Ammonium and Phosphate Removal from Aqueous Solutions by Zeolite and Gravel: Kinetics and Adsorption Isotherms

Abstract

Although constructed wetlands (CWs) are a viable solution for wastewater treatment, substrate selection significantly affects their performance. This study evaluated the adsorption behavior of ammonium and orthophosphate on natural zeolite (coarse- and fine-grained) and coarse gravel using kinetic and isotherm experiments. Coarse materials are intended for use as filler media in CWs to address problems such as clogging. Ammonium removal due to adsorption reached 96.20% and 96.49% for coarse and fine zeolite, respectively, and 16.84% for gravel. For orthophosphate, the removal was 11.46% and 12.81% for coarse and fine zeolite, respectively, and 6.70% for gravel. Kinetic analysis showed that the adsorption of both nutrients followed the pseudo-second-order model. Zeolite exhibited high ammonium adsorption capacities (181.87 and 174.23 mg/kg), with granulometry showing minimal effect. The orthophosphate adsorption capacities were lower (11.76 and 12.35 mg/kg for zeolite; 6.44 mg/kg for gravel). Isotherm modeling indicated that ammonium adsorption fitted better to the Langmuir model (monolayer adsorption), while orthophosphate followed the Freundlich model (heterogeneous surface adsorption). Ζeolite adsorbed six times more ammonium and twice as much phosphate as gravel. These findings suggest that natural zeolite is an effective and sustainable CW substrate, enhancing nutrient removal and serving as an economical and environmentally friendly alternative to traditional filler media.

1. Introduction

Large volumes of nitrogen and phosphate pollution are released into water bodies as a result of the social economy’s growth and extensive industrial and agricultural production, posing a major threat to domestic water safety and productivity [1]. Furthermore, the eutrophication of rivers due to the combined pollution of phosphorus and nitrogen can quickly put human life and the aquatic environment at even greater risk [2]. Currently, some removal technologies, including ion exchange, membrane methods, electro-dialysis, coagulation sedimentation, and biological processes, are frequently employed for the removal of phosphorus and nitrogen [3,4,5]. Nevertheless, the majority of these techniques are not appropriate for the large-scale advanced treatment of phosphorus and nitrogen wastewater and involve substantial operating costs.

Constructed wetlands (CWs) are now a commonly accepted solution for wastewater treatment. Unlike natural ecosystems, this sustainable ecotechnology uses a more controlled setting to remove pollutants using physical, chemical, and biological processes found in natural wetlands [6]. Wastewater remediation can be carried out more effectively, sustainably, and economically with CWs. It also has a high rate of nutrient recovery and low maintenance/operational costs in an environmentally beneficial manner [7].

The vegetation type and flow patterns within the system can be used to categorize CWs into two groups: surface flow constructed wetlands (SF CWs) and subsurface flow constructed wetlands (SSF CWs). SF CW systems are often designed with a water column above a substrate layer, which is typically soil. Gravel beds with vertical flow (VF) or horizontal subsurface flow (HSSF) are commonly used as SSF CWs. Further differentiation of emergent and submerged macrophyte wetlands is based on the kind of vegetation. Rooted emergent macrophyte systems are the most prevalent. The combination of many CW types in a single facility is referred to as a hybrid wetland system [8,9].

CWs can manage wastewater from a variety of sources, including storm runoff, acid mine waste, municipal, agricultural, industrial, livestock, and landfill leachate [6,10,11,12]. Through microbial degradation, plant uptake, substrate adsorption, filtering by the packed media, and sedimentation, CWs can remove a range of contaminants from wastewater, such as total suspended solids (TSSs), total Kjeldahl nitrogen (TKN), total phosphorus (TP), total coliforms (TCs), biochemical oxygen demand (BOD), chemical oxygen demand (COD), heavy metals, antibiotics, pesticides, textile dyes, hormones, petroleum, and explosives [11,13].

The three key elements of these biologically designed systems are plants, microorganisms, and porous filler materials. Microorganisms are essential for the elimination of pollutants in CWs because they break down organic contaminants and convert different nutrients [14,15]. Aquatic plants, or macrophytes, are essential to CWs since they are the main means of moving dissolved oxygen (DO) through the system. Also, macrophytes reduce eutrophication in downstream streams through the uptake of pollution and the absorption of nutrients like phosphate and nitrogen [16]. In addition to serving as a filtration and/or adsorption medium for pollutants in the water, the materials in the media (soil, sand, pebbles, and gravel) offer a vast surface area for microorganisms to adhere to, promoting the growth of macrophytes [17]. The microorganisms that create biofilms in the rhizosphere and porous media are responsible for the mechanisms that change organic matter and nutritional compounds [9,18].

However, the removal performance of CWs is not consistent. There have been numerous issues with nitrogen and phosphorus when using CWs globally. The removal of COD, BOD, TSS, and other contaminants can be accomplished well by CWs, but the removal of nitrogen and phosphorus is comparatively less effective. Reports indicate that CWs based on gravel remove only about 45% and 35% for NH₄-N and orthophosphate, respectively [12,19].

