enabling the recovery of high-value end products. However, the current P-adsorption capacity of immobilized PBP systems lags behind that of other materials. Improvements such as integrating support materials to facilitate higher PBP loading or PBPep engineering to reduce the protein footprint to a shorter peptide sequence while retaining P-specificity may improve capacity. In tandem with technical advancements, the cost and environmental implications of protein- or peptide-based bioadsorbents must be considered.

The development of emerging technologies such as bioadsorbents should be informed a priori by sustainability considerations.³² In particular, prospective techno-economic assessment (TEA) and life cycle assessment (LCA) include economic and environmental performances as design constraints rather than afterthoughts; however, this approach is challenging given the inherent lack of existing data.^32,33 Examples of TEA and LCA guiding the development of emerging water treatment technologies include catalytic nitrate reduction,³⁴ struvite precipitation,³⁵ and perchlorate reduction.³⁶

As a nascent technology, the economics and environmental impacts of bioadsorbents for P removal and recovery have not been previously evaluated; however, this information is vital for guiding the research for sustainable development pathways. Here, we integrated performance models with TEA and LCA in a Monte Carlo framework to do the following: (1) assess the economic and environmental implications of using the current PBP state of technology to remove and recover P from municipal wastewater; (2) plot the research and development of PBP, including a reduction in the surface area of the support matrix and a transition to a peptide-based sorption unit, and evaluate the transition to the porous adsorption and reuse cycles required to achieve sustainable P removal and recovery; and (3) explore regulatory drivers to motivate P removal and recovery. Protein- and peptide-based P removal and recovery were benchmarked against the P-selective LayneRT anion exchange resin. The findings can inform technology development pathways for protein- and peptide-based bioadsorbents.

Methodology

Bioadsorption Modeling Scenarios: Advancing PBP and PBPep Research

To evaluate the sustainable development pathways for P removal and recovery at low concentrations, six protein and peptide scenarios were evaluated via TEA and LCA: (1) PBP immobilized on Sepharose beads without reuse, (2) PBP immobilized on iron oxide particles (IOPs) without reuse, (3) PBPep immobilized on IOPs without reuse, (4) PBP immobilized on IOPs with reuse, (5) PBPep immobilized on IOPs with reuse, and (6) PBPep in covalently linked aggregates (Table 1). Scenario 6 was used to study advancements in extended reuse cycles and increased P capacity per unit mass of biosorbent material.

Table 1. PBP Technology Development Pathways Evaluated in the Six Study Scenarios^a.

^aImages created with BioRender.com.

Scenario 1 (the baseline scenario) was developed based on lab-scale experimental results.²⁷ Scenarios 2–5 reflect potential improvements in bioadsorbent performance compared to the baseline scenario. Specifically, scenario 2 explores increases in the P-adsorption capacity by increasing the surface area of the immobilization matrix, modeled here as IOPs.³⁷ IOPs feature a smaller diameter and a larger surface-area-to-volume ratio compared to lab-grade Sepharose beads (greater detail on the selected immobilization matrices is provided in the SI, Section 3). Scenarios 3–5 reflect prospective future technological improvements. In scenario 3, increases in the P-adsorption capacity were derived from using a higher matrix surface area (IOPs) and replacing the PBP with a smaller PBPep, here modeled as a P-loop peptide, which has been shown to effectively bind P.³⁸ The smaller size of the PBPep relative to the PBP (having a radius of 5.3 Å versus 21.5 Å, respectively) improves the value proposition by increasing the P-adsorption capacity metric (mass P:volume sorbent). Scenarios 4 and 5 illustrate the impact of reusing the bioadsorbent for multiple sequential adsorption/desorption cycles. Consistent adsorption performance when using PBP immobilized on Sepharose beads has been previously observed over 10 cycles.³⁰ Here, 20 cycles were modeled to further explore the influence of extended reusability in scenarios 4 and 5.

A sixth scenario analyzed a range of ratios for the P-binding sites to support the matrix volume and reuse cycles. The range included up to 236× the P-adsorption capacity of PBPep–IOP (scenarios 3 and 5). This range represents the transition from a surface-based adsorbent to a porous nanoparticle adsorbent. The reuse cycles were increased an order of magnitude to 400 cycles (see the SI, Section 2 for more information). To minimize the noise associated with the influent P concentration, static values of 10 or 1 mg P L^–1 in the influent and 100 μg P L^–1 in the effluent were used.

