zero-valent iron is an effective and abundant reducing agent with a standard redox potential of −0.44 V for NOx− removal from groundwater and wastewater. According to the particle size, ZVI can be classified as nanoscale ZVI (nZVI) and microscale ZVI (mZVI), and nZVI generally has a higher reductive capacity than mZVI due to its greater specific surface area. The mechanisms of the ZVI reaction with NOx− mainly involve (1) direct electron transfer from ZVI to NO3− to form lower-valence nitrogen species, as shown in Eq. 1–3 (Jeong et al., 2012) in Table 1; (2) indirect reduction of NO2− by the produced Fe(II), as in Eq. 4–5 (Liu and Wang, 2019); and (3) removal of NO3− via the H₂ secondarily generated by hydrogenotrophic nitrate-respiring bacteria, as in Eq. 6–8 (Kim and Cha, 2021). In addition, NO3− can be transformed to NH4+ by adding both ZVI and Fe(II) with strong reducibility (Eq. 9). Fe(II)EDTA generally has a high adsorption capacity for NO, and the ZVI added in the reactor could convert Fe(II)EDTA-NO to NH4+, providing substrates for Anammox and decreasing the toxicity of NO in activated sludge (Eq. 10; Zhang et al., 2019a).

Although ZVI has been adopted to remediate nitrate-containing wastewater pollution, the shell of the iron oxides formed on the surface of ZVI and the agglomeration of ZVI particles significantly affect the nitrate reduction efficiency. Hence, physical and chemical strategies have been developed to enhance the transformation efficiency and iron corrosion. ZVI composites are prepared to increase the adsorption capacity for NO3− as well as to increase the number of active sites for electron transfer to NO3−, mainly through (1) doping ZVI with metal or inorganic components such as Cu, Pt, Al and activated carbon (AC); (2) supporting ZVI composites with matrixes such as calcium alginate and AC; and (3) adding reducing-state additives such as sulfur to ZVI. Meng et al. (2020) encapsulated mZVI and AC into porous calcium alginate to achieve good dispersion of mZVI particles and enhance the iron-carbon galvanic cell effect. An acid mine drainage-based nZVI was synthesized to couple NO3− reduction and norfloxacin oxidation with ultrasound irradiation to overcome passivation (Diao et al., 2019).

In early 1966, Chao and Kroontje (1966) studied the feasibility and intermediates of inorganic NO3− transformation through Fe(II) oxidation. Only under acidic and high-temperature conditions, can Fe(II) be chemically oxidized by NO3−. Even Fe(II) can theoretically react with NO3− in a thermodynamically favored reaction at near-neutral pH and environmental temperature, but the kinetic rate is extremely low. Although the redox potential of the Fe(III)/Fe(II) couple (−314–14 mV) is lower than that of all nitrogen couples (NO3−/NO2−, +430 mV; NO₂⁻/NO, +350 mV; NO/N₂O, +1,180 mV; N₂O/N₂, +1,350 mV) in denitrification, Fe(II) oxidation coupled with NO3− reduction cannot proceed smoothly without a catalyst. The common catalysts are Cu²⁺, iron (hydr)oxides and even cell surface enzymes. For example, under anoxic conditions in soils and sediments, the dissimilatory reduction of nitrate to ammonia has been observed in the presence of green rust compounds [Fe^II₄Fe^III₂(OH)₁₂SO₄•yH₂O; Hansen et al., 1996], representing the occurrence of chemical denitrification or chemodenitrification. The unique structure of green rust may self-catalyze this reaction.

In addition, Fe(II)-EDTA could promote the adoption of aqueous NO and provide a substrate for the production of N₂ by Anammox bacteria (Eq. 11–12). N₂O is a natural product primarily derived from biotic denitrification, but chemodenitrification processes also contribute to N₂O release to the atmosphere. Among them, NO2− can be reduced by Fe(II) to produce N₂O along with Fe(III; hydr)oxides as byproducts (Eq. 13–14; Jones et al., 2015). In the intermediate step of nitritation, the produced HNO₂ could be chemically reduced by Fe(II) to release N₂O, as shown in Eq. 15 (Yang et al., 2018). Nitrite can rapidly be absorbed on goethite at low pH, and surface Fe(II)-goethite complexes show variable reactivity with NO2− to produce N₂O (Dhakal et al., 2021).

Hydroxylamine (NH₂OH) is an intermediate in short-cut nitrification and can abiotically react with Fe(III) to produce nitrous oxide (N₂O), as shown in Eq. 19, and this process is greatly affected by pH. Generally, abiotic reactions play an essential role in the production of total N₂O at pH values less than 5. To prevent excessive greenhouse gas emissions, short-cut nitrification reactors usually control pH at near-neutral conditions to reduce the release of N₂O from abiotic reactions (Hikino et al., 2021).