In CW systems, choosing suitable substrates is essential for maximizing treatment effectiveness since these substrates are important for the production of biofilms and adsorption process [16,20]. Zeolite substrates for CWs have a broad variety of applications and a good removal efficacy. Zeolites occur in a wide variety and are available worldwide. There are many types of zeolites, including synthetic and natural zeolite. In addition to expanding the range of applications for zeolite in wastewater treatment, using it as a CW substrate increases the material’s usefulness as a green absorbent material [21]. It possesses a high ion exchange capacity, a large specific surface area, a robust adsorption capacity, and the ability to support microbial colonization. Its incorporation has been associated with enhanced nitrogen removal and biofilm formation [22,23,24,25], while its impact on plant and microbial interactions remains a subject of active investigation. Also, zeolite may compensate for the shortcomings of a single substrate and greatly enhance removal efficiency when combined with other materials, such as the CW substrate [26].

The removal of ammonium nitrogen is one of the most important uses of zeolite. Research shows that CWs with zeolite substrates outperform traditional substrates like sand or gravel, achieving ammonium removal rates of over 95% [27]. Zeolite is essential for retaining phosphorus and removing heavy metals in addition to nitrogen. Hexavalent chromium Cr(VI) removal through microbial-mediated processes was greatly enhanced with modified zeolite coatings (e.g., Zn-layered double hydroxides, Zn-LDHs), whereas zeolite-filled tidal flow wetlands in hybrid CW systems achieved 46–64% phosphate removal. Zn-LDH–zeolite substrates enhanced the synthesis of extracellular polymeric substances (EPSs) and microbial activity (e.g., Novosphingobium, Brevundimonas), both of which are essential for the reduction of Cr(VI). Zeolite is also exceptionally effective at removing pesticides and pharmaceuticals. Zeolite substrates removed >98% of antibiotics, such as ciprofloxacin and erythromycin, from tidal flow CWs treated with pharmaceutical wastewater, mostly due to microbial decomposition (61% contribution) [28]. In a comparable way, zeolite-enhanced CWs outperformed gravel-only systems in removing fluopyram from pesticide-laden rinsing water by 72%, with bioaugmentation further improving results [29].

A major problem that occurs in CWs with an impact on their sustainable development is clogging. This occurs through various mechanisms, such as the sedimentation of suspended solids, the accumulation of organic matter in the substrate, biofilm development, and plant root system development. The factors that influence and have the greatest impact on CW clogging are the porosity of the substrate and the organic load. The smaller the porosity of the substrate, the easier it is to clog. Porosity depends on the type, grain size, and unevenness coefficient of the substrate. When the grain size is 3–4 mm, there is a risk of clogging, while substrates with larger grain sizes effectively prevent clogging [30,31].

While various substrates have been investigated for their ability to enhance TKN, NH₄-N, and TP removal in CW systems, a significant gap remains regarding natural zeolite as a substrate. Especially for natural zeolites, the quantification of the adsorption capacity and a comprehensive description of the adsorption mechanisms have not been investigated, although the zeolite substrate in CWs guarantees effective nitrogen and phosphorus removal [32]. In the literature, there are studies describing the ability of zeolites to remove ammonia and orthophosphates through adsorption, but they refer to zeolite that is either in powder or granular form with a very small grain size (i.e., <4 mm) [33,34,35,36]. However, these forms are unsuitable for use as a substrate in CWs due to the clogging problem, as described above. The aim of this study is to evaluate the adsorption capacity of natural zeolite for the removal of ammonium and orthophosphate from aqueous solutions to assess its potential use as a substrate in CWs. Two granulometries are evaluated and compared: a small one (grain size 1.0–2.5 mm) that is not recommended for use as a filler in CWs due to clogging problems and a larger one (grain size 8–16 mm) that can be used as a filler in CWs. The adsorption capacity of zeolite is also compared with that of gravel (grain size 8–16 mm), which is widely used as a substrate in CWs. For this purpose, kinetic and isotherm adsorption models are used.

2. Materials and Methods

2.1. Experimental Design of Adsorption Kinetics

In order to evaluate the adsorption of ammonium cations (NH₄⁺) and orthophosphate anions (PO₄⁻³) on porous media, adsorption kinetics and isotherm experiments were conducted at the Laboratory of Ecological Engineering and Technology, Department of Environmental Engineering, Democritus University of Thrace (DUTh), Xanthi, Northern Greece. The selected porous media (or substrate) were (a) coarse-grained zeolite (code name: CG-Z, range 8–16 mm), (b) fine-grained zeolite (code name: FG-Z, range 1–2.5 mm), and (c) coarse-grained gravel (code name: CG-G, range 8–16 mm). The zeolite (provided by “ZEOLIFE”, Thessaloniki, Greece) consists primarily of SiO₂ (~69.62%), Al₂O₃ (~13.62%), and other minor components including CaO (~3.28%), K₂O (~2.94%), MgO (~0.9%), Na₂O (~0.55%), Fe₂O₃ (~0.75%), and TiO₂ (~0.11%), with a loss of ignition (LOI at 1050 °C) of 8.23% and a clinoptilolite content of at least 85%. The gravel was of igneous origin with the following composition: Si 28.50%, Al 7.95%, Fe 4.22%, Ca 3.62%, Mg 1.76%, and P 0.11% and was obtained from the bed of a local river. The chemical composition of the gravel was measured at the Laboratory of the Institute of Geological and Mining Research, Xanthi, and the granulometric analysis was performed at the Laboratory of Building Materials, Department of Civil Engineering, DUTh, Xanthi. The porous materials were washed with deionized water to remove dust and placed in an oven for 24 h at 105 °C for drying.