Wastewater Characteristic Distributions for the Models

To varying extents, the P adsorption performance depends on wastewater characteristics when using different materials.^14,39,40 Accordingly, effluent P concentrations were acquired from the U.S. Environmental Protection Agency’s (EPA) 2019 Enforcement and Compliance History Online (ECHO) database for POTWs in all of the U.S. states. Discharge P concentrations were limited to ≤10 mg L^–1, and data for potentially competitive anions (i.e., bicarbonate, chloride, nitrate, and sulfate) were collected. Additional environmental factors influencing the technologies, such as the wastewater temperature (−1.11 to 37.78 °C) and pH (6–9), were incorporated into the analysis for each technology based on the ECHO data, as described below.

In addition to the scenarios focused on advancing PBP and PBPep research, we performed a Morris one-step-at-a-time sensitivity analysis to evaluate the sensitivity of the LCA/TEA models to initial P concentrations, which were modeled as the empirical data for existing U.S. POTW discharge regulations (summarized in the SI, Section 1). This approach enabled the exploration of the costs and environmental impacts of bioadsorbent usage in response to regulatory drivers (i.e., what were to happen if existing permits were reduced to the ultralow discharge target of 100 μg L^–1). The values generated in this theoretical assessment are limited in scope, as they do not account for realistic site variation beyond initial P concentrations, e.g., variations in the water quality parameters, costs of labor and transportation, etc.

Phosphate Adsorbent Models

Phosphate-selective adsorbent models were developed in MATLAB 2015a to enable the calculation of the number of bed volumes treated. Two models were utilized: one for the bioadsorbent scenarios and the other for LayneRT hybrid anion exchange (HAIX) media. LayneRT anion exchange media was included in this study as a benchmark for comparison because it is a commercially available adsorbent with demonstrated P selectivity and pH-controlled reusability.^14,39,40

The bioadsorbents were modeled as P-specific adsorbents unaffected by competing ions using single-layer Langmuir isotherm models.^23,27,38,41 Previous works have showed that the Langmuir coefficient (K_L) of PBP is dependent on pH and temperature.^27,30 Because the pH and temperature are expected to vary in wastewater, these dependencies were incorporated into our deterministic model. Discrete data were used to develop continuous functions by using multilinear regression. Langmuir coefficients were fit using quadratic equations for both temperature and pH (eq 1; SI, Section 5). Interdependency between the variables was allowed in the regression analysis, but the best fit (R² = 0.87) indicated no interdependency.

1

In eq 1, K_L is the Langmuir coefficient in L (μmol P)⁻¹, and T is temperature in °C.

Like the bioadsorbents, the commercially available benchmark adsorbent LayneRT also offers high specificity for P adsorption; however, it is impacted to a greater extent than the bioadsorbents by the presence of other anions in solution.^23,40,42 Therefore, it was essential to determine the selectivity coefficients of LayneRT corresponding to the influent concentrations of competing anions. Using the selectivity coefficient, the P removal performance was modeled assuming plug flow and the instantaneous equilibrium of the sorbed and aqueous phases, in accordance with previous studies.^34,36,43

Previous experimental work was used to determine the selectivity coefficients of the LayneRT resin.⁴⁴ Experimental data included the pH; the initial and final aqueous-phase concentrations of phosphate, bicarbonate, chloride, nitrate, and sulfate; and the sorbed-phase concentrations of the anionic constituents. Measurements differentiating between the monovalent and divalent phosphates in solution and those sorbed to resin were not provided. Accordingly, pH and pK_a were used to determine the relative ratio of the variably charged P species in solution. Because the ratio of the speciation of the sorbed P species and the selectivity coefficients for each species were unknown, the sum of the squared error difference between the experimental results and the selectivity coefficient equation (eq 2) was minimized to determine values. The error between the total molar mass of sorbed P measured in the previous experiments⁴⁴ versus the calculated total molar mass was minimized by varying the selectivity coefficients of the mono- and dibasic phosphate exchange with chloride.

2

In eq 2, K^BC_A is the selectivity coefficient of B (e.g., dibasic phosphate) exchanged with A (e.g., chloride), and C is the charge of B. The bar above the constituents indicates the sorbed phase.

