Using selectivity to evaluate aqueous- and resin-phase denitrification during biological ion exchange

Nitrate (⁠⁠) is a contaminant that contributes to eutrophication of surface water and groundwater and has adverse human health effects in drinking water (Ward et al. 2018). is effectively removed from water by ion exchange (IEX), but regeneration of these resins with concentrated salt brines requires onsite storage of concentrated chemicals and disposal of brine waste. This regeneration requirement is a significant barrier to the implementation of IEX for decentralized contamination that accounts for a majority of the loading to surface waters globally (Carpenter et al. 1998; Howarth et al. 2000; Howarth 2008). Biological ion exchange (BIEX) is a type of IEX where resins are regenerated biologically using a biofilm on the surface of the resins and has been suggested as an alternative to overcome barriers associated with chemical regeneration for application to decentralized treatment (Archna et al. 2012; Huno et al. 2018).

BIEX has been implemented in continuous-flow systems, where water flows into and out of the IEX reactor at a constant rate, for the removal of natural organic matter (NOM) from surface waters (Schulz et al. 2017; Amini et al. 2018; Winter et al. 2018; Edgar & Boyer 2021), resulting in a 100% reduction rate in brine waste and achieving as much as 60% continuous DOC removal. BIEX for removal, however, has been limited to batch configurations. For example, when laden brines (McAdam & Judd 2008) or resins (Meng et al. 2014; Ye et al. 2019) were contacted with microbial cultures for >4 h, 95–100% brine waste reduction and complete regeneration of resins were achieved. Batch BIEX experiments have relied on the desorption of adsorbed to the resin surface into the aqueous phase, followed by the microbial denitrification of aqueous-phase (Meng et al. 2014; Ye et al. 2019). Denitrification of adsorbed on the surface of IEX resins has not been experimentally evaluated. It is currently unknown if the denitrifying microorganisms can reduce the adsorbed to the IEX resin.

The desorption of adsorbed constituents from IEX resins by other constituents has been coined secondary ion exchange (SIEX) and is a key mechanism in both continuous and batch BIEX experiments (Schulz et al. 2017; Liu et al. 2020, 2022; Edgar & Boyer 2021, 2022). In a continuous-flow BIEX system, the desorption of by SIEX is undesirable because in the aqueous phase will flow out of the reactor before microbial denitrification can occur. This is because the timeframes of microbial denitrification are in the order of hours to days (Tiedje 1983; Archna et al. 2012; Huno et al. 2018), while most IEX resin columns have retention times of approximately 5 min. Previous work has identified the common SIEX constituents in different microbial cultures (wastewater sludge, lab-cultured microorganisms, and natural microorganisms from wood mulch) and quantified their affinities for IEX using separation factors (Edgar & Boyer 2022). The separation factors calculated in Edgar & Boyer (2022) show that buffer and salt solutions exist that could be used to culture microorganisms while minimizing SIEX, promoting biological denitrification on the resin surface rather than the aqueous phase as would be desired in a continuous-flow system.

The goal of this study was to determine if resin-phase microbial denitrification is possible and identify potential configurations for continuous-flow BIEX implementation. The specific objectives of this work were to (1) identify a microbial mineral medium with low IEX selectivity to minimize secondary IEX with ⁠, (2) achieve aqueous-phase microbial denitrification in a low IEX selectivity microbial mineral medium, and (3) determine if microbial denitrification is possible using adsorbed to IEX resins (i.e., resin-phase denitrification). These objectives were accomplished using batch denitrification experiments in solutions with low IEX selectivity as identified in Edgar & Boyer (2022).

