Introduction

Ammonia nitrogen (NH3-N) pollution is a ubiquitous inorganic pollutant in water bodies1, characterized by extensive sources and large discharge volumes. Significant quantities of industrial wastewater, including those from oil refining wastewater, pharmaceutical wastewater, and food industry wastewater, as well as landfill leachate, contain high concentrations of NH3-N, imposing continuous environmental pressure. Elevated NH3-N concentrations trigger eutrophication and induce anoxic/anaerobic conditions in aquatic systems, thereby disrupting the ecosystem stability2. Moreover, dissolved ammonium (NH4+) can be converted to nitrite (NO2-) by nitrifying bacteria, posing substantial risks to drinking water safety3. Thus, developing a cost-effective and environmentally friendly approach for NH3-N removal is of significant practical importance.

Conventional NH3-N remediation technologies include biological processes4,5, chemical precipitation6,7, and adsorption8,9. Among these, adsorption has garnered widespread attention due to its operational simplicity, cost-effectiveness, high efficiency, and environmental benignity, making it a promising technology for practical applications10. Adsorption methods primarily utilize materials such as ion-exchange resins11,12, metal-organic frameworks (MOFs)13,14, and clay minerals. Compared to ion-exchange resins and MOFs, clay mineral-based adsorbents have emerged as ideal candidates owing to their favorable selective adsorption properties, ion-exchange capabilities, and alignment with sustainable development goals15. Studies have demonstrated that clay minerals, such as bentonite, zeolite and attapulgite, when subjected to thermal activation16, alkalization17, or salt modification18, exhibit significantly enhanced NH3-N removal capacity. Among these minerals, natural vermiculite (NV), a layered clay material, has been increasingly applied in wastewater purification due to its excellent cation-exchange capacity, adsorption performance, thermal expansibility, and cost-effectiveness.

NV and its modified derivatives have received considerable attention as adsorbents for wastewater treatment. Relevant studies have shown that NV exhibits significant adsorption capacity for heavy metal ions such as Cr(VI), Pb(II), and Cu(II) in aqueous solutions, and its adsorption performance can be further enhanced through chemical modification19,20,21. In recent years, vermiculite-based materials have also been increasingly employed for NH3-N adsorption. Dias et al.22 demonstrated that vermiculite, after modification with sodium hydroxide (NaOH) and hydrochloric acid (HCl), undergoes cation exchange. Adsorption tests revealed that HCl modification led to the replacement of K+ by Al3+, whereas NaOH modification resulted in its replacement by Mg2+ and Na+. For both modifiers, the optimal concentration was 0.1 mol·L-1. Ferreira et al.23 used expanded vermiculite to adsorb NH3-N from landfill leachate. In bench-scale tests employing a fixed-bed column, two bed heights and three flow rates were evaluated, and these tests yielded an average uptake capacity of 33.4 mg·g-1 at an NH3-N concentration of 2,560 mg·L-1. In pilot-scale tests conducted with a single stainless-steel column, an average uptake capacity of 18.1 mg·g-1 was obtained with an initial NH3-N concentration of 1,193 mg·L-1. Ren et al.24 developed a convenient method for removing high concentrations of NH3–N from wastewater using NaCl-modified vermiculite. They prepared a series of modified vermiculite samples (Na–V) using NaCl solutions at varying concentrations. The NH3–N adsorption performance of Na–V was significantly enhanced compared to that of natural vermiculite (NV). Among these, vermiculite modified with 180 g·L-1 NaCl yielded the highest ammonium adsorption capacity (Qm) of 11.569 mg·g-1, which was 63.43% higher than that of NV (Qm = 7.079 mg·g-1). Currently, most studies focus on modifying NV using different chemical modifiers to enhance its adsorption performance. However, such single modification methods rely relatively heavily on the inherent cation-exchange capacity of the matrix itself, which limits further improvement in the adsorption performance of NV-based adsorbents. Therefore, exploring a low-cost, environmentally friendly modification method that can efficiently enhance the adsorption capacity for NH3-N removal from water is of great practical significance.