Sustainability Analysis

Using TEA and LCA (per the ISO 14040/14044 standards), cost and life cycle environmental impacts were determined for a functional unit of 1 m³ of municipal wastewater treated to a designated level of 100 μg P L^–1.^45,46 In select scenarios, the functional unit of 1 kg of recovered P was also determined. Included in the system boundary were the production and operation of the P adsorption technology, which incorporated the manufacturing, chemical inputs, transportation of all components (P-adsorption technology and operational consumables), and electricity at each stage. The construction and demolition of the facility were excluded. Because this evaluation was primarily focused on the technologies, the useful life of the technologies was the basis for comparison (LayneRT: three years, which represents a median value of 821 reuse cycles; PBP and PBPep: variable, depending on the implementation of reuse cycles). The life cycle inventories and processes used to produce the bioadsorbents and LayneRT resin are available in Figure S4 (a materials flowchart) and Table S2 (the quantities).

All performance simulations and the life cycle impact and cost assessments were performed in MATLAB 2015a. The unit environmental characterization factors that were input into MATLAB were obtained using the ecoinvent database accessed via SimaPro (v9; PRé Sustainability) using unit allocation at the point of substitution (APOS, U). The Tool for the Reduction and Assessment of Chemical and Other Environmental Impacts (TRACI, v2.1) was used to determine the environmental impact of the raw materials and direct emissions for normalization. No other normalization was performed, except in the comparison of the eutrophication potential from the untreated release of P versus the eutrophication potential of P removal and recovery, where kg N equiv was converted to kg P equiv using a conversion factor of 0.42 kg P (kg N)⁻¹.⁴⁷ The global warming potential (GWP) was reported in detail, with additional environmental impacts noted in the SI (Sections 6–8).

Latin hypercube sampling and Monte Carlo analyses were used to evaluate the uncertainty in 46 parameters. This analysis was conducted under stochastic influent characteristics (10 000 random combinations of influent characteristics within the specified ranges described above and summarized in the SI, Table S4) for each of the scenarios tested here.

Statistical Analysis

The sensitivity of the treatment technologies to water quality inputs was determined using the Spearman’s rank-order correlation.^48,49 In scenario 6, Loess smoothing was used to observe the impacts of changing the P-binding sites to support the matrix volume and reuse cycles. Data were evaluated in the Origin 2018 and GraphPad Prism softwares. Unless noted otherwise, median values are presented.

Results

Targeted P Removal: State of Technology

To assess the potential economic and environmental impacts of targeted P removal in tertiary wastewater treatment, an initial assessment of PBP and LayneRT was performed. To ensure meaningful results, realistic distributions of water quality values were used, which were based on the U.S. EPA’s ECHO database. This baseline scenario (scenario 1, Figure 1) was selected because no substantial additional research and development would be required, and it represents the current state-of-the-art technology for both treatment systems. The removal of P to 100 μg L^–1, a stringent yet realistic regulation scenario emerging in permits across the U.S., was modeled. Additional research and development scenarios are considered later.

Figure 1.

(a) The costs and global warming potential (GWP) of PBP as modeled in scenario 1 for 1 m³ of wastewater treated to a target effluent P concentration of 100 μg L^–1. (b) Spearman’s rho (ρ) correlation values for the LayneRT ion exchange model inputs compared to the outputs (cost, $ m^–3) for the upper portion of the graph, as well as inputs compared to the performance (bed volumes treated) for the lower portion of the graph. Note that DB = dibasic and MB = monobasic. (c) Spearman’s ρ correlation values for the PBP technology model inputs, including production (P) and operation (O) factors, compared to the outputs ($ m^–3).

Under the PBP single-use scenario 1, the median cost to recover P was $3100 m^–3 (Figure 1), which is an unrealistic value, given that wastewater treatment costs range between $0.40 and $4.50 m^–3,⁵⁰ even if these costs do not account for P recovery. The median GWP was 3.6 kg CO₂ equiv m^–3. This value is closer to realistic indirect GWP emissions for wastewater treatment (5.8 kg CO₂ equiv m^–3).⁵¹ The costs for PBP treatment were predominantly driven by the production of PBP. The elevated costs for PBP treatment might be assumed to be attributed to the production of the protein, as has been found in a previous report examining protein applications in water treatment.³⁶ However, a breakdown of the costs revealed that the media required to produce PBP only accounted for approximately 1% of the total cost. Instead, the primary cost driver was the nickel affinity resin used in protein purification (8.8%) and the Sepharose resin used to immobilize PBP during treatment (89.4%), which together accounted for 98.2% of the total cost associated with producing the PBP bioadsorbent.