One of the more important physical characteristics of an absorbent used to remove pollutants in CWs is surface area. The precise calculation of surface area using the Brunauer–Emmett–Teller (BET) isotherm analysis may help to explain the differences among the investigated materials. However, it was decided to perform batch adsorption studies in order to include the ions under consideration (i.e., ammonium and orthophosphate) and to obtain a comprehensive picture of the sorption capacity of the investigated materials for these ions. For the kinetic adsorption experiments on ammonium ions, approximately 100 g of adsorbent material (i.e., CG-Z, FG-Z and CG-G) and 500 mL of NH₄Cl solution with an ammonium nitrogen concentration of 35 mg/L were placed in glass bottles. The pH of the solution was adjusted to be approximately 7.0–7.5 (natural value in constructed wetlands). For each adsorbent material (or porous medium), 3 series of 7 bottles each (a total of 63 bottles) were prepared. Furthermore, seven control bottles were prepared using only ammonia solution. The bottles were sealed airtight and placed in a thermostatic chamber at 20 °C and continuously stirred (start of experiment, time 0 h). At a certain time after the beginning of the experiment, i.e., 3, 6, 12, 24, 48, 96, and 144 h, one bottle of each adsorbent material was taken and the ammonium nitrogen (NH₄-N) concentration in the solution was measured.

For the orthophosphate adsorption kinetic experiments, the same procedure was followed as in the case of ΝH₄-Ν. In this case, 500 mL of an aqueous solution with a PO₄-P concentration of 20.0 mg/L was added to glass bottles containing approximately 100 g of adsorbent material each. This solution was prepared by adding KH₂PO₄ and K₂HPO₄ in an appropriate amount and ratio in distilled water so that the pH of the final solution was 7.2 ± 0.2. A total of 72 bottles were prepared (i.e., for each adsorbent, 3 rows of 8 bottles each). Furthermore, eight control bottles were prepared using only orthophosphate solution. The bottles were placed in a thermostatic chamber at 20 °C and continuously stirred. At a certain time after the beginning of the experiment, i.e., 6, 9, 12, 24, 48, 96, 192, and 288 h, one bottle of each adsorbent was taken and the phosphorus concentration (PO₄-P) in the solution was measured. The control bottles with the ammonium solution and the orthophosphate solution remained in the heated chamber for 144 and 288 h, respectively.

The amount of ammonium and orthophosphate adsorbed at equilibrium was calculated with the following equation:

$${\mathrm{q}}_{\mathrm{e}}={\displaystyle \frac{{(\mathrm{C}}_{\mathrm{i}}-{\mathrm{C}}_{\mathrm{e}})}{\mathrm{M}}}\mathrm{V}$$

(1)

where q_e (μg/g) is the adsorption capacity at equilibrium; C_i and C_e (mg/L) are the initial and equilibrium concentrations of NH₄-N or PO₄-P in the solution, respectively; V (L) is the volume of solution; and M (kg) is the mass of the adsorbent (i.e., CG-Z, FG-Z, and CG-G added to each glass bottle).

The amount of NH₄-N and PO₄-P adsorbed, q_t (μg/g), at time t (h) was calculated with the following equation:

$${\mathrm{q}}_{\mathrm{t}}={\displaystyle \frac{{(\mathrm{C}}_{\mathrm{i}}-{\mathrm{C}}_{\mathrm{t}})}{\mathrm{M}}}\mathrm{V}$$

(2)

where C_i and C_t (mg/L) are the initial concentration and the concentration at time t of NH₄-N or PO₄-P in the solution, respectively; V (L) is the volume of solution (i.e., 500 mL); and M (kg) is the mass of the adsorbent (i.e., CG-Z, FG-Z, and CG-G added to each glass bottle).

The data collected from the above adsorption experiments were used to find which adsorption kinetic model best describes the adsorption of ammonium and orthophosphate on the three adsorbent materials. In the present work, three kinetic models were tested, the pseudo-first-order and pseudo-second-order models [37,38,39] and the intra-particle diffusion model (or the Weber and Morris model) [40,41].

The pseudo-first-order model and its linear form and the pseudo-second-order model are described by Equations (3), (4), and (5), respectively:

$${\mathrm{q}}_{\mathrm{t}}={\mathrm{q}}_{\mathrm{e}}\left[1-\exp \left(-{\mathrm{k}}_1\mathrm{t}\right)\right]$$

(3)

$$\log \left({\mathrm{q}}_{\mathrm{e}}-{\mathrm{q}}_{\mathrm{t}}\right)=\log {\mathrm{q}}_{\mathrm{e}}-{\displaystyle \frac{{\mathrm{k}}_1}{2.303}}\mathrm{t}$$

(4)

$${\displaystyle \frac{1}{{\mathrm{q}}_{\mathrm{t}}}}={\displaystyle \frac{1}{{\mathrm{q}}_{\mathrm{e}}}}+{\displaystyle \frac{1}{{\mathrm{k}}_2{\mathrm{q}}_{\mathrm{e}}^2\mathrm{t}}}$$

(5)

where q_t (mg/kg) is the amount of adsorbed substance (i.e., NH₄-N or PO₄-P) at time t (h) and k₁ (1/h) and k₂ (kg/(mg h)) are the constant rates of the pseudo-first- and pseudo-second-order equations, respectively.

The following criteria were used to evaluate the fit of the kinetic models to the experimental data: (a) the correlation coefficient (R²), (b) the root mean square error (RMSE) and the normalized objective function (NOF), and (c) the mean absolute error (MAE) [42].