IOM are widely distributed in the bacterial and archaeal domains (as shown in Table 2). Moreover, IOM are mainly classified into Proteobacteria, Actinobacteria (Zhang et al., 2016a), Firmicutes (Li and Pan, 2019) and Euryarchaeota (Hafenbradl et al., 1996). In 1991, Brons et al. (1991) reported that the nitrate-reducing bacteria Escherichia coli reduced NO3− to NO and N₂O in the presence of Fe(II) and L-lactic acid under anaerobic conditions. In 1996, Straub et al. (1996) enriched and isolated three strains of gram-negative bacteria, which can use Fe(II) as an electron donor for NO3− reduction. Subsequently, NDAFO IOM were found in river and lake sediments, paddy soils, water treatment reactors and constructed wetlands (CWs) and were mainly categorized as α-Proteobacteria (Kumaraswamy et al., 2006; Sorokina et al., 2012; Jiang et al., 2015; Park et al., 2018), β-Proteobacteria (Anirban and Flynn, 2013; Zhang et al., 2016b, 2019c, 2020a; Liu et al., 2019b; Chen et al., 2020a; Ma et al., 2020; Mai et al., 2021; Xu et al., 2021), γ-Proteobacteria (Etique et al., 2014; Li et al., 2015, 2018a; He et al., 2017) and δ-Proteobacteria (Su et al., 2020), which play important roles in Fe(II) reduction coupled with NOx− oxidation.

Nitrate/nitrite-dependent IOM can be divided into autotrophic and mixotrophic microorganisms in terms of metabolic types. Autotrophic IOM do not require organic carbon and use only Fe(II) as an electron donor to generate energy and fix carbon dioxide. However, such bacteria are rarely reported or distributed in wild environments. In contrast, mixotrophic IOM require organic carbon as a common reducing substrate with Fe(II). At present, it has been observed that 90% of nitrate-reducing bacteria can oxidize Fe(II) in the presence of organic matter. In other words, many heterotrophic denitrifying bacteria are also IOM in nature and are mostly heterotrophic or mixotrophic. These facts indicate great possibilities for the application of NDAFO with abundant and readily available functional microorganisms sources.

Currently, the precise electron transfer mechanism of enzyme-catalyzed NDAFO is not well understood. According to existing studies, electrons supplied by Fe(II) are transferred to NO3− and intermediate nitrogen species during denitrification by iron oxidoreductase, the quinone pool, NAR, cytochrome bc₁ complex, etc., coupled with the conventional heterotrophic electron transfer chain (Figure 1). Although no specific enzyme has been identified to accept an electron from extracellular Fe(II), it has been reported that cytochrome c (c-Cyts) is likely involved in the electron transfer between Fe(II) and IOM; however, more direct evidence is needed for verification, and other enzyme proteins need to be evaluated (Liu et al., 2019a). In addition, Fe(II) can penetrate the outer membrane into the periplasm through porin-containing proteases and chemically react with NO2− to produce gaseous NO, N₂O and N₂.

Feammox plays an important role in nitrogen cycling globally, especially in Fe(III)-rich soils or wetlands. IRM are the dominant microorganisms in the Feammox process, which oxidizes NH4+ to NO2−, NO3− and N₂ to produce bioenergy for cell growth (Eq. 20–22; Tan et al., 2021). In 1993, Lovley et al. (1993) first reported that Geobacter metallireducens could use Fe(III) as the sole electron acceptor to oxidize short-chain fatty acids, alcohols and monoaromatic compounds, which proved the existence of IRM. In 2005, Clément et al. (2005) accidentally detected the production of NO2− and Fe(II) in riparian wetland soils for the first time and observed dissimilar iron reduction by ammonium oxidation under anaerobic conditions. In 2006, Sawayama (2006) observed that NH4+ was oxidized to NO2− in an anaerobic fixed-bed reactor supplemented with Fe(III)-EDTA and named the process “Feammox” (ferric ammonium oxidation). Exiguobacterium sp. was found to be the dominant IRM from high-throughput sequencing targeting the 16S rRNA gene. Subsequently, many different IRM have been observed to participate in the Feammox process, including Proteobacteria (Benaiges-Fernandez et al., 2019; Li et al., 2019; Ding et al., 2020a,b; Wang et al., 2021; Yang et al., 2021a), Actinobacteria (Shuai and Jaffé, 2019; Ma et al., 2021) and Firmicutes (Sawayama, 2006; Qin et al., 2019; Yang et al., 2019; Yao et al., 2020; Table 3). Among them, Geobacteraceae sp. and Shewanella sp. have been widely reported and studied. Although 16S rRNA high-through sequencing is promising to identify more IRM in recent studies, pure strain isolation in enrichment cultures is still needed to explore the mechanisms underlying Feammox.

