1. Introduction
Phosphorus (P) is a nonrenewable resource required for all plant growth [
1,
2]. About 90% of P is used in the global food chain, mainly as fertilizer, and it is estimated that P reserves will be depleted in 50 to 100 years at its current consumption rate [
3,
4,
5]. Total P (TP) can be classified as particulate P (PP) or total soluble P (TSP). TSP remains in the solution after water is filtered using a 0.45-µm filter to remove PP. PP includes living and dead plankton, P precipitates, and P adsorbed to particulate matter [
6]. Runoff containing PP often enters surface water, where it settles to the bottom of lakes and streams, making the P less available to algae [
7]. P can be further classified as inorganic or organic. Inorganic P is not bound to carbon and hydrogen and includes orthophosphates and polyphosphates. Orthophosphate, also known as soluble reactive P (SRP), is the form used by plants. Polyphosphates are strong complexing agents for metal ions commonly found in detergents and can convert into orthophosphate [
6,
8]. Organic P is bonded to plant or animal tissues and can be found in excreta and pesticides [
6,
8].
P losses from agricultural land are a loss of a valuable nutrient. Manure and fertilizers containing P are the main contributors to nonpoint source pollution, both in PP and SRP forms [
9]. Other sources include point sources, such as effluent from wastewater treatment plants or industrial facilities [
9]. In surface water, this P contributes to eutrophication, including the growth of harmful algal blooms, a higher frequency of hypoxia events, poisonous seafood, losses to aquaculture enterprises, long-term ecosystem changes, and loss of biodiversity [
10,
11], impacting tourism, human health, environmental safety, and property values [
4]. In one case, agricultural runoff partially caused the eutrophication of Lake Erie, leaving upwards of a
$100 million annual impact on Ohio’s economy [
12,
13,
14].
Much research has been conducted on PP but only limited on SRP into freshwater bodies. SRP is 95% bioavailable to algae, meaning that it is easily utilized for plant growth, increasing the risk for eutrophication [
7]. SRP originates from subsurface drainage or surface runoff [
15]. Approximately 18 to 28 million hectares (180,000 to 280,000 km
2) of cropland in the Midwest region of the USA use subsurface (tile) drainage [
16]: perforated drain tubes placed two to four feet (61 to 122 cm) below the crop land that allow water to enter the drain [
17]. There is a positive correlation between subsurface drain outlets connected to surface water and the amount of P present in those water bodies [
16,
18]. P leaching through soil into subsurface drains depends on the drainage system design, soil properties, hydrology, precipitation, season, and land-management practices, such as tillage [
16,
19].
Subsurface drainage properties can vary from field to field due to different soil types, land management practices, geology, hydrology, and climate [
20]. In this research, P adsorption media (PSM) was proposed to capture SRP from agricultural subsurface drainage. Adsorption is the transfer of solutes in a liquid phase, adsorbates, onto a solid adsorbent material, also known as media [
21]. Phosphate (PO
43−) is absorbed to positively charged ions (cations) such as iron, magnesium, calcium, and aluminum by Van der Waal interactions [
22].
Examples of PSM include those derived from natural and waste materials and manufactured nano-engineered media (NEM). Different types of PSM exhibit different performance kinetics and adsorption capacities. Natural material-based PSM include zeolite, limestone, and soils. Waste material-based PSM are by-products from other processes such as slags from metal processing plants, fly ash, and water treatment residuals. NEM are manufactured specifically to adsorb phosphorous and have a large surface area and high concentrations of positively charged ions. The base material is typically a ceramic, resin, or biochar that is then coated with metal oxide nanoparticles. These modifications increase the number of adsorption sites on the media, enhancing the PSMs’ overall adsorption capacity [
23,
24,
25,
26,
27,
28].
Table 1 shows different PSM types.