In this study, NV was used as the base matrix and modified with ethylenediaminetetraacetic acid disodium salt (EDTA-2Na) to improve the NH3-N removal capacity. This approach leverages the strong complexing ability of EDTA for cations25, and the ion exchange between its Na+ and the Mg2+, Al3+, and Fe3+ in NV. The objective was to develop an efficient method for removing NH3-N from water. To investigate the removal mechanism and performance of EDTA-2Na-modified vermiculite (E-V), the powder structures of NV and E-V were characterized. Their adsorption isotherms and kinetic characteristics were determined. Combined with static adsorption experiments, the effects of solution pH, E-V dosage, and common coexisting cations (K+, Ca2+, Mg2+) on NH3-N adsorption performance were systematically evaluated.

Materials and methods

Materials

NV was obtained from Hebei Province, China. Prior to use, the raw NV was subjected to pretreatment involving crushing, purification, and grinding. All chemical reagents used were of analytical grade. Ammonium chloride (NH4Cl), ethylenediaminetetraacetic acid disodium salt (EDTA-2Na), and potassium sodium tartrate (NaKC4H4O6) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Hydrochloric acid (HCl) and sodium hydroxide (NaOH) were procured from XiLong Science Co., Ltd. (Guangzhou, China). Deionized water was used throughout the experiments. Simulated NH3-N wastewater was prepared by dissolving an appropriate amount of NH4Cl in deionized water.

Modification of NV

The NV samples were ground in a mortar, sieved through a 200-mesh sieve, and stored for subsequent use. EDTA-2Na solutions with concentrations of 0.1, 0.2, and 0.3 mol·L-1 were prepared in beakers. As schematically illustrated in Fig. 1, the NV was added to the EDTA-2Na solutions at a solid-to-liquid ratio of 1:30 (g: mL). The mixtures were transferred to a constant-temperature magnetic stirrer for continuous agitation at 160 rpm for 12 h. Subsequently, solid-liquid separation was conducted, and the solid residues were thoroughly rinsed with deionized water to remove any unreacted EDTA-2Na. The washed solids were dried in an oven at 105 °C for 2 h, followed by regrinding and sieving through a 200-mesh sieve. This procedure yielded three modified vermiculite samples: 0.1E-V, 0.2E-V, and 0.3E-V, corresponding to modifications with 0.1, 0.2, and 0.3 mol·L-1 EDTA-2Na, respectively.

Fig. 1
figure 1

Preparation steps of the E-V adsorbent.

Full size image

Material characterization

The surface morphology and microzone composition of the materials were characterized by scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDS, ZEISS-Sigma300, Germany). The crystal structure, phase composition, and lattice parameters were analyzed using an X-ray diffractometer (XRD, Rigaku-SmartLab 9KW, Japan). The chemical structures, functional groups, and molecular vibrations were identified by Fourier-transform infrared spectroscopy (FT-IR, Thermo Fisher-Nicolet iS 5, USA). The specific surface area and pore volume distribution of the samples were determined from nitrogen adsorption-desorption experiments performed on a surface area analyzer (Micromeritics-ASAP 2420, USA). The specific surface area was calculated using the multi-point Brunauer-Emmett-Teller (BET) method, and the total pore volume was evaluated by the Barrett-Joyner-Halenda (BJH) method. The zeta potential of 0.E-V was analyzed using a zeta potential analyzer (Model: Zetasizer Nano ZS90, Brookhaven Instruments Corporation, USA).

Static adsorption experiments

Equal masses of 0.1E-V, 0.2E-V, and 0.3E-V adsorbents were added to separate 50 mL aliquots of simulated NH3-N wastewater, each with an identical initial concentration. The pH of each system was adjusted to 7.0 using 0.1 mol·L-1 HCl or NaOH solutions. The mixtures were then agitated on a constant-temperature magnetic stirrer at 25 °C and 160 rpm for 2 h. Upon completion of the adsorption process, solid-liquid separation was achieved by centrifugation at 5000 rpm for 5 min. The residual NH3-N concentration in the supernatant was determined spectrophotometrically using Nessler’s reagent. All experiments were conducted in triplicate, and the results are expressed as mean values. Based on these results, the 0.3E-V sample, which demonstrated the highest NH3-N removal efficiency, was selected for all subsequent experiments.

Effect of adsorbent dosage

The dosage of 0.3E-V was varied from 0 to 20 g·L-1, while the initial NH3-N concentration was maintained at 10 mg·L-1 and the solution pH was maintained at 7. The mixtures were then agitated at 25 °C and 160 rpm for 4 h. Subsequently, the samples were centrifuged, filtered, and analyzed to determine the residual NH3-N concentration.

