Granular Natural Zeolites: Cost-Effective Adsorbents for the Removal of Ammonium from Drinking Water

Abstract

Increasing food demand has resulted in an ever increasing demand for nitrogen fertilizers. Ammonium is the main constituent of these fertilizers and is a threat to aquatic environments around the world. With a focus on the treatment of drinking water, the scope of this study was to investigate the influence of key parameters on the suitability of granular natural zeolites as adsorbents for ammonium. Sorption experiments were performed in artificial matrices by varying the grain size, contact time, ammonium concentration, pH, content of competing ions, and regeneration solutions used. Additionally, natural matrices and the point of zero charge (pzc) were investigated. With an initial ammonium concentration of 10 mgN/L, the grain size was shown to have no significant effect on the sorption efficiency (97–98%). The experimental data obtained was best described by the Langmuir adsorption model (R² = 0.99). Minor effects on sorption were observed at different pH values and in the presence of competing anions. In addition, the pH_PZC was determined to be between pH 6.24 and pH 6.47. Potassium ions were shown to be better than sodium ions for the regeneration of previously loaded zeolites, potassium is also the main competitor to ammonium sorption. The use of tap, bank filtrate, river, and groundwater matrices decreased the ammonium sorption capacity of granular natural zeolites by up to 8%. Based on our results, granular natural zeolites are promising cost-effective adsorbents for drinking water treatment, especially in threshold and developing countries.

1. Introduction

The increase in industrialization that has occurred since the early 18th century has resulted in a 16-fold increase in the living standard of the average world citizen [1]. Concomitantly, the world’s population has risen from 1 billion to 7.7 billion between 1800 and 2019 [2,3]. Inevitably, an increase in food production was necessary to support this population increase. To enable food production on such a tremendous scale, the use of fertilizers in agro-cultural applications is essential. The development of an energy-intensive ammonia synthesis method (10 kWh/kg NH₃) [4] by Fritz Haber and Carl Bosch in the early 20th-century led to nitrogen fertilizer production increasing by 776% between 1961 and 2014 [5].

The excessive use of nitrogen fertilizers has enormously accelerated the global nitrogen (N) cycle and has resulted in an omnipresent form of soil and water pollution [6,7,8,9]. Ammonium (NH₄⁺) is the main constituent of nitrogen fertilizers and, for this reason, is one of the principle concerns to address; other concerns include ammonification, septic systems, landfills, animal wastes, irrigation, or municipal and industrial wastewater [10]. Further compounding the issue, high N concentrations in surface and groundwater lead to additional concerns: increased oxygen demands, stimulation of eutrophication, soil acidification, and others. Broadly speaking, the examples of NH₄⁺ removal discussed in the literature can be divided into two types: removal from wastewater and removal from drinking water. These examples differ fundamentally in terms of their characteristics, particularly their pollutant loads.

The pollutant loads, including ammonium levels, tends to be much higher in wastewater applications: (1) generally 30 mgN/L in municipal wastewater [11,12]; (2) up to 1500 mgN/L in anaerobic digestion liquor from sludge treatment [13]; and (3) up to 6000 mgN/L in high strength ammonium industrial wastewater [14].

On the other hand, drinking water sources are often derived from groundwater and bank filtrate. They tend to exhibit lower pollutant loads due to various elimination processes (e.g., microbial conversion, adsorption) during soil passage. Usually, the natural NH₄⁺ concentration in aerobic groundwater and surface water is <0.2 mg/L. However, under anaerobic groundwater conditions, the concentration may increase more than tenfold [15]. For instance, a number of sample sites have been documented to contain concerning levels of ammonium: 14 mg/L in the shallow groundwater of the Jianghan Plain, China [16]; up to 28 mg/L in the bank filtrate of Reijerwaard along the river Nieuwe Maas, Netherlands [17]; up to 100 mg/L in groundwater systems in the metropolitan area in Hanoi, Vietnam [18,19]; as well as up to 390 mg/L in a coastal aquifer-aquitard system in the Pearl River Delta, China [20]. Recent investigations have also reported a significant correlation between high NH₄⁺ concentrations in groundwater and high contents of dissolved iron and manganese [21]. Considering that the world population’s food demands continue to increase, it is reasonable to assume that NH₄⁺ levels will also increase in various drinking water matrices.

In terms of human toxicity, NH₄⁺ only becomes harmful over 100 mg/kg body weight per day [15]. However, during drinking water treatment processes, the formation of toxic nitrite (NO₂⁻) and harmful chlorination side products, as well as a higher chlorine consumption, must be considered. Generally, NH₄⁺ represents a pollution indicator for drinking water and must not exceed specific threshold values following purification (e.g., 0.5 mg/L in Germany) [22]. Therefore, NH₄⁺ contaminated waters can be treated by various technologies and processes including: (1) activated sludge and fixed bed reactor processes; (2) air-stripping; (3) membrane processes; and (4) ion-exchange by zeolites [23,24].