Many IRM rely on a series of redox-sensitive proteins or multiple heme cytochromes to transfer extracellular electrons by direct contact with ion minerals, which form an extracellular electron transfer (EET) pathway that binds the cell’s internal respiratory chain to external solid Fe(III) minerals. Shewanella sp. and Geobacter sp. often use c-Cyts to transfer electrons, exploiting the many solvent-exposed hemes as electron transfer centers, as heme comprises iron atoms and porphyrin rings.

Shewanella oneidensis has 39 genes that encode c-Cyts, which are considered to be electron transport mediators, including CymA in the cytoplasmic membrane, Fcc3 and STC in the periplasm, and MtrCBA in the outer membrane (Figure 2A). CymA is a dehydrogenase that oxidizes quinols to release electrons and then transfers electrons directly or indirectly via Fcc3 and STC to MtrA. For the MtrCBA structure, MtrA is inserted into the porin-like protein MtrB and then interconnects with MtrC in the outer membrane, which finally forms a ternary trans-outer membrane complex. In addition, Shewanella sp. can also secrete small molecules, such as flavin and quinones, as electron shuttles (ESs) to achieve long-distance EET (Shi et al., 2016).

Regarding Geobacter sp., G. sulfurreducens has 111 genes encoding these c-Cyts, which can be divided into three categories (Figure 2B): (1) ImcH and CbcL first oxidize quinols to release electrons in the cytoplasmic membrane; (2) PpcA and PpcD receive and transfer electrons in the periplasm; and (3) Omas (B and C) and Omcs (B, C or Z) form trans-outer membrane protein complexes, which combine with porin-like proteins OmbB and OmbC to eventually transfer electrons to the extracellular space (Liu et al., 2018). Overall, electrons are derived from quinols in the cell’s inner membrane through periplasm transport and outer membrane emission to the final electron acceptor outside the cell.

The reduction of NO3− to NO2− is the first step of microbial denitrification, which is catalyzed by membrane-bound nitrate reductase (NAR) or nitrate reductase in the periplasm (NAP). Depending on environmental conditions, such as the temperature, concentration of organics, pH, and aeration time, NO2− can accumulate to mM concentrations in the partial nitrification process. Microbially produced NO2− can be reduced by Fe(II) to gaseous NO, N₂O, and N₂ via chemodenitrification. In anoxic iron-containing environments, this process results from integrating biotic and abiotic processes, which jointly promote total nitrogen (TN) removal in the aqueous phase. This coupling provides a solution to the problem where Fe(II) does not react with NO3− in certain environments. However, accurate contribution ratios for the biotic and abiotic pathways should be quantitatively determined for an in-depth understanding of the denitrification process.

Some Fe(III) minerals, as electron acceptors, could biotically oxidize NH4+ in IRM-mediated Fe(III) reduction, and the Fe(II; hydr)oxides produced in the above process might be utilized as electron donors and react with NOx− in anaerobic conditions (Figure 3). The biotic enzymatic Fe(III) reduction and Fe(II) oxidation rates are generally higher than those of abiotic chemical reactions in nitrogen-rich conditions (Lovley, 1997). This integration has important implications for microbial ecology since specific bacteria can gain chemical energy for growth from the two energy generation processes. Although the IRM and IOM might be spatially or temporally separated, a syntrophic relationship is formed in which iron-bearing minerals mutually support the growth of each group (Zhao et al., 2015). Yang et al. (2020) developed an Anammox-like process to treat sludge digest effluent with high-concentration NH4+ using NO3− as a terminal electron acceptor in the Fe(III)/Fe(II) cycle, in which NH4+ was mainly oxidized by Fe(III) to N₂ through Feammox. The produced Fe(II) triggered NDAFO to reduce NOx− to achieve effective TN removal. The coupled processes provide a safe and efficient method to treat NH4+ and NO3− wastewater simultaneously.

In the Fe(II)-driven denitrification system, biogenic Fe(III; hydr)oxides largely depend on different inoculated denitrifiers. A Thiobacillus-dominated mixed culture converted poorly crystalline ferrihydrite to crystalline akaganeite, while T. denitrificans and Pseudogulbenkiania strain 2002 preferentially produced maghemite and hematite (Kiskira et al., 2019). This in-depth understanding of microbial factors is critical to predicting the fate and recovery of Fe species in natural or engineered conditions.

Ferrihydrite is a thermodynamically metastable iron (hydr)oxide that can be gradually transformed to crystalline goethite and hematite. (Bi)carbonate levels that are elevated (pCO₂, ~2%) compared to atmospheric CO₂ levels (~0.04%), a critical geochemical parameter in sediments, could increase the occurrence of hematite through olation, ligand exchange, and rearrangement (Li et al., 2020b). Under suboxic conditions, aqueous Fe(II) can catalyze ferrihydrite transformation to more stable lepidocrocite (Lp) and goethite (Gt), and citrate greatly affects the ratio of Lp to Gt (Sheng et al., 2020a).