Effect of solution pH

The initial pH of the simulated wastewater was adjusted to values ranging from 3 to 10 using 0.1 mol·L-1 HCl or NaOH, while the dosage of 0.3E-V and the initial NH3-N concentration were fixed at 10 g·L-1 and 10 mg·L-1, respectively. The adsorption experiments were conducted under identical stirring and separation conditions to those previously described.

Effect of coexisting cations

Simulated wastewater containing 10 mg·L-1 NH3-N and different concentrations of K+, Ca2+ and Mg2+ (2, 5, 10 mmol·L-1), introduced as KCl, CaCl2 or MgCl2, was prepared. The dosage of 0.3E-V and the solution pH were maintained constant at 10 g·L-1 and 7, respectively. The adsorption and analysis procedures were identical to those mentioned earlier.

Isothermal adsorption experiments

The initial pH of the solution was adjusted to 7, and the dosage of 0.3E-V was fixed at 10 g·L-1. The initial NH3-N concentration was varied from 0 to 40 mg·L-1. The mixtures were agitated at 25 °C and 160 rpm for 4 h to ensure adsorption equilibrium was reached. After the adsorption process, the mixtures were centrifuged at 5000 rpm for 5 min, and the supernatants were filtered through a 0.45 μm membrane filter. The equilibrium NH3-N concentration in the filtrate was determined, and the corresponding equilibrium adsorption capacity (Qe) was calculated. All experiments were conducted in triplicate, and average values were used for data fitting.

Adsorption kinetics experiments

The initial NH3-N concentration (10 mg·L-1), solution pH (7), and 0.3E-V dosage (10 g·L-1) were kept constant. The mixtures were stirred at 25 °C and 160 rpm, and samples were collected at predetermined time intervals (0–500 min). The collected samples were centrifuged, and the supernatants were filtered through a 0.45 μm membrane for the determination of residual NH3-N concentration. The adsorption capacity at each time interval (Qt) was calculated, and kinetic models were employed to fit the experimental data. All experiments were performed in triplicate.

Equations used

  • Calculate removal rate percentage (R):

$$R=\frac{{\left( {{C_0} - {C_e}} \right)}}{{{C_0}}} \times 100\%$$
(1)

where C0 and Ce are the initial and equilibrium concentrations of NH4+ (mg·L-1).

  • Calculate the adsorbed amount of NH4+ (Qe, mg·g-1):

$${Q_e}=\frac{{\left( {{C_0} - {C_e}} \right)V}}{m}$$
(2)

where V is the volume of solution (L), and m is the mass of the adsorbent (g).

  • Langmuir model:

$$\frac{{{C_e}}}{{{Q_e}}}=\frac{{{C_e}}}{{{Q_m}}}+\frac{1}{{{Q_m} \cdot {K_L}}}$$
(3)

where Qm is the theoretical maximum adsorption capacity (mg·g-1), and KL is the Langmuir constant (L·mg-1).

  • Freundlich model:

$$\ln {Q_e}=\ln {K_F}+\frac{1}{n}\ln {C_e}$$
(4)

where KF is the Freundlich constant [(mg·g-1)·(L·mg-1)1/n], and n is a constant indicative of adsorption intensity.

  • Separation factor:

$${R_L}=\frac{1}{{1+{K_L}{C_0}}}$$
(5)

where RL is the separation factor for the Langmuir isotherm.

  • Pseudo-first-order model:

$$\ln \left( {{Q_t} - {Q_e}} \right)=\ln {Q_e} - {K_1}t$$
(6)

where Qt is the amount of ammonium adsorbed at time t (mg·g-1), Qe is the amount adsorbed at equilibrium time (mg·g-1), and K1 is the pseudo-first order (PFO) rate coefficient (min-1).

  • Pseudo-second-order model:

$$\frac{t}{{{Q_t}}}=\frac{1}{{Q_{e}^{2}{K_2}}}+\frac{t}{{{Q_e}}}$$
(7)

where Qt is the amount of ammonium adsorbed at time t (mg·g-1), Qe is the amount adsorbed at equilibrium (mg·g-1), and K2 is the pseudo-second-order (PSO) rate coefficient (g·mg-1·min-1).