The large discrepancy between the cost and the GWP of the PBP treatment was also attributed to the use of the nickel affinity and Sepharose resins. While research-scale costs are available for the two materials, the environmental impacts for the two types of resins are not available in the literature or the environmental assessment databases. Accordingly, surrogates that were available in the database were modified and selected to incorporate the environmental impact of the two resins (Table S2). These surrogates represent emissions from full-scale production processes, which could account for a portion of the discrepancy between the (lab scale-based) cost and GWP estimates.

The correlation analysis corroborates the cost and GWP analyses, with the strongest correlation of 0.999 between the cost and the materials used to manufacture the PBP resin. This suggests that future development should focus on reducing the unit costs of the matrix and/or facilitating longer use (reuse) in treatment systems. Other factors strongly correlated with cost and GWP include resin disposal (0.984) and the total influent P concentration (0.998). Interestingly, pH (−0.01) and temperature (0.02) were not strongly correlated with the cost. The correlations could be affected by the parabolic shape of the surface response model for pH and temperature with the PBP performance (Figure S5). However, performing correlation analyses for trends on either side of the pH and temperature vertices did not improve the correlations. The cost correlation with pH from 6–6.5 and 6.5–9.0 was 0.03 and 0.02, respectively. Similarly, evaluating the correlation on either side of the maximum P loading temperature of 20 °C did not strengthen the correlation (−0.01 and 0.02, respectively).

In comparison, the LayneRT resin had a median treatment cost of $4.04 m^–3 and a median GWP of 0.19 kg CO₂ equiv m^–3 (Figure 1). These cost and GWP values are approximately 3 orders of magnitude and 1 order of magnitude lower than those of the PBP bioadsorbent, respectively. These results align with the expectations for a more mature technology, given that LayneRT was patented in 2004⁵² and has built on a long history of ion exchange resins used in water treatment. The factors driving the LayneRT resin cost included the resin (26.8%) and sodium hydroxide (69.1%) used to regenerate the resin over its useful life. In this analysis, onsite regeneration was assumed and would likely be required, as the resin had a median performance of 72 bed volumes. Including the transportation required to regenerate the resin would likely significantly increase the cost of the treatment. Onsite regeneration of the resin and the recovery of P, along with the development of more cost-effective regeneration solutions, could be a targeted research improvement for the LayneRT technology.

In this initial assessment, the costs and environmental impacts of the PBP and LayneRT technologies were greater than those of standard wastewater treatments, which may or may not include enhanced nutrient removal (and do not account for P recovery). Encouragingly, the costs driving the use of PBP were not dominated by the production of the protein but rather the purification and immobilization of the protein. Overall, similar trends were observed for all of the additional environmental impacts except for ozone depletion (Figures S6 and S7).

Research and Development Scenario Analysis

Emerging technologies frequently require research and development investment prior to full-scale translation, and protein-based technologies are no exception. Research and development pathways for PBP technologies were proposed and evaluated in five additional scenarios (Table 1). As described in the Methodology section, the research improvement pathways were compared against our baseline analysis (scenario 1). We incorporated an increase in the bioadsorbent capacity by increasing the surface-area-to-volume ratio of the support material (scenario 2) or decreasing the diameter of the functional P-binding group (scenario 3). Further improvements evaluated the ability to reuse the technology (scenario 4) and the incorporation of all of the improvements (scenario 5).