A stock salt solution and trace element solutions (A and B) were prepared as described in Robles et al. (2021), where the tabulated composition of the stock solutions can be found (Robles et al. 2021). The stock salt solution was used to make a reduced anaerobic mineral medium and test media (modified the reduced anaerobic mineral medium with a low affinity for IEX). The reduced anaerobic mineral medium was prepared in a 2 L bulb flask filled with 1 L de-ionized (DI) water, 10 mL salt stock solution, 1 mL trace element solution A, and 1 mL trace element solution B. About 0.25 mL of 0.1% resazurin was added to ensure anoxic conditions in the mineral medium. The medium was prepared according to the Hungate method and was reduced with N₂ gas. The medium was buffered with 15 mM HEPES and pH was adjusted to 7.5 with 4 M NaOH. The mineral medium was reduced by adding 0.4 mM L-cysteine and 0.2 mM Na₂S × 9H₂O. While sparging, 100 mL aliquots of growth media were transferred from the flask into 10 separate 120 mL serum bottles and then immediately capped with rubber stoppers and aluminum crimp caps. The bottles were then autoclaved and ready to be used for the growth of denitrifying bacteria.

Two liters of test media were prepared similarly to the reduced anaerobic mineral medium via the Hungate method with a few changes that minimized salt concentrations. No stock salt solution, resazurin, L-cysteine, or Na₂S × 9H₂O was added. Furthermore, the pH was adjusted to 6.7 (instead of 7.5) using 4 M NaOH. This modified reduced anaerobic mineral medium was then ready to be used as a low IEX selectivity microbial growth medium in the batch denitrification experiments.

The resin used in this experiment was A520E strong base anion exchange resin (Purolite). A520E is a gel, macroporous chloride-form resin with a polystyrene base. The resin has quaternary ammonium functional groups and a capacity of 0.9 eq/L, and was chosen for its high selectivity. The chloride-form resin was converted into the form by mixing 10 g of resin (dry weight) in a 150 mL concentrated KNO₃ solution (50× resin capacity) at 100 rpm for 24 h. The resin was then rinsed with DI water until the rinse water conductivity was <30 μS/cm to remove any remaining non-adsorbed KNO₃. The initial solution and the solution after 24 h of mixing with IEX resin were analyzed to determine the amount of adsorbed ⁠.

In an anaerobic glove chamber, two bottles of the reduced anaerobic mineral medium were spiked with 20 mM ⁠, 35 mM acetate, and 1 g/L yeast extract. Then, 600 μL anaerobic digester sludge from Mesa Northwest Water Reclamation Facility in Mesa, Arizona, was added as inoculum to each bottle. The inoculum collected was identified as rich in denitrifying bacteria by plant operators due to rapid denitrification observed in the anaerobic digester. The bottles were incubated at 32 °C and shaken on a shaker table at 100 rpm. A 1 mL sample was taken every 48 h and analyzed for ⁠. If was less than 5 mM, the culture was respiked with 20 mM ⁠. This process continued for 2 weeks.

After 2 weeks, six more bottles were spiked with 20 mM ⁠, 35 mM acetate, and 1 g/L yeast extract. The inoculum consisted of a 5 mL culture from the previously grown denitrifying culture cultivated for 2 weeks. These six bottles were incubated on a shaker table at 100 rpm and 32 °C and respiked with ⁠, acetate, and yeast extract identically to the previous bottles until the culture was sufficiently grown (visible flocs achieving complete reduction within 24 h of a spike).

Culture samples were collected from the microbial growth bottles and then concentrated and washed as follows. Bottles were transferred to an anaerobic chamber, and the contents were divided into 50 mL centrifuge tubes and centrifuged using an Eppendorf microcentrifuge 5415R (Hauppage, NY) at 13,200 rpm for 15 min. The supernatant was discarded, and pellets from individual bottles were combined and resuspended into the mineral medium. The tubes were then centrifuged again, the supernatant was discarded, and the pellets were resuspended two additional times. This process ultimately resulted in a single 50 mL centrifuge tube containing the cultured bacteria that had been sufficiently rinsed for the batch denitrification experiment.

The fresh reduced anaerobic mineral medium was prepared at the start of the batch denitrification experiments. The following six experimental conditions used are presented in Table 1: condition 1 provided a control for the test water; Condition 2 provided a control for microorganisms in test water without ⁠; Condition 3 was a control for microorganisms in test water with aqueous ⁠; Condition 4 was a control for the interactions between test water and resin; Condition 5 was the test condition containing test water, resin, and microorganisms; and Condition 6 was a control for resin in DI water. Each condition was tested in triplicate resulting in 18 total 120 mL glass serum bottles.