  • Intra-particle diffusion model:

$${Q_t}={K_d}{t^{1/2}}+C$$
(8)

where Kd is the rate constant for intraparticle diffusion (mg·g-1·min-1/2), and C is a constant related to the boundary layer thickness (mg·g-1).

Results and discussion

Material characterization

SEM-EDS analysis

Figure 2 presents the SEM images of NV, 0.3E-V (before adsorption), and 0.3E-V after NH3-N adsorption (designated as 0.3E-V + NH3-N). As shown in Fig. 2a, NV displays a characteristic irregular lamellar structure with a relatively smooth surface, which aligns with the morphology reported for similar materials in the literature26. Following modification with 0.3 mol·L-1 EDTA-2Na (Fig. 2b), the layered structure of the resulting 0.3E-V appears partially exfoliated, revealing a more developed porous architecture alongside a notable increase in surface roughness27. These structural modifications are conducive to increasing the density of adsorption sites on the material surface, thereby creating favorable conditions for NH3-N adsorption. Upon completion of the adsorption process (Fig. 2c), the surface of the 0.3E-V + NH3-N exhibits significantly greater flatness compared to its state before adsorption. Complementary EDS elemental analysis (Table 1) shows that the mass fraction of nitrogen on the surface of 0.3E-V + NH3-N reaches 2.16 wt%, which is 1.82 times higher than that of the pristine 0.3E-V. Concurrently, the mass fraction of sodium decreases from 3.76 wt% to 1.70 wt%. These characterization results collectively verify the successful loading of NH3-N onto the 0.3E-V material.

Fig. 2
figure 2

(a) SEM image of NV, (b) SEM image of 0.3E-V, (c) SEM image of 0.3E-V after adsorption of NH3-N, (d) EDS energy spectrum before and after adsorption of 0.3E-V.

Full size image
Table 1 Changes in major elemental composition of 0.3E-V before and after adsorption.
Full size table

XRD analysis

Figure 3 shows the XRD patterns of NV, 0.3E-V, and 0.3E-V + NH3-N. The diffraction pattern of NV is dominated by peaks corresponding to quartz (SiO2, PDF#46-1045) and vermiculite [(Mg, Fe, Al)3 (Al, Si)4O10(OH)2·4H2O, PDF#16–0613]. This phase composition is consistent with literature reports for NV28. Compared to NV, the 0.3E-V sample shows a significant reduction in the intensity of the characteristic vermiculite peaks and the emergence of a new diffraction peak. This new peak, referred to herein as the V-P peak, is indicative of a vermiculite-phlogopite interstratified mineral structure24. Its appearance signifies a notable modification of the NV interlayer structure, which we attribute to structural collapse following the incorporation of alkaline cations (Na+, K+, Rb+, Cs+)29. The treatment with EDTA-2Na facilitates the chelation of EDTA anions with interlayer cations (Mg2+, Ca2+)25, which accelerates the ion exchange of Na⁺ for the original interlayer cations and reduces the stability of the layered structure. Following NH3-N adsorption, the intensity of the V-P diffraction peak in the NH3-N-laden sample decreases relative to that in 0.3E-V. This observation is likely a consequence of the ion exchange between NH4+ and the intercalated Na+, leading to a further structural alteration that diminishes the characteristic V-P peak intensity30.

Fig. 3
figure 3

XRD spectra of NV, 0.3E-V and 0.3E-V + NH3-N (V - Vermiculite; V-P - vermiculite-phlogopite interlayer mineral structure; Q - Quartz).

Full size image

FT-IR analysis

Figure 4 displays the FT-IR spectra of NV, 0.3E-V, and 0.3E-V + NH3-N. All samples exhibit characteristic bands at 3435 cm-1 and 1639 cm-1, corresponding to O-H stretching and bending vibrations of adsorbed water31,32. The strong peak at 1013 cm-1 along with those at 691 cm-1 and 465 cm-1 are assigned to Si-O, Al-O, and Si-O-Si vibrations, confirming the vermiculite structure33. In 0.3E-V, the notable weakening of peaks at 691 cm-1 and 465 cm-1 suggests structural disruption, while new peaks at 2926 cm-1, 2850 cm-1, and 1398 cm-1, characteristic of -CH3, -NH, and C = O groups, verify the successful anchoring of EDTA-2Na34. After NH3-N adsorption, the appearance of an NH4+ peak at 1432 cm-1 and the decrease in water-related peak intensities indicate ion exchange between interlayer Na+ and NH4 +35. The smaller hydration radius and lower hydration energy of NH4+ promote dehydration, potentially resulting in interlayer collapse36. These spectral findings align well with SEM and XRD analyses.