The RMSE, NOF, and MAE are expressed by the following equations:

$$\mathrm{RMSE}=\sqrt{{\displaystyle \frac{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{N}}{\left({\mathrm{P}}_{\mathrm{i}}-{\mathrm{O}}_{\mathrm{i}}\right)}^2}}{\mathrm{N}}}}$$

(6)

$$\mathrm{NOF}={\displaystyle \frac{\mathrm{RMSE}}{{\mathrm{O}}_{\mathrm{mean}}}}$$

(7)

$$\mathrm{MAE}={\displaystyle \frac{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{N}}\left|{\mathrm{O}}_{\mathrm{i}}-{\mathrm{P}}_{\mathrm{i}}\right|}}{\mathrm{N}}}$$

(8)

where O_i represents the measured (or observed) values; P_i represents the predicted values; O_mean is the mean of the measured values; and N is the total number of measurements. The acceptable value of NOF is between 0.0 and 1.0, and the optimum value of both NOF and MAE is 0.0.

The intra-particle diffusion model is described by the following equation [40,41]:

$${\mathrm{q}}_{\mathrm{t}}={\mathrm{k}}_{\mathrm{i}\mathrm{d}}{\mathrm{t}}^{0.5}+\mathrm{C}$$

(9)

where k_id (mg/(kg·h^0.5)) is the intra-particle diffusion rate constant and C (mg/kg) is the intercept, which indicates the thickness of the boundary layer.

2.2. Experimental Design of Adsorption Isotherm

In 750 mL glass bottles, 100 g of adsorbent material (i.e., CG-Z, FG-Z and CG-G) and 500 mL of NH₄Cl solution containing different initial NH₄-N concentrations (i.e., 20, 30, 40, 50 and 60 mg/L) were added. In the glass bottles with CG-G as porous media, the initial NH₄-N concentrations ranged from 2 to 60 mgN/L (2, 5, 10, 20, 40, and 60 mgN/L). For the adsorption isotherm experiments on orthophosphates, the same procedure was followed as in the case of NH₄-N, with the difference being that an aqueous solution of orthophosphates with a concentration ranging from 2.5 to 20 mgP/L (i.e., 2.5, 5, 10, 15 and 20 mgP/L) was added to the bottles with the adsorbent material (i.e., CG-Z, FG-Z, and CG-G). The experiments were carried out at two temperatures, 10 °C and 25 °C. Samples from the bottles were taken on the 6th day, which was considered the threshold where, as shown by the kinetic adsorption experiments, substrate saturation occurs.

The experimental data were fitted to Langmuir and Freundlich isotherm models, which are frequently used in the literature [43,44]. The Langmuir and Freundlich models (linear form) are given by Equations (10) and (11), respectively:

$${\displaystyle \frac{{\mathrm{C}}_{\mathrm{e}}}{{\mathrm{q}}_{\mathrm{e}}}}={\displaystyle \frac{{\mathrm{C}}_{\mathrm{e}}}{{\mathrm{q}}_{\mathrm{m}}}}+{\displaystyle \frac{1}{{\mathrm{k}}_{\mathrm{L}}{\mathrm{q}}_{\mathrm{m}}}}$$

(10)

$$\log {\mathrm{q}}_{\mathrm{e}}=\log {\mathrm{k}}_{\mathrm{F}}+{\displaystyle \frac{1}{\mathrm{n}}}\log {\mathrm{C}}_{\mathrm{e}}$$

(11)

where q_m (mg/kg) is the theoretical maximum adsorption efficiency, q_e (mg/kg) is the amount of substance (i.e., ammonium or orthophosphates) adsorbed on the porous media at equilibrium, C_e (mg/L) are the equilibrium concentrations of the solute (i.e., ammonium or orthophosphates) in the solution, k_L (L/mg) is the Langmuir adsorption constant and represents the adsorption energy, k_F (mg/kg) is the Freundlich adsorption constant, and n is the adsorption intensity constant. The n value determines the type of adsorption as follows: if n = 1, the adsorption is linear; if n < 1, the adsorption process is chemical; and if n > 1, the adsorption is a favorable physical process [45,46]. The coefficient of determination (R²) was used to evaluate the fit of the isothermal models to the experimental data.

2.3. Thermodynamic Study

The relationships between thermodynamic parameters such as the Gibbs free energy (ΔG), enthalpy (ΔH), and entropy (ΔG) are shown in the following equations [33,47,48]:

ΔG = ΔH − T ΔS

(12)

ΔG = −R T lnK_d

(13)

where K_d (mL/g) is the equilibrium constant and is calculated from the slope of q_e vs. C_e, R is the universal gas constant (J/(mol K)), and T is the absolute temperature (K).

Combining Equations (12) and (13) yields

$$\ln {\mathrm{K}}_{\mathrm{d}}={\displaystyle \frac{\mathrm{\Delta }\mathrm{S}}{\mathrm{R}}}-{\displaystyle \frac{\mathrm{\Delta }\mathrm{H}}{\mathrm{R}\mathrm{T}}}$$

(14)

Assuming that ΔH is approximately independent of temperature, ΔH and ΔS are obtained from the slope and intercept, respectively, of the lnK_d versus 1/T plot.