Decreasing the average diameter of the support material from 90 to 1 μm resulted in orders-of-magnitude decreases in the costs and GWPs of PBP technologies (Figure 2; other environmental impacts are shown in Figure S6). The median cost and GWP for scenario 2 were $109.95 m^–3 and 0.58 kg CO₂ equiv m^–3, respectively. The improvements were realized through the availability of more reactive surface area per unit volume of the support matrix. Similar to the observations in scenario 1, the purification resin and support matrix were still the primary drivers of the total costs and environmental impacts in producing the PBP–IOPs (4.9% of costs associated with PBP, 25.3% with nickel, and 69.8% with IOPs). Moreover, the use of IOPs in place of Sepharose beads decreased the relative proportion of the cost associated with the immobilization matrix (69.8% versus 89.4%). Increasing the adsorption capacity per volume of the support matrix dramatically improved the economics of the PBP bioadsorbents. The adsorption capacity was further increased in scenario 3 by decreasing the size of the P-binding unit by using PBPep in lieu of PBP, effectively decreasing the average radius from 21.5 to 5.3 Å. This increased capacity further reduced the cost and GWP impact to median values of $19.66 m^–3 and 0.02 kg CO₂ equiv m^–3, respectively. The costs and materials associated with the production of PBPep were modeled as equivalent to those associated with the production of PBP, given that no data yet exist for this process (see the SI, Section 4). Future studies are needed to establish the life cycle inventory associated with PBPep.

Figure 2.

(a) Cost ($ m^–3) and (b) global warming potential (GWP, kg CO₂ equiv m^–3) of phosphate-binding protein (PBP) and phosphate-binding peptide (PBPep) bioadsorbents and LayneRT anion exchange resin for 10 000 scenarios. The different research achievements highlighted by the scenario analysis are summarized in Table 1. The median value for each scenario is represented by the horizontal line in each box. The edges of each box are the 25th and 75th percentiles, while the whiskers represent the 10th and 90th percentiles.

Implementing P recovery and protein/peptide reuse was the final strategy we evaluated to improve the economic and environmental impacts. Scenario 4 evaluated 20 cycles of reuse of PBP immobilized on IOPs. These improvements brought costs to $5.58 m^–3 and GWP to 0.001 kg CO₂ equiv m^–3. Evaluating the combination of the technological improvements, including the immobilization of PBPep on IOPs with reuse, yielded a cost of $1.06 m^–3 and and GWP of 0.001 kg CO₂ equiv m^–3.

Throughout the analysis, the discrepancy between the cost and the GWP persisted. In contrast, as highlighted above, the economic cost for wastewater treatment was as high as $4.50 m^–3, which was similar in magnitude to the GWP (5.8 kg CO₂ equiv m^–3). This highlights an underlying assumption in the current study: the base costs for the technologies were assumed to be static and had much higher unit costs than the unit GWP. However, if P capture technologies were to be implemented on a wider scale, efficiencies in the costs for the input materials might be realized, thus better aligning the costs and GWPs.

Targeted Improvement to Achieve Sustainable P Capture

To further evaluate the improvements, the number of active sites per unit volume and the number of reuse cycles were assessed over a range (scenario 6, see Figure 3). These improvements decrease the cost to recover P, with approximately 34% of the 10 mg L^–1 scenarios achieving costs less than $6 (kg P)⁻¹, a critical economic threshold for implementing sustainable P recovery (see the SI, Section 9). At a higher influent concentration (10 mg P L^–1), the cost to treat 1 m³ of water is higher than in the 1 mg P L^–1 scenario (with a median value of $0.077 m^–3 versus $0.069 m^–3, respectively); this result was expected, as more sorbent material is required to remove the higher concentration. However, when normalized to kg P removed and recovered, the trend is the opposite, where the higher concentrations have lower costs (Figure 3; $7.79 (kg P)⁻¹ versus $76.40 (kg P)⁻¹); this represents the requirement to pump ∼10× more water volume to capture 1 kg P, indicating that operational costs play an important role in the cost to recover P. The operation of multiple P recovery strategies (e.g., enhanced biological phosphorus removal (EBPR), followed by a polishing step) may not yield maximum sustainability benefits. However, in instances where conventional P treatment alternatives cannot achieve ultralow P levels, a polishing step may be necessary to satisfy stringent P discharge limits.

Figure 3.

Contour plots showing the (a) cost ($ (kg P)⁻¹) and (b) GWP (kg CO₂ equiv (kg P)⁻¹) over a range of reuse cycles and ratios of increased P-sorption capacity relative to the PBPep–IOP (scenarios 3 and 5) sorption capacity at 10 mg L^–1 influent. Ratios >1 assume a transition from surface-based sorption to a porous-based sorption mechanism. The black contour line in (a) emphasizes the $10 (kg P)⁻¹ threshold from our study, which is comparable to the associated cost of enhanced biological phosphorus removal, as noted in previous studies.⁶⁴ The black contour line in (b) emphasizes the 1.6 kg CO₂ equiv (kg P)⁻¹ threshold, which is comparable to the approximate 1.45 kg CO₂ equiv (kg P)⁻¹ required to mine P. (c) Box and whisker plot of the PBPep–IOP and LayneRT technologies with fixed influent concentrations at 1 and 10 mg P L^–1. The middle lines in the boxes represent the median values. The edges of the boxes represent the 25th and 75th percentiles, while the whiskers represent the 10th and 90th percentiles.