To determine the media that has the best potential to remove SRP from subsurface drainage, the literature was consulted and preliminary studies conducted [
39]. These studies entailed testing P removal in subsurface drainage using eight types of PSM: PO
4Sponge generation 1 and 2, FerrIXA33E, HIX(Zr)-Nano, steel furnace slag (SFS), blast furnace slag (BFS), calcium magnesium biochar, and ferrous sulfate biochar [
39]. The PO
4-Sponge generation 1 and 2, FerrIXA33E, and HIX(Zr)-Nano were types of NEM that were able to remove SRP below detection limits in subsurface drainage with an initial SRP concentration of 0.20 mg/L [
39]. The authors also found that the tested NEM had increased removal with increasing media amounts [
39]. The BFS and calcium magnesium biochar PSM did not remove SRP from the subsurface drainage [
39]. A quantity of 0.15 g of ferrous sulfate biochar removed 21%, 23%, 26%, and 28% after 2, 4, 6, and 24 h of contact time, respectively, at an initial SRP concentration of 0.500 mg/L [
39]. The SFS was able to achieve 7%, 12%, and 28% removal using 0.3, 0.6, and 1 g(s) of media, respectively, at an initial SRP concentration of 0.200 mg/L, showing that it was also capable of increased removal with increasing media amounts [
39]. Based on this preliminary testing, SFS and PO
4Sponge, as a representative NEM, which will be referred to as a generalized NEM (GNEM), were selected for further study because of their capacity for a given contact time, hydraulic conductivity, durability, cost, and commercially availability [
39].
SFS is a by-product of the steel industry. The slag is formed after lime is injected during the smelting process as a fluxing agent, where it chemically bonds the silicates, aluminum oxides, magnesium oxides, manganese oxides, and ferrites [
40]. SFS is then poured, cooled, and processed to remove free metallics and sized for commercial use [
40]. The removal mechanism for SFS is a reaction of calcium minerals on the SFS surface with phosphate or bicarbonate ions to produce either calcium phosphate or calcium carbonate, respectively [
41]. The optimal conditions for calcium phosphate precipitation are when the pH is 8 or above and there are high concentrations of soluble calcium ions [
41,
42]. Penn et al. (2020) recommends utilizing SFS to treat subsurface drainage for not more than four to six months due to an accumulation of calcium carbonate minerals on the SFS surface. Calcium carbonates precipitate on the slag instead of calcium phosphate when bicarbonate and dissolved forms of CO
2 are present in the subsurface drainage, which results from water infiltrating through calcareous soils and microbial respiration [
41,
42]. This decrease in capacity is due to (1) the bicarbonate and phosphate ions competing to adsorb to the calcium minerals and (2) the decrease in pH due to the formation of calcium carbonate and soluble calcium concentration, which negatively impacts the SFS’s ability to precipitate phosphate ions as calcium phosphate [
41,
42]. Specifically, the SFS has a decrease in P removal via calcium phosphate precipitation when the pH of the solution is below 8.5 [
41,
42]. Gonzalez et al. (2020) found that surface runoff was not an issue for the SFS, because it did not contain bicarbonate concentrations causing calcium carbonate mineralization [
41].
The 2020 cost of SFS is
$0.03/kg [
43]. Steel slag fines with a particle diameter of 0.075 mm have a hydraulic conductivity of 6.12 × 10
−3 cm/s [
44]. The specific gravity of SFS ranges from 3.2 to 3.6 [
45]. Blanco et al. (2016) used SFS in a batch adsorption study with an initial P concentration of 5-mg P/L and achieved an adsorption capacity of 0.12 to 1.20-mg P/g media [
46]. Sheng-gao et al. (2008) conducted a batch adsorption study with SFS using an initial concentration of 1000-mg P/L and achieved an adsorption capacity of 33.3-mg P/g media [
47]. Both Sheng-gao et al. (2008) and Blanco et al. (2016) noted that increasing initial P concentrations increased the adsorption capacity of the media [
46,
47].