Table 1

Experimental conditions tested in triplicate in this study

Condition .	Medium .	Resin .	Microbes .	Acetate .
1 (Test water only)	Test water	No	No	Yes
2 (Inoculum only)	Test water	No	Yes	Yes
3 (Aqueous ⁠)	Test water + aqueous	No	Yes	Yes
4 (Test water and resin)	Test water	Yes	No	Yes
5 (Resin and inoculum)	Test water	Yes	Yes	Yes
6 (Resin and DI)	DI water	Yes	No	No

Condition .	Medium .	Resin .	Microbes .	Acetate .
1 (Test water only)	Test water	No	No	Yes
2 (Inoculum only)	Test water	No	Yes	Yes
3 (Aqueous ⁠)	Test water + aqueous	No	Yes	Yes
4 (Test water and resin)	Test water	Yes	No	Yes
5 (Resin and inoculum)	Test water	Yes	Yes	Yes
6 (Resin and DI)	DI water	Yes	No	No

The text in parentheses next to each test condition number represents how each condition is referenced hereafter.

View Large

Ten milligrams of resin-adsorbed was added to each resin bottle, which is equivalent to 0.18 g of dry saturated resin (calculated during resin saturation). The appropriate medium was then added to each bottle, i.e., test water without ⁠, test water with ⁠, or DI water (Table 1). Next, the concentrated microbial culture was divided between the nine bottles for Conditions 2, 3, and 5 (5.5 mL concentrated culture). Test Condition 3 received 0.32 mL of 500 mM solution (equating to 10 mg of ⁠, identical to resin conditions). Conditions 1–5 received 0.28 mL of 1 M acetate solution, equivalent to adding 17.5 mg of acetate to each bottle. The bottles were incubated in a shaker table at 100 rpm and 32 °C. Samples were collected at t = 0 h, t = 1 h, 2 h, 1 day, 2 days, 3 days, 7 days, and 14 days. Filtered samples were analyzed for inorganic ions and acetate. A horizontally fixed frictionless syringe (borosilicate glass) was used to measure gas generation in bottles prior to collecting fluid samples. The syringe plunger was allowed to expand freely until stopping, at which point the gas volume was recorded and the plunger was fully depressed before the syringe was disconnected from the bottle. After 7 days, one bottle from each test condition had 2 mL of gas removed for gas chromatography analysis.

After day 14, samples were collected, and the contents of all bottles were centrifuged, rinsed in DI water, re-centrifuged, and the resin and microorganisms were recovered. Sonication was performed for 5 min to detach microorganisms from the resin in Condition 5. The recovered resin and microorganisms were dried and weighed. Resins were regenerated by mixing in 100 mL of 3,500 mg/L Na₂SO₄ solution (⁠ equivalent to 3× the capacity of each resin sample) for 24 h. Initial and final regenerant samples were collected and analyzed for inorganic ions and organic acids.

A batch adsorption equilibrium experiment was conducted to determine the final active site composition of resin and test water mixtures with various concentrations of ⁠, ⁠, and acetate. Four conditions were used for the experiment, each completed in triplicate 120 mL serum bottles as follows: (1) DI and resin control, (2) 100 mg/L (equivalent concentration to adsorbed ⁠), (3) 100 mg/L acetate (equivalent concentration to adsorbed ⁠), and (4) 20 mg/L + 175 mg/L acetate (identical to the final amount of and acetate present in Condition 5 from the denitrification experiment). About 10 mg of resin-adsorbed was added to each bottle and mixed on a shaker table for 24 h, then resins were regenerated as in the denitrification experiment. Samples were collected before adding resin, after 24 h of mixing, and after regeneration. All samples were analyzed for acetate, ⁠, ⁠, and Cl⁻.