Exploration of Regulatory Drivers

Environmental realities drive technology adoption in the ultralow P discharge range, which is the target application for PBP/PBPep bioadsorbents. Thus, we performed an additional theoretical analysis of the impact of regulatory drivers, focused on the reductions of U.S. POTW permits to 100 μg L^–1 (Figure 4). Supporting the previous correlation analyses, the influent P concentration strongly influenced both the cost and environmental impacts, where wider P ranges translated to wider ranges of impacts. Among the U.S. EPA regions, the median cost of treatment was relatively stable for each scenario analyzed. Median GWP fluctuated slightly more than cost, which indicates its sensitivity to P inputs. Similar trends were observed for the other environmental impacts (Figures S8 and S9).

Figure 4.

a) U.S. EPA regions. In this study, only data from the 50 states were included. Regional variations in PBPep’s cost (b, d) and GWP (c, e), both normalized to mass of P removed, are shown for the theoretical reductions of existing U.S. POTW P discharge limits to an ultralow level of 100 μg L^–1. (b) and (d) were modeled using scenario 5 parameters, while (c) and (e) were modeled using scenario 6 parameters. Additional details of the model inputs are provided in the SI, Section 9. The data set numbers represent the mean (and median, also signified by the horizontal lines in the boxes). The edges of the boxes are the 25th and 75th percentiles, while the whiskers represent the 1st and 99th percentiles.

Discussion

The financial costs to remove P are consistently higher than the P recovery value (i.e., struvite-P requires a potential minimum selling price of 1.32 kg P^–153 to be economically viable, which is still higher than the diammonium phosphate rock price of 0.87 kg P^–1;^54−56,57 see the SI, Section 9 for additional discussion). However, additional incentives for P recovery emerge when accounting for not only the value of the P product but also the value of externalities, such as the environmental damages that are avoided by preventing the discharge of excess P.⁵ The environmental impacts of mining P may make a compelling case for wastewater P recovery. For example, the ecoinvent database lists the GWP of mined P for phosphate fertilizer as diammonium phosphate and fertilizer-grade phosphoric acid as 1.41 and 1.45 kg CO₂ equiv (kg P)⁻¹, respectively. These values compare favorably to our range analysis median GWP of 1.60 kg CO₂ equiv (kg P)⁻¹. An even stronger motivation comes from an examination of the eutrophication potential. Every kg of P removed from POTWs prevents 1 kg P of eutrophication potential. The eutrophication impact of preventing that kg P release (i.e., treatment to remove and recover P) is just 0.002 kg P equiv, and similar to GWP, this compares favorably to the eutrophication potential of mining 1 kg P with a eutrophication potential of 0.01 kg P equiv.

Implications of Regulatory Drivers

Given the emergence and upward trajectory of ultralow P discharge regulations (which contribute to advancing SDS targets 6.3 and 6.6, which are focused on improving water quality and protecting ecosystems), technologies capable of achieving μg L^–1-level targets are imperative. Our regulatory analysis underscores the finding that it is more cost- and environmentally intensive to incrementally reduce P at the low end of the concentration range due to diminishing returns on investment to reduce the last few μg L^–1 P (i.e., normalized to the amount of P removed, it is more expensive to reduce P from 500 to 100 μg L^–1 than it is to start at 1 mg L^–1 and reduce P to 100 μg L^–1). This is exemplified by the outlier point in region 1 (Figure 4), where the unit life cycle costs of reducing P by 1 μg L^–1 are very high, as it is not economical to build and operate a P removal system targeting this margin of performance.