2.4. Chemical Analysis

The ammonia nitrogen (NH₄-N) and orthophosphate phosphorus (PO₄-P) concentrations were measured according to the Standard Methods [49]. For the measurement of NH₄-N, the titrimetric method (4500-NH₃ C) was used after preliminary distillation (the 4500-NH₃ B method). In this method, the sample is distilled in boric acid solution and the ammonia is titrated in the distillate with a standard titration solution of 0.02 N H₂SO₄ until the indicator turns pale lavender. For the PO₄-P measurement, the vanadomolybdophosphoric acid colorimetric method (4500-P C) was used. In this method, filtered samples (0.45 μm, Whatman) were reacted with vanadate–molybdate reagent. After color development (about 10 min), the absorbance of the sample was measured against a blank at a wavelength of 470 nm. The phosphorus concentration was determined using the prepared calibration curve.

3. Results and Discussion

3.1. Adsorption Removal

The removal of NH₄-N and PO₄-P due to adsorption in the three porous media (i.e., CG-Z, FG-Z, and CG-G) is presented in Figure 1. The experimental findings showed that the type of porous medium and the duration of contact had a significant impact on the removal efficiency of orthophosphate and ammonium from the aqueous solutions. It is clearly seen that for both ammonium and orthophosphates, zeolite achieves greater removal rates compared to gravel. Specifically, Figure 1a shows that 65.33% of the ammonium was removed due to adsorption in CG-Z and FG-Z in the first 3 h. Then the ammonium removal rate gradually decreased until 48 h, with the removal reaching 96.23% and 96.34% for CG-Z and FG-Z, respectively. From this time point until 144 h, there was essentially an equilibrium, as there was no change in the concentration of NH₄-N in the aqueous solution. The average removal rate at equilibrium was 96.20% and 96.49% for CG-Z and FG-Z, respectively. Regarding coarse gravel (CG-G), on the other hand, ammonium removal gradually increased from 5.65% in the first 3 h to 15.48% at 24 h. It then remained almost constant at these levels until 144 h (in equilibrium state), finally achieving an average ammonium removal rate of 16.84%. Furthermore, the removal efficiencies in this study are comparable to those of Montalvo et al. [50], who, using an integrated system with zeolite and lime, achieved ammonium removal efficiencies of approximately 98%.

The removal of orthophosphates from the aqueous solution by adsorption in the studied porous media was much lower and occurred at a slower rate compared to that of NH₄-N (Figure 1b). The removal rate in the first 6 h was 1.72%, 4.79%, and 2.31% for CG-Z, FG-Z, and CG-G, respectively. Equilibrium was reached after 96, 192, and 96 h for CG-Z, FG-Z, and CG-G, respectively, with an average removal rate of 11.46%, 12.81%, and 6.70%, respectively, until 288 h (Figure 1b). These results are in agreement with those reported by Andrés et al. [51].

In both ammonium and orthophosphate removal, fine-grained zeolite generally performed better than the other substrates, highlighting the importance of material type and particle size in adsorption efficiency. Despite being frequently utilized in constructed wetlands, gravel showed only a slight ability to retain any pollutant under the conditions evaluated. More specifically, these results show that zeolite adsorbs approximately six times more NH₄-N and twice as much PO₄-P compared to gravel. Also, there was no significant difference in the removal of both ammonium and orthophosphate between CG-Z and FG-Z. Therefore, when designing constructed wetlands for wastewater treatment, for ammonia and orthophosphate removal, the total or partial replacement of the filling material, which is usually gravel [52], with zeolite would improve the performance of the system. Furthermore, using a coarser material as the filling material in the CW (i.e., CG-Z instead of FG-Z) had a minimal effect on contaminant removal while improving the hydraulic conductivity and avoiding clogging, as finer materials contribute more to clogging and related hydraulic problems [53].

According to the adsorption mechanisms documented in the literature, the experimental results clearly show differences in adsorption efficacy between zeolite and gravel substrates. The higher ammonium removal by zeolites (96.20–96.49% efficiency) is consistent with earlier research showing that clinoptilolite’s remarkable cation exchange capacity is caused by its negatively charged surface and aluminosilicate structure [31]. According to He et al. [26], the quick initial adsorption phase (65.33% removal in 3 h) is caused by easily accessible exchange sites on zeolite surfaces. The low ammonium uptake by gravel (16.84%) is in line with research by Zhou et al. [19], who highlighted the poor ion exchange characteristics of gravel-based built wetlands and observed comparable low removal rates.

Vera et al. [54] examined two mesocosm-scale constructed wetland systems operating in parallel. In one system, the filling material was zeolite, and in the other, it was gravel. Each system consisted of two CWs. The concentration of orthophosphate was measured at the inflow and outflow of both systems. The results showed that the orthophosphate removal efficiency was 70% in the zeolite medium, which was statistically greater (p < 0.05) than the results obtained from the gravel medium. The limited orthophosphate removal observed in our system (Figure 1b) is likely attributed to the absence of artificial aeration and, generally, to the different environmental conditions compared to those of constructed wetlands. More specifically, in a study by Vera et al. [54], phosphorus removal efficiencies of approximately 70% were achieved using clinoptilolite-based zeolite; however, this was under continuous or cyclic artificial aeration. In addition, the presence of macrophytes in constructed wetlands can further enhance phosphorus removal through direct uptake by plant tissues and through rhizosphere-mediated processes, where plant roots provide oxygen and exudates that stimulate microbial activity and promote phosphorus immobilization.