The pH and conductivity measurements were completed using an Orion Dual Star Multiparameter meter, an Orion 9156BNWP (Thermo Fisher Scientific, Waltham, MA, USA) combination pH probe, and an Orion Star A212 conductivity probe. Concentrations of ⁠, ⁠, sulfate (⁠⁠), nitrite (⁠⁠), and chloride (Cl⁻) anions and sodium (Na⁺), ammonium (NH₄⁺), potassium (K⁺), calcium (Ca²⁺), and magnesium (Mg²⁺) cations in the aqueous phase were measured using ion chromatography (Dionex ICS 5000 + , Sunnyvale, CA, USA). Anion chromatography was conducted using a Dionex AS-18 column with KOH eluent, and cation chromatography was conducted using a Dionex CS-15 column with methanesulfonic acid eluent. Both cation and anion chromatography utilized a 0.1–100 ppm calibration curve for all measured ions. Organic acids including acetate were measured using high-performance liquid chromatography (HPLC; Shimadzu LC-20AT) equipped with an Aminex HPX-87H column (Bio-Rad Laboratories, Hercules, CA, USA) as detailed previously (Joshi et al. 2021). Gaseous N₂, dihydrogen (H₂), nitric oxide (NO), nitrous oxide (N₂O), and carbon dioxide (CO₂) were measured using a Shimadzu GC-2010 Pro gas chromatograph.

After using the aqueous control condition to determine a baseline for biological transformation, aqueous ion concentrations are observed in the test conditions to determine if the experimental conditions successfully minimized the influences of SIEX. The resin and inoculum condition showed constant low concentrations of chloride, phosphate, and sulfate. Because the concentrations of these inorganic ions remained low and unchanged throughout the experiment, SIEX involving these ions is believed to have little to no influence. There was some initial desorption over the first 2 days due to acetate adsorption, resulting in an aqueous concentration of 14 mg/L that began decreasing after 3 days and dropped to 5.3 mg/L after 14 days. As in the aqueous condition, the disappearance of aqueous and the generation of occur concurrently and signify biological transformation. At the end of the experiment, the amount of aqueous and accounted for approximately 15 and 6%, respectively, of the total N initially added to the system as adsorbed ⁠. Therefore, approximately 21% of the total N was in the aqueous phase at the end of the experiment.

The aqueous- and gas-phase analyses presented in this section help to identify the biological transformations taking place and confirm that the experimental conditions used were appropriate for both the stimulation of biological transformation and the inhibition of SIEX. Both aqueous and gas data confirm biological activity in the expected bottles, but the data are also indicative of incomplete denitrification that stops at ⁠. Since this observation holds true for the aqueous control bottles, it is apparent that the low selectivity solution used is the likely cause of incomplete denitrification rather than the lack of availability of adsorbed N species. The aqueous ion trends observed in test conditions were indicative of limited SIEX by inorganic ions, but as much as 21% of total N may have been desorbed by acetate.

The resin and inoculum condition showed less adsorbed acetate (7.7% of active sites) and much less adsorbed (53% of active sites) at the end of the experiment compared to the resin and test water condition without inoculum. Most notably, 25% of active sites were occupied by ⁠. This resin-phase composition highlights the possibility of resin-phase biotransformation of ⁠, since a large fraction of has been converted to ⁠, while only a small fraction was observed to have been desorbed by acetate. Additionally, the amount of adsorbed acetate highlights the possibility that adsorbed acetate was also used by microorganisms. Because the initial desorption in resin and inoculum conditions was nearly identical to resin and test water conditions (approximately 0.023 meq desorbed ⁠), one would have expected to find similar amounts of adsorbed acetate in the resin and inoculum conditions, but the observed amount of adsorbed acetate was much less (7.7% in resin and inoculum conditions versus 15% in resin and test water conditions). The aqueous- and resin-phase ion compositions observed in the resin and inoculum condition are strong indicators for resin-phase biological denitrification.