Ligand and ESs have been reported to facilitate electron transfer between Fe(II) and IOM, enhancing microbial Fe(II) oxidation in wastewater treatment. c-Cyts is a key protein involved in IOM extracellular electron transfer. Organic ligands were detected with a rapid stopped-flow spectrometer and found to significantly accelerate the reaction between c-Cyts and Fe(II), with a reaction rate order of EDTA > citric acid > ammonia triacetate > malonic acid > glycine amino acid > control (Wang et al., 2019b). In addition, Fe(II) can be oxidized by ESs, e.g., oxidized riboflavin, flavin mononucleotide (FMN) and quinone, providing bioreinforcement strategies for the reduction of redox-sensitive nitrogen species (Zhang et al., 2020c). Thus, ESs are demonstrated a feasible and efficient strategy to promote NDAFO process.

The addition of iron ligands, such as nitrilotriacetic acid (NTA), ethylenediaminetetraacetic acid (EDTA), N-methyliminodiacetic acid (MIDA) and polyphosphate significantly accelerated Fe(III) reduction coupled to ammonium oxidation. Synergetic effects of submicromolar Fe(II) addition and ligands catalyzed the dissolution of lepidocrocite due to interfacial electron transfer of Fe(III)/Fe(II) and detachment of Fe ligands (Biswakarma et al., 2020). The extent of Feammox and the reduction of ferrihydrite were enhanced when amended with the ES AQDS (Zhou et al., 2016). Biochar facilitated long-term microbial reduction of hematite at a two fold higher rate than that in the control since the semiquinone groups on biochar likely participated in the redox reactions (Xu et al., 2016). Hence, ES-mediated Feammox could lead to greater N loss and a promoted reaction rate in wastewater.

NOM is abundant in aquatic environments and has a critical impact on FeBNR. In a ZVI-oxidizing supported autotrophic denitrification process based on iron-carbon microelectrolysis and iron scraps (ME-ISs-AD), an optimum dosage of 1.0 mg-COD/mg-TN can significantly increase the denitrification load from 0.19 to 0.44 kg-N/(m³·d) and reduce the accumulation of N₂O/NO₂⁻. The study demonstrated that organic carbon simulated the bioconversion of iron compounds and enhanced ZVI passivation, increasing the production of H₂ and Fe(II) and promoting autotrophic denitrification (Deng et al., 2020). NOM has positive and negative influences on the NDAFO process. On the one hand, NOM can behave as an electron donor to enrich mixotrophic IOM and increase bacterial activity. On the other hand, Fe(II)-NOM complexes cannot enter the periplasm to participate in electron transport chains. Aqueous Fe(II) may react enzymatically or abiotically with NO2− in the periplasm or at the surface of cells. The inhibition of denitrification by Fe(II)-OM was observed due to the large size and negative charge of such complexes (Peng et al., 2018).

Under anaerobic conditions with ferric ions as the sole electron acceptor, NOM can compete with NH4+ as the electron donor for Feammox. Therefore, NH4+ is preferentially utilized in the absence of organic carbon. In wild environments, such as saline-alkaline paddy soils, NH4+ is generally uncorrelated or negatively related to NOM under Fe(III)-dominant reducing conditions (Liu et al., 2021). However, recent studies showed that NOM was favorable for Feammox because (1) organic carbon can promote the release of structural Fe in clay minerals, supporting the release of amorphous Fe and promoting the bioavailability of Fe minerals (Glodowska et al., 2020); (2) some organic carbon, such as a humic substances, can act as electron donors and ESs to link insoluble Fe(III) and IRM (Stern et al., 2018); and (3) the degradation of organic carbon can release protons, which leads to a decrease in pH and directly provides H⁺ to the Feammox process. Although organic carbon is not required for Feammox metabolism, it does play a key role in mediating and enhancing Feammox rates.

Nitrogen pollution in groundwater is a pervasive and increasing global issue mainly due to the intense application of nitrogen fertilizers, affecting human health if exposure is long-term. Nitrate is usually reduced to N₂ by denitrifying bacteria and removed from drinking water, but this process consumes extensive amounts of organic carbon and possibly causes secondary pollution. Techniques utilizing Fe minerals provide alternative cost-effective solutions to remediate nitrogen-polluted groundwater.

A CW could be designed and built on an in situ polluted site to control nitrogen transformation, decrease transportation fees, and avoid excessive interference with the local environment. However, nitrogen removal is always limited by the absence of electron donors in CWs. Autotrophic denitrification, especially with iron supplements, has become a practical alternative method to enhance nitrogen removal. The complicated interactions between environmental parameters, substrates, and microorganisms significantly affect nitrogen transformation in iron-based CWs and have been the focus of many studies.