The median costs for reducing existing permits to 100 μg L^–1 dropped by 1 order of magnitude, closer to levels competitive with EBPR, after reflecting additional technological improvements in the model (Figure 4b shows the results for scenario 5, while Figure 4c shows the results for scenario 6, with an increased porous:surface sorption ratio and more reuse cycles). The greatest return on investment is realized when a large number of POTWs with existing permits farther from the target of 100 μg L^–1 are reduced to ultralow levels. For example, wide-ranging regulatory reductions in a P-sensitive area such as the Great Lakes region of the U.S. would yield greater P removal, although at a higher cumulative cost and environmental impact than regions with fewer P discharge permits. Wisconsin, a state adjacent to the Great Lakes (and part of the widely regulated region 5), introduced a revised Phosphorus Rule in 2010, specifying ≤100 μg L^–1 standards for surface waters, which impacted total maximum daily loads (TMDLs) and, in turn, POTW P discharge limits. Compared to California (a state part of region 9, which has relatively few P-permitted POTWs), broad sweeping reductions of existing permits to 100 μg L^–1 would impact approximately 300× more facilities in Wisconsin, thus accruing approximately 280× more removed P (normalized to volume) with a 330× higher cumulative cost.

Contextualizing the Improvements Evaluated in This Study

The improvements evaluated in this study included increasing the number of reuse cycles and increasing the number of P-binding sites per unit volume. Both of these improvement targets, while ambitious, were based on available literature values for similar water treatment-based or protein-based technologies. Ion exchange technologies have life spans of 5–15 years in drinking water treatment facilities. The time between regeneration cycles ranges from 12 to 48 h. On the basis of these ranges, the reuse cycles range from 450 to 10 950. We limited our reuse cycles for protein-based technologies to the more conservative end of the spectrum at 400 cycles, which still represents an ambitious target for such technologies. Most studies have only evaluated a limited reuse cycle range of 10–20 (e.g., refs (30) and (58)); however, it is noteworthy that minimal sorption capacity was lost over those cycles.

The transition to peptide-based bioadsorbents is an emerging but promising field that addresses a range of environmental contaminants, including P³⁸ and heavy metals.⁵⁹ Our study, however, highlights the need to transition away from the predominant approach involving the surface attachment of protein and peptide bioadsorbents, as this approach limits the number of binding sites to the matrix support volume. Other drinking water treatment technologies consistently use higher porous:surface area sorption ratios. For example, based on a maximum P-sorption capacity of 4.1 mg P (g resin)⁻¹,⁶⁰ the LayneRT resin has a ratio of 747 (see the SI, Section 2 for the calculations). The protein-based technology analysis included a more conservative ratio of 236. While these ratios may be common for traditional drinking water treatment technologies, achieving these values for protein-based technologies requires additional research and development.

One promising example found in enzyme literature is covalently linked enzyme aggregates and crystals (CLEAs and CLECs, respectively), which couple the protein structure and the support matrix.^61,62 Our conservative calculations of the porous:surface area ratio from a study on lipases⁶³ indicated that a ratio of 14 was achieved out of a theoretical limit of 125 (see the SI, Section 2). However, this study did not fully evaluate the causes of the discrepancy between the actual and theoretical limits, including the introduction of mass transfer limitations, steric hindrance, and the inactivation of the enzymes associated with damage from the covalent linkers. In fact, the use of protein-based bioadsorbents may lend itself well to addressing these fundamental questions, as mass transfer limitations and steric hindrance can be studied independent of the kinetic limitations (i.e., enzymes consistently turn over the substrate, limiting the system’s ability to reach equilibrium). While the covalent linking of peptides to surfaces has been employed to capture rare earth elements (e.g., terbium),⁵⁸ the ability to covalently link peptides together to create a functional matrix remains a critical research gap for enabling the sustainable application of bioadsorbents for environmental systems. Overall, our analysis shows that PBP and PBPep systems may be comparable P removal and recovery technologies compared with current treatment processes, especially in scenarios with higher P concentrations. While PBP and PBPep systems can serve as a polishing step (going from 1 mg P L^–1 to 100 μg P L^–1), careful consideration of the unit costs to recover 1 kg P is recommended.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c04016.

• Description of the POTW discharges and permits and regulatory analysis, discussion on the immobilization matrices and proposed technology improvements evaluated in this study, model parameters, additional environmental impact category data, and the economic value of and the cost to recover P (PDF)

The authors declare no competing financial interest.

Special Issue

Published as part of the Environmental Science & Technologyvirtual special issue “Accelerating Environmental Research to Achieve Sustainable Development Goals”.