The results presented in Figures 2–4 clearly show biological activity under the provided conditions, and that SIEX only occurred with <15–18% of adsorbed due to desorption by acetate. These results provide a positive outlook for denitrifying BIEX implementation in a continuous-flow system due to the clear stepwise transformation observed for adsorbed ⁠. The potential utilization of adsorbed acetate by microorganisms is an added benefit, as well as the adsorption of ⁠, which will need to be further transformed for complete denitrification. Although these conditions are favorable for implementation in a continuous-flow system, further investigation is still required to determine if biological transformations actually occur on the resin surface or if the ultimate resin-phase composition observed is due to continuous adsorption and desorption of acetate, ⁠, and as the aqueous- and resin-phase concentrations are driven toward an equilibrium state that is continuously fluctuating due to biological transformations. To further address this theory, an adsorption equilibrium experiment was completed.

Condition 2, which started with equivalent (meq:meq) concentrations of aqueous and adsorbed ⁠, resulted in an average of 39% desorption. Condition 3, which started with equivalent concentrations of aqueous acetate and adsorbed ⁠, resulted in 11.3% of desorbed ⁠. This is consistent with literature that has suggested IEX selectivity follows the order of > > acetate (Marina et al. 1997; Pohl et al. 1997; Li & Yang 2015; Edgar & Boyer 2022). Condition 4, which mimicked the test conditions in the denitrification experiment, resulted in active site compositions of 17.1% of and 8.8% of acetate, which is very similar to the ultimate active site composition observed in the denitrification experiment, which was 24.9% of and 7.7% of acetate. The similar active site composition in both experiments indicates that adsorption equilibria are the major driving force for desorption in both experiments. However, the higher amount of desorbed in the denitrification experiments indicates that the presence of denitrifying bacteria encourages further desorption by decreasing aqueous concentrations and increasing aqueous concentrations.

Though these results indicate that resin-phase biotransformation was unlikely, this gradual desorption still has positive implications for a continuous-flow denitrifying BIEX system, since the small amounts of released from the resin over time could potentially be denitrified in the short retention times used in IEX treatment systems, particularly if desorbed could stay present in the extracellular matrix of the biofilm during diffusion to and from active sites through the biofilm (de Beer & Stoodley 1995; Horn & Morgenroth 2006). accumulation is a common barrier to achieving denitrification under a range of conditions (Glass & Silverstein 1998; Cao et al. 2013; Yuan et al. 2013; Rocher et al. 2015), and the accumulation observed in this work may be due to pH limitations, inappropriate C/N ratio, or a lack of available nutrients, vitamins, or other trace elements. pH is a likely contributor as all bottles had a pH value from 6.3 to 6.5 at the end of the denitrification experiment. The next step toward a continuous-flow denitrifying BIEX system should be flow-through experiments that determine if washout occurs at a variety of flow rates, water compositions, resin types, and at various stages of biofilm growth.

This work has identified a low selectivity solution, containing constituents with low IEX affinity, and IEX resin pair that are amenable to achieving biotransformation of in a batch solution with approximately 47% of desorption, showing promise for BIEX implementation in a continuous-flow denitrification system. Denitrifying bacteria were shown to transform desorbed  to ⁠, driving adsorption equilibria toward additional desorption. The gradual desorption facilitated by this biotransformation presents an opportunity for a continuous-flow denitrifying BIEX configuration where desorbed can be denitrified in the short retention times used for IEX systems. The success of a continuous-flow configuration will still be dependent on overcoming the barrier of accumulation, which means maintaining the appropriate pH conditions, vitamins, minerals, and trace elements required for denitrifying bacteria while still minimizing SIEX. The successful biotransformation achieved under these test conditions has enhanced the understanding of BIEX and specifically indicated the remaining gaps in knowledge that must be addressed for its implementation in a continuous-flow configuration.

The research described herein was supported by the National Science Foundation Engineering Center for Bio-mediated and Bio-inspired Geotechnics (CBBG) (grant no. ERC-1449501). Any opinions or positions expressed in this article are the authors only and do not reflect any opinions or positions of the NSF.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.