3.2. Adsorption Kinetics

To assess the rate and mechanism of the adsorption process, the kinetics of the adsorption of ammonium and orthophosphate onto the investigated substrates were further examined. In order to achieve this, two common kinetic models—the pseudo-first-order and pseudo-second-order equations—were fitted to the experimental data. Figure 2 and Figure 3 show the resulting adsorption patterns and model fits for ammonium and phosphate, respectively, while

Table 1. Kinetic parameters of ammonium and phosphate adsorption onto various substrates.

Substrate	qe,m mg/kg	Pseudo-First-Order Model					Pseudo-Second-Order Model
		qe,c mg/kg	k1 1/h	R2	NOF	MAE	qe,c mg/kg	k2 kg/(mg h)	R2	NOF	MAE
Ammonium
CG-Z	181.87	168.57	0.083	0.94	0.14	18.51	181.81	0.0036	0.99	0.03	4.66
FG-Z	174.23	179.47	0.088	0.95	0.09	9.92	178.57	0.0034	0.99	0.03	2.87
CG-G	29.34	30.94	0.067	0.97	0.09	1.96	34.84	0.0039	0.87	0.13	2.26
Phosphate
CG-Z	11.76	10.84	0.0163	0.98	0.10	0.66	13.38	0.0025	0.98	0.07	0.46
FG-Z	12.35	11.30	0.0162	0.92	0.24	1.71	12.24	0.0093	0.95	0.06	0.53
CG-G	6.44	5.56	0.0159	0.92	0.29	1.04	6.82	0.0123	0.98	0.07	0.26

q_e,m = measured adsorption capacity at equilibrium; q_e,c = calculated adsorption capacity at equilibrium; NOF = normalized objective function; MAE = mean absolute error; CG-Z = coarse-grained zeolite; FG-Z = fine-grained zeolite; CG-G = coarse-grained gravel.

summarizes the kinetic parameters obtained from the model fitting. A thorough analysis of these findings is provided in the section that follows, emphasizing the variations in the materials’ adsorption behaviors as well as the models’ applicability in explaining the trends that were noticed.

The adsorption of ammonium proceeded rapidly during the initial hours in all cases, and its rate decreased as equilibrium approached (Figure 2). The adsorption process was nearly finished in the first 24 h for both coarse-grained and fine-grained zeolite, and the measured equilibrium adsorption capacities (q_e,m) were 181.87 mg/kg and 174.23 mg/kg for CG-Z and FG-Z, respectively. Moreover, the calculated adsorption capacity at equilibrium (q_e,c) from the pseudo-second-order model closely matched the q_e,m values, as they were 181.81 mg/kg and 178.57 mg/kg for CG-Z and FG-Z, respectively (Table 1). Furthermore, the pseudo-second-order model showed a better fit compared to the pseudo-first-order model for zeolite data (i.e., CG-Z and FG-Z), with high values for the correlation coefficient (R² = 0.99) and low values for both the normalized objective function (NOF value of 0.03) and mean absolute error (MAE values of 4.66 for CG-Z and 2.87 for FG-Z) (Table 1). These results are in agreement with other previous studies [33,36,55,56].

The two granulometries of the examined zeolite showed similar ammonium adsorption, indicating that a reduction in the grain size does not significantly affect the adsorption capacity. Ammonium adsorption on the zeolite occurs both on the external surface of the grains and on the internal surfaces of the pores and channels that the zeolite has. With a reduction in grain size, the external surface increases, but the internal pores and channels do not. Thus, for a small grain size, there is a larger external surface but a smaller internal surface compared to a large grain size [55,57,58]. However, other researchers have reported that decreasing the grain size increases the adsorption capacity of zeolite. Mery et al. [59] reported that a zeolite particle size of 0.5 mm had the highest ammonium adsorption capacity, which was 64% and 31% higher than that of zeolite with 1.0 and 2.0 mm grain sizes, respectively. Senila et al. [60], in their study, used clinoptilolite-rich zeolite with a similar chemical composition to that used in the present study. Using techniques applicable to the physicochemical and structural characterization of zeolites, they confirmed their porous morphology and substantial specific surface area, which favor both ammonium retention through ion exchange mechanisms and biofilm development due to enhanced surface roughness and porosity.

On the other hand, ammonium uptake was significantly reduced in coarse gravel, with q_e,m equal to only 29.34 mg/kg, six times lower than the uptake in zeolite. In this case, the pseudo-first-order model shows a better fit for the coarse-grained gravel data (Figure 2c, Table 1). It was confirmed that gravel has limited affinity for ammonium, as the adsorption stabilized more slowly and reached much lower maximum levels. The results for the kinetic adsorption of ammonium and orthophosphate are of great importance for the design of treatment systems such as CWs. Kinetic models help to clarify adsorption mechanisms, which depend on the physical and/or chemical characteristics of the adsorbent. The pseudo-second-order fit shows that the dominant process of ammonium adsorption on zeolite is chemisorption, in which ammonium ions are first adsorbed onto the zeolite surface and then into the adsorption sites located in the pores and channels of the zeolite [61].

According to the results, phosphate adsorption in the three porous media tested was much slower and less efficient than ammonium adsorption. The measured adsorption capacity at equilibrium (q_e,m) for CG-Z, FG-Z, and CG-G was 11.76, 12.35, and 6.44 mg/kg, respectively. As in the case of ammonium, the pseudo-second-order model showed a better fit compared to the pseudo-first-order model, as it achieved higher R² values and lower values of both the NOF and MAE compared to the corresponding values of the pseudo-first-order model. Moreover, the measured q_e,m values were very close to the q_e,c values calculated from the pseudo-second-order model (Table 1).