Fig. 4
figure 4

FT-IR patterns of NV, 0.3E-V, 0.3E-V + NH3-N.

Full size image

BET analysis

Figure 5 presents the N2 adsorption-desorption isotherms and pore size distributions of NV and 0.3E-V, with corresponding BET parameters summarized in Table 2. As shown in Fig. 5a, both materials exhibit typical type IV isotherms, characteristic of mesoporous structures. The gradual increase in adsorption capacity indicates the progression from monolayer to multilayer adsorption of N2 molecules on the mesoporous surface19. A type H3 hysteresis loop is observed, commencing at a relative pressure P/P0 >0.4348, which suggests the prevalence of slit-like pores with irregular geometries within the 0.3E-V particles. The pore size distribution (Fig. 5b) further confirms the typical mesoporous nature for 0.3E-V, showing a concentration of pores within this size range.

Fig. 5
figure 5

(a) N2 adsorption-desorption isotherms and (b) pore size distribution of NV, 0.3E-V adsorbent materials.

Full size image

As indicated in Table 2, 0.3E-V possesses a larger specific surface area (9.7801 m2·g-1 vs. 7.0916 m2·g-1, a greater total pore volume (2.2709 × 10− 2 cm3·g-1 vs. 2.0858 × 10− 2 cm3·g-1), and an increased micropore volume (9.58 × 10− 4 cm3·g-1 vs. 5.02 × 10− 4 cm3·g-1), compared with NV, while its average pore size decreases from 11.7652 nm to 9.2879 nm. These alterations are attributed to ion exchange between Na+ and interlayer Mg2+/Ca2+ during modification, which disrupts the interlayer structure of NV and generates numerous new micropores and mesopores37. This structural transformation consequently increases the volumes of mesopores and micropores while reducing the average pore size. The enhanced specific surface area and pore volume of 0.3E-V provide more adsorption sites, thereby facilitating NH3-N adsorption. The observed variations in specific surface area and pore structure corroborate the conclusion of interlayer disruption derived from XRD analysis and are consistent with relevant studies on clay mineral adsorption38,39.

Table 2 BET data for NV, 0.3E-V.
Full size table

Adsorption mechanism of NH3-N

Based on multi-scale material characterization and experimental data, the modification mechanism of EDTA-2Na and the hierarchical adsorption mechanism for NH3-N were elucidated. The process involves two key steps. First, EDTA anions selectively chelate and extract free Ca2+/Mg2+ from the interlayer spaces of NV. This process not only exposes additional adsorption sites but also creates space for subsequent Na+ intercalation, thereby significantly enhancing the ion exchange capacity of the material. Second, the introduced Na+ ions promote the expansion and reorganization of the NV interlayer structure, leading to an increased volume of mesopores and micropores and a larger specific surface area. These structural modifications further augment the density of available adsorption sites. Furthermore, NH4+, owing to its smaller hydrated ionic radius and lower hydration energy compared to Na+, is preferentially adsorbed and readily undergoes ion exchange with the intercalated Na+. Collectively, these effects endow the 0.3E-V adsorbent with a greater specific surface area and more abundant adsorption sites, enabling a synergistic hierarchical adsorption process dominated by ion exchange and complemented by physical adsorption. This mechanism aligns with the adsorption characteristics of layered clay minerals and effectively overcomes the limited adsorption capacity of traditional adsorbents.

Adsorption isotherms and kinetics

Adsorption isotherms

The adsorption isotherms of NV and 0.3E-V were fitted using the Langmuir and Freundlich models, with results presented in Fig. 6; Table 3. For 0.3E-V, the Freundlich model provided a superior fit, as indicated by a higher coefficient of determination (RF2 = 0.9981). A Freundlich exponent (1/n = 0.9228 < 1) indicates a thermodynamically favorable adsorption process. The proximity of the exponent to unity (1/n ≈ 1) suggests a relatively homogeneous adsorbent surface, which also accounts for the reasonably good fit achieved with the Langmuir model. The theoretical maximum adsorption capacity derived from the Langmuir model was 18.45 mg·g-1, approximately twice that of NV (9.15 mg·g-1). Nevertheless, based on both the goodness-of-fit and the physical interpretation of the model parameters, the Freundlich model offers a more accurate description of the adsorption mechanism within the studied concentration range. This identifies the process as multilayer adsorption on a heterogeneous surface, which is consistent with a mechanism dominated by ion exchange40.