The GNEM is a proprietary NEM composed of iron oxide nanocrystals of oxyhydroxide with an alumino-silicate-bonded porous structure containing 80% interconnected pores, a hydraulic conductivity between 3 to 7 cm/s, a base surface area of 15 m
2/g, a surface area over 70 m
2/g with nano-modification, and a density of approximately 0.53 g/cm
3 [
22,
26,
48,
49]. Important to note is that the granular form of the GNEM used in this research is only produced for small-scale or laboratory purposes [
50]. The manufacturer recommends a monolith of GNEM shaped into cubes, discs, or other shapes for full-scale processes [
51]. The GNEM has an adsorption capacity ranging from 25-mg P/g media for low P concentrations (<2 mg/L) and 80-mg P/g media for high P concentrations (>5 mg/L) [
30,
51]. However, the capacity for very low P concentrations (<0.5 mg/L) is about 10 mg/L [
30]. Competing ions found in subsurface drainage are not believed to be a concern for the GNEM [
22,
50]. P removal is achieved at concentrations from 0.1 mg/L seen in agricultural drainage to 150 mg/L seen in industrial wastewater at food-processing plants [
48]. The GNEM can remove P down to levels below 0.09 mg/L for lakes, streams, and agricultural water runoff [
30]. Safferman et al. (2015) tested the media using effluent from multiple wastewater treatment plants and found that it reduced SP levels from 1 mg/L to less than 0.3 mg/L [
49].
The 2020 cost of the GNEM is
$19.00/kg [
26]. According to the manufacturer, the GNEM can be regenerated 15 to 20 times, and the regeneration process lowers the average media cost by 80% when compared to the cost of a single, nonregenerated use of this media [
48]. Regeneration is the process of stripping ions off the adsorption media using a pH of 10 or higher [
21]. P is also easily recovered from the caustic solution as a calcium phosphate precipitate after the regeneration process. For example, Sengupta and Pandit (2011) used sodium chloride and sodium hydroxide to remove phosphate ions off a hydrated ferric oxide (HFO) adsorption media, then calcium or magnesium salt to precipitate phosphate out as a solid-phase fertilizer by-product [
52].
Based on the literature review, there is a lack of data and economic considerations on the use of PSM to remove P from subsurface drainage. The objective of this research is to determine the PSM option best suited for managing and removing SRP from a subsurface drainage and gain an understanding of economic factors. This was achieved through the following tasks: (1) characterizing real subsurface drainage water (RSDW) to enable the formulation of synthetic subsurface drainage water (SSDW) for laboratory use studies, (2) running column experiments on selected media to estimate the SRP removal and media capacity under different conditions, and (3) conducting an economic analysis for a study site as a design example and to determine the important implementation consideration.
4. Conclusions
GNEM removed SRP to very low levels, below the 50-µg/L analytical detection limit, at an EBCT as low as 5 min. The resulting GNEM capacity at these low SRP levels and rapid EBCT was 1.2-mg SRP per g of media. The SFS had a similar performance but only had a capacity of 0.17 mg of SRP per g of media. Other types of media tested did not successfully remove the SRP to the very low target level.
Using the SFS to remove P and then disposing it after use was the most cost-effective option based on a specific study site. Using the GNEM media to remove P and then regenerating it onsite was slightly more expensive than the SFS, with a percent relative difference of 26%. These costs greatly depend on the many site-specific assumptions and are not transferable, but the methodology can be applied to other locations. Additionally, this was a theoretical analysis, as the media was not actually deployed at the study site. However, shipping the GNEM back to a centralized location was comparably very expensive, having a percent relative difference between 33% and 53%, and will likely not be effective for most situations. Further, media costs and the value of the recovered phosphorus is also expected to vary. As GNEM becomes more common, new products will increase competition and manufacturing and performance optimization will likely lower costs. Conversely, as the use of SFS increases, costs may increase. However, if disposal of the SFS on-site is feasible or it can be used as a beneficial soil amendment and SRP recovered, costs for disposal will be eliminated.
Although implementation costs were found to be equal or greater than crop revenues for the study site case study, there are opportunities for efficiencies, and governmental conservation incentives may be possible. Further, a more optimal design approach should be explored. The very simplistic approach used for the study site results in SRP concentrations always below the breakthrough level of 50% of the influent concentration. This results because the equilibrium mechanisms of the PSM result in greater SRP removal when there is less SRP adsorbed to the media and less removal when more is adsorbed. A system where two columns of PSM are used would reduce the overall required amount of media. The first column would be a roughing system, and the second would lower the SRP to detection limits. Once SRP levels start to increase, this second column would be switched to the first position and serve as a roughing system, and new SFS or regenerated GNEM would be used in the polishing (second) column. This SRP-free subsurface drainage water would then be blended with equal amounts of untreated subsurface water, which would theoretically maintain the blended effluent to below the breakthrough target of 50% for the case study. However, this would be at the expense of more complex operations, which may negate any benefits.