Zeolite is an aluminosilicate mineral and therefore has a negative surface charge and very low anion exchange capacity, showing low phosphorus adsorption. Phosphate removal occurs mainly through substituent exchange and surface complexation with calcium and aluminum in zeolites, in contrast to the exchange of ammonium cations [62]. The description of phosphate binding to numerous surface groups by Ji et al. [63] is supported by the heterogeneous adsorption sites. The low phosphate adsorption capacity of gravel without chemical alterations is confirmed by the difference in efficiency between zeolites and gravel, which is consistent with the findings of Xu et al. [64].

Figure 4 and Figure 5 for ammonium and phosphate, respectively, show that the Weber and Morris model (Equation (9)) fits the experimental data quite well if they are divided into two or more linear regions. The ammonium adsorption process on the three substrates (i.e., CG-Z, FG-Z and CG-G) is divided into two linear regions (Figure 4). The first region with the steepest slope (blue dots) represents adsorption controlled by external film diffusion, and the second region (orange triangles) corresponds to intra-particle diffusion into the porous structure of zeolite and gravel. As for the phosphate adsorption process on the three substrates, it is divided into three linear regions (Figure 5) [41,65]. The first region (blue dots) represents adsorption controlled by external film diffusion, the second region (orange triangles) is probably controlled by pore diffusion (or intra-particle diffusion), and the third region (green rectangles) is the equilibrium stage. Each plot (Figure 4 and Figure 5) shows the equations of the individual linear regions, which have the form of Equation (9), and the kinetic parameters of these equations are indicated in

Table 2. Intra-particle diffusion model parameters for adsorption of ammonium and phosphate onto various substrates.

Parameters		CG-Z	FG-Z	CG-G
Ammonium
Kid1	mg/(kg·h0.5)	17.58	16.61	5.45
C1	mg/kg	93.95	91.69	3.01
R12		0.98	0.93	0.80
Kid2	mg/(kg·h0.5)	0.52	0.82	0.01
C2	mg/kg	176.63	166.1	6.03
R22		0.66	0.69	0.12
Phosphate
Kid1	mg/(kg·h0.5)	1.66	1.37	1.12
C1	mg/kg	0.32	2.17	0.11
R12		0.98	0.78	0.99
Kid2	mg/(kg·h0.5)	0.99	0.43	0.36
C2	mg/kg	2.06	6.49	2.58
R22		0.99	0.99	0.98
Kid3	mg/(kg·h0.5)	0.08	0.01	0.01
C3	mg/kg	10.83	12.16	6.03
R32		0.79	1.0	0.12

.

The results show that the adsorption of both ammonium and phosphate occurs as a combination of surface film diffusion and pore diffusion, with the former having a steeper slope than the latter, as shown in Figure 4 and Figure 5, with the values of the constants k_id1 being larger than the corresponding values of the constants k_id2 (Table 2). This is in agreement with the results reported in previous studies [41,66]. In phase (I), approximately 97%, 96%, and 93% of ammonium and 55%, 68%, and 62% of orthophosphate were adsorbed by CG-Z, FG-Z, and CG-G, respectively, within the first 24 h (t^0.5 value of 4.9 h). This is attributed to the immediate utilization of the available adsorption sites on the adsorbent’s surface. Phase (II) can be attributed to the very slow diffusion of the adsorbates from the surface of the film into the pores, which are the least accessible adsorption sites.

3.3. Adsorption Isotherms

The adsorption isotherms obtained for the removal of ammonium and phosphate by various substrates (CG-Z, FG-Z and CG-G) are shown in Figure 6 and Figure 7 and

Table 3. Adsorption isotherm parameters of ammonium.

Substrate	T (°C)	Langmuir Isotherm			Freundlich Isotherm
		qm (mg/kg)	kL (L/mg)	R2	1/n	kF (mg/kg)	R2
CG-Z	10	275.02	0.548	0.89	0.26	118.03	0.74
	25	286.93	0.340	0.88	0.29	103.87	0.76
FG-Z	10	268.00	0.352	0.81	0.28	100.45	0.71
	25	285.71	0.142	0.91	0.41	63.03	0.83
CG-G	10	54.79	0.059	0.98	0.56	4.34	0.93
	25	48.33	0.068	0.93	0.40	7.31	0.88

and

Table 4. Adsorption isotherm parameters of phosphate.

Substrate	T (°C)	Langmuir Isotherm			Freundlich Isotherm
		qm (mg/kg)	kL (L/mg)	R2	1/n	kF (mg/kg)	R2
CG-Z	10	18.14	0.12	0.90	0.72	2.05	0.97
	25	23.71	0.14	0.90	0.63	3.34	0.96
FG-Z	10	24.15	0.08	0.98	0.69	2.12	0.99
	25	16.80	0.36	0.72	0.47	4.71	0.87
CG-G	10	12.81	0.43	0.87	0.40	4.20	0.96
	25	15.65	0.36	0.80	0.48	4.34	0.95

. The adsorption behavior was characterized by fitting the data to Langmuir and Freundlich isotherm models at different temperatures (10 °C and 25 °C) and initial concentrations.