Fig. 6
figure 6

Adsorption isotherms for NV, 0.3E-V: (a) Langmuir curve, (b) Freundlich curve.

Full size image
Table 3 Isothermal adsorption model fitting results.
Full size table

Adsorption kinetics

The adsorption kinetics of NH3-N onto 0.3E-V, as shown in Fig. 7a, exhibited a rapid initial phase within the first 2 h, followed by a gradual deceleration until equilibrium was reached. This commonly observed pattern is attributed to the high initial NH3-N concentration and abundant available sites on the adsorbent surface, which collectively drive fast adsorption. As the reaction proceeds, the diminishing concentration gradient and the progressive occupation of active sites jointly reduce the mass transfer driving force, leading to the observed slowdown and eventual equilibrium.

Fig. 7
figure 7

(a) Shows the effect of contact time on the adsorption capacity of vermiculite, (b) intra-particle diffusion model, (c) PFO model for adsorption of NH3-N, (d) PSO model for adsorption of NH3-N.

Full size image

To further elucidate the rate-limiting steps, the intra-particle diffusion model was applied (Fig. 7b; Table 4). The adsorption process was found to comprise three consecutive stages. The first involves rapid diffusion of NH4+ to the external surface and initial site occupation. This is followed by diffusion into the internal pores, and finally a stage approaching equilibrium where the rate declines markedly due to enhanced electrostatic repulsion and mass transfer resistance. The multi-linear nature of the Qt versus t1/2 plot, where none of the fitted lines pass through the origin, indicates that the adsorption process is co-controlled by liquid film diffusion and intraparticle diffusion, rather than being dominated solely by the latter41.

Further analysis using pseudo-first-order and pseudo-second-order kinetic models (Fig. 7c and d; Table 5) showed that the pseudo-second-order model provided a superior fit (Rs2 = 0.9998). This suggests that chemisorption-related processes act as a rate-limiting step, consistent with the ion exchange-dominated mechanism proposed in this study and thereby offering additional mechanistic support.

Table 4 NV, 0.3E-V internal diffusion models.
Full size table
Table 5 Adsorption kinetics of NH3-N by NV, 0.3E-V.
Full size table

Factors influencing NH3-N adsorption

Effect of adsorbent dosage

Figure 8 illustrates the influence of adsorbent dosage on both the NH3-N removal efficiency and the unit adsorption capacity. Under consistent experimental conditions, as the dosage of 0.3E-V increased from 1 to 60 g·L-1, the NH3-N removal efficiency rose markedly from 66.79% to 97.09%. This enhancement can be attributed to the increased dosage providing a greater specific surface area and more active sites, thereby facilitating the adsorption process42. Notably, when the dosage reached 20 g·L-1, the removal efficiency achieved 95.88%, representing efficient removal. A further increase in dosage from 20 to 60 g·L-1 resulted in a considerably diminished growth rate of the removal efficiency. Concurrently, the unit mass adsorption capacity exhibited a continuous decline with increasing dosage. This trend is primarily due to the overlapping of active sites between adsorbent particles. An excessive dosage leads to a surplus of adsorption sites in the system, which not only reduces site utilization efficiency but also represents an inefficient consumption of adsorbent material.

Fig. 8
figure 8

Effect of dosage on the performance of adsorbed NH3-N.

Full size image

Effect of initial solution pH

The influence of the initial solution pH, ranging from 3 to 10, on the NH3-N adsorption performance of NV and 0.3E-V was systematically investigated at the optimal adsorbent dosage of 20 g·L-1. As shown in Fig. 9a, 0.3E-V exhibited superior NH3-N adsorption capacity and removal efficiency compared to NV across the entire pH range studied, confirming the effectiveness of the EDTA-2Na modification. The maximum NH3-N removal efficiency of 0.3E-V reached 95.88% at pH 7, representing a 32.7% increase over the removal efficiency of 63.18% achieved by NV.

Fig. 9
figure 9

(a) Effect of pH on the performance of adsorbed NH3-N, (b) The potential values of 0.3E-V samples under different pH conditions and (c) The isoelectric point of 0.3E-V samples.

Full size image

Zeta potential measurements (Fig. 9b) revealed that the surface charge of 0.3E-V remained negative throughout the studied pH range. This persistent negativity stems from the dissociation of carboxyl groups in EDTA, which impart a high density of negative charges onto the vermiculite surface, providing direct electrochemical evidence of successful modification43.

The zeta potential data measured at pH 1, 2, and 3 were selected for fitting analysis44, with the results shown in Fig. 9c. The pH value at which the zeta potential is zero is defined as the point of zero charge (pHPZC); the pHPZC of 0.3E-V was determined from the fitting curve to be 0.213. This low pHPZC underscores the strong dependence of the adsorption mechanism on solution acidity. In acidic media, the high concentration of H+ competes with NH4+ for active sites, with the higher adsorption energy of H+ (-41.2 kJ·mol-1) compared to NH4+ (-28.7 kJ·mol-1) significantly inhibiting NH3-N adsorption45. Under alkaline conditions, the rise in pH promotes the conversion of NH4+ to NH3·H2O, while metal ions on the adsorbent surface react with OH- to form hydroxide colloidal layers, collectively further hindering NH3-N adsorption.

Effect of coexisting cations

Common cations in natural water bodies (K+, Ca2+, Mg2+) may compete with NH4+ for adsorption sites, influencing NH3-N removal efficiency. As illustrated in Fig. 10, the adsorption of NH3-N by 0.3E-V was inhibited in a concentration-dependent manner with increasing levels of coexisting cations (2, 5, 10 mmol·L-1), and the inhibition strength followed the order: K+ >Ca2+ >Mg2+. This phenomenon is primarily governed by the hydration radius effect: the hydration radii of Mg2+ (4.28 Å) and Ca2+ (4.12 Å) are larger than that of NH4+ (3.31 Å), whereas K+ shares an identical hydration radius (3.31 Å) with NH4+, rendering K+ more likely to occupy the same adsorption sites as NH4+46. Additionally, although divalent cations (Ca2+, Mg2+) exhibit stronger electrostatic affinity, their high hydration energies (Ca2+: 1577 kJ·mol-1; Mg2+: 1921 kJ·mol-1) restrict their migration to the solid-liquid interface, resulting in a relatively weaker competitive inhibition effect compared to K+.

Fig. 10
figure 10

Effect of co-existing ions on the adsorption performance of NH3-N.

Full size image

Conclusions

  1. 1.

    Vermiculite was modified with EDTA-2Na at varying concentrations (0.1–0.3 mol·L-1), with the 0.3 mol·L-1 variant (0.3E-V) identified as optimal. Under conditions of pH 7 and a dosage of 20 g·L-1, 0.3E-V achieved a maximum NH3-N removal efficiency of 95.88%.

  2. 2.

    Competitive cation experiments revealed that K+, Ca2+, and Mg2+ all inhibit NH4+ adsorption, with K+ showing the strongest effect due to its identical hydration radius to NH4+ (3.31 Å).

  3. 3.

    Adsorption isotherm analysis indicated that the Freundlich model (RF2 = 0.9981) better described the process, with a Freundlich exponent 1/n = 0.9228 (< 1) confirming favorable adsorption on a heterogeneous surface. Although the near-unity 1/n value allowed the Langmuir model (RL2 = 0.9573) to also provide a reasonable fit, the Freundlich model was deemed more representative across the studied concentration range, suggesting multilayer adsorption behavior. The theoretical maximum adsorption capacity of 0.3E-V was 18.45 mg·g-1, 2.02 times that of natural vermiculite (9.15 mg·g-1).

  4. 4.

    Kinetic studies showed that adsorption followed the pseudo-second-order model (Rs2 = 0.9998, k2 = 1.193 × 10− 3 g·mg-1·min-1), indicating chemisorption as the rate-controlling step. The process was characterized by an initial rapid phase dominated by surface adsorption, followed by a diffusion-controlled approach to equilibrium.

  5. 5.

    These results collectively support a synergistic adsorption mechanism for NH3-N onto EDTA-2Na-modified vermiculite, dominated by ion exchange and complemented by physical adsorption.