The Langmuir model fit all substrates better for ammonium adsorption (Figure 6, Table 3), as shown by higher R² values (0.81–0.98), than the Freundlich model (0.71–0.93). The CG-Z substrate exhibited the highest theoretical maximum adsorption efficiency (qₘ) of 286.93 mg/kg (at 25 °C). This was followed by FG-Z, with an efficiency of 285.71 mg/kg (at 25 °C), and CG-G, with an efficiency of 54.79 mg/kg (at 10 °C) (Table 2). With an adsorption affinity of 0.548 L/mg at 10 °C, CG-Z had the greatest Langmuir constant (k_L), indicating increased ammonium binding to zeolite surfaces. CG-Z and FG-Z showed moderate increases in qₘ with increasing temperature from 10 °C to 25 °C, while CG-G showed a decrease in qₘ (Table 3). Many studies report that the Langmuir model best describes the adsorption of ammonium onto natural or modified zeolite [67,68,69], while others report that the Freundlich model has the best fit [55,70,71,72].

Regarding the phosphate adsorption isotherm, the Freundlich model showed higher R² values for all three tested substrates, indicating that this model fits the experimental data better than the Langmuir model (Table 4). In all cases, the 1/n value was less than 1, indicating that phosphate adsorption on the three substrates (i.e., CG-Z, FG-Z, and CG-G) is favorable. According to the literature, the optimal adsorption isotherm for phosphate adsorption on zeolite depends on its characteristics and may vary. Several studies report that the Freundlich model best describes phosphate adsorption on zeolite (natural or modified) [34,73], while others report that the Langmuir model has the best fit [62,74].

According to Sandoval et al. [75], smaller particles offer a greater specific surface area for microbial adhesion, and chemical interactions are supported by the particle size effect (fine-grained > coarse-grained zeolite). However, Ofiera et al.’s [76] findings on thermally enhanced zeolites are in contrast to the moderate temperature sensitivity of phosphate adsorption on fine zeolites, indicating the limitations of natural zeolites in high-temperature applications.

3.4. Adsorption Thermodynamics

The results of the thermodynamic study are presented in

Table 5. Thermodynamic parameters of ammonium and orthophosphate adsorption on coarse-grained zeolite (CG-Z), fine-grained zeolite (FG-Z) and coarse-grained gravel (CG-G).

Substrate	T (K)	Kd	ΔG (kJ/mol)	ΔH (kJ/mol)	ΔS (kJ/(mol K))
Ammonium
CG-Z	283	5.41	−4.00	0.54	0.016
	298	5.48	−4.18
FG-Z	283	4.84	−3.71	2.81	0.023
	298	5.14	−4.06
CG-G	283	1.10	−0.22	−16.79	−0.059
	298	0.77	0.66
Orthophosphate
CG-Z	283	0.94	0.15	6.69	0.023
	298	1.08	−0.20
FG-Z	283	0.87	0.34	4.34	0.014
	298	0.95	0.13
CG-G	283	0.56	1.36	20.09	0.066
		0.86	0.37

. Regarding ammonium adsorption, the ΔG value is negative for all substrates, indicating that the adsorption process is spontaneous. The positive ΔH values for zeolite substrates indicate that the adsorption process is endothermic and favored by increasing temperature, while for gravel, it is exothermic and favored by decreasing temperature. The positive value of ΔS indicates increased randomness at the solid–solution interface during adsorption. In other words, the ammonium ions on the adsorbent’s surface are in a more chaotic distribution compared to those in the aqueous solution. Regarding orthophosphate adsorption, the ΔH values are positive for all substrates, indicating that the adsorption process is endothermic. Comparing the results of the present work with those in the literature [33,35,48,77,78], it emerges that the effect of temperature on the adsorption of ammonium and orthophosphates and the nature of the adsorption process (endothermic or exothermic) depends both on the type and characteristics of the adsorbent and on the selected experimental conditions.

4. Conclusions

According to this study, natural zeolite—both coarse- and fine-grained—has a much greater capacity to adsorb ammonium and phosphate than gravel, which makes it a great option for CWs. Specifically, it achieved up to 96.49% ammonium removal and up to 12.81% phosphate removal, outperforming gravel by a wide margin. The adsorption of ammonium and phosphate on the three substrates studied followed the pseudo-second-order model, indicating chemisorption as the main removal mechanism, except for the adsorption of ammonium on coarse gravel, which followed the pseudo-first-order model. Phosphate adsorption fits the Freundlich model better, indicating multilayer adsorption on heterogeneous sites, but ammonium adsorption was best represented by the Langmuir model, indicating monolayer coverage on homogeneous surfaces, according to isotherm modeling. Zeolites performed better than gravel in the removal of both ammonium and phosphate, and the adsorption mechanisms were impacted by temperature, substrate type, and interactions specific to the pollutant. These results demonstrate the remarkable effectiveness and sustainability of natural zeolite as a substrate for the removal of both ammonium and phosphorus nutrients from wastewater in constructed wetlands. According to the results, zeolite-based systems can greatly improve wastewater treatment efficiency while remaining economically and ecologically friendly. However, the method should be tested using natural wastewater containing both nutrients. Furthermore, future studies should examine potential synergies between zeolite substrates and microbial communities and focus on long-term performance evaluations in full-scale operating wetland systems. In addition, although zeolite addition to CW substrates can promote plant–microbe interactions, enhance nitrogen removal, and increase microbial diversity, excessive addition may affect redox dynamics and substrate regeneration, requiring careful optimization.