When considering the treatment processes highlighted above, natural zeolites, particularly clinoptilolite, stand out as promising adsorbents for drinking water treatment. The use of zeolites is more economical because high energy expenditures are avoided by excluding aeration processes and because additional chemicals and high pressures are not required. Furthermore, the use of zeolites is sustainable; ammonia synthesis can be reduced since NH₄⁺-saturated particles can be repurposed as slow-release fertilizers in agriculture [25] or can be regenerated in a cost-effective manner using sodium solutions [26,27,28]. Another positive feature of natural zeolites is their abundant availability and their low cost. Approximately 1 million tons of natural zeolites were mined globally in 2020, with the major producers being China, South Korea, Indonesia, and Slovakia [29]. Mining costs are generally low and achieving the desired zeolite reduction ratio during the milling process is the main expense. For example, the price of untreated finer zeolite products is USD 50 to 120 per ton and twice as high as for granular products, at USD 30 to 70 per ton [30,31]. In summary, outstanding NH₄⁺ removal efficiencies, the convenience of application, their sustainable characteristics, and their broad range of applications means that natural zeolites hold tremendous promise as highly cost-effective adsorbents [32,33,34].

Zeolites are constructed by a three-dimensional tetrahedral structure of AlO₄ and SiO₄, covalently connected by an oxygen atom to form interconnected channels within the zeolites’ framework. The higher the AlO₄⁻ content, the higher the negative surface charge. This surface charge is compensated for by adsorbed cations such as Na⁺, K⁺, Ca²⁺, and Mg²⁺, which can be exchanged for NH₄⁺ with remarkable selectively [35].

Previous studies have investigated the removal of heavy metals (Zn, Cr, Pb, Cd, Cu, etc.) [36,37,38], organic contaminants [39], and high NH₄⁺ loads by natural zeolites during wastewater treatment [39]. Natural zeolites have also been applied to the removal of NH₄⁺, humic acid, Fe, and Mn from surface and groundwater matrices. Two similarities were identified by comparing both areas of study: (1) similar zeolite grain sizes (powder-like and fine particles); and (2) zeolites were investigated in artificial, wastewater, or surface and groundwater matrices. In general, the composition, modification and quantity of the used zeolite, the initial concentration of the contaminants, pH, temperature, and contact time were identified to have the strongest influence on the removal efficiencies for various contaminants.

Until now, there has been little research on the effect of different water matrices on the efficiency of NH₄⁺ removal using the same natural zeolite. Furthermore, to the best of our knowledge, granular natural zeolites have not been investigated for NH₄⁺ removal. Finally, a trend can be seen when combining sorption onto granular natural zeolites and biological treatment in flow-through fixed bed reactors to enhance NH₄⁺ removal [40,41,42]. From a process engineering viewpoint, the clogging of filters in flow-through fixed-bed columns is less likely to occur when using granular natural zeolites, thus improving filter runtime, and increasing intervals between backwashing.

The main objective of this study was to demonstrate the suitability of cost-saving granular natural zeolites (clinoptilolite) in terms of their NH₄⁺ removal efficiency and to compare them directly with the finer grain sizes usually used in drinking water treatment. In a departure from previous studies, multiple water matrices such as artificial, as well as surface and river water, groundwater, and bank filtrate were investigated. Various process parameters were investigated for future applications of larger zeolite grain sizes, by conducting a broad range of batch experiments. This investigation demonstrated that the use of granular natural zeolites is more cost-effective than the use of finer zeolites. By avoiding milling costs and combining this with high NH₄⁺ removal efficiencies and sustainable characteristics (e.g., N re-use for agriculture), granular natural zeolites are a highly promising route to improving the quality of drinking water, especially in threshold and developing countries.

2. Materials and Methods

2.1. Zeolite Characteristics

The natural zeolite (clinoptilolite) investigated, CLP85+, was supplied by Zeolith Umwelttechnik Berlin GmbH, Berlin, Germany. In

Table 1. Chemical composition and zeolite characteristics, data provided by supplier [43].

Composition	Value (%)	Characteristics
SiO2	65.00–71.30	Exchange capacity	1.2–1.5 mol/kg
Al2O3	11.50–13.10	Selectivity	NH4+ > K+ > Na+ > Ca2+ > Mg2+
CaO	2.70–5.20	Mean pore diameter	0.4 nm
K2O	2.20–3.40	Specific surface	30–60 m2/g
Fe2O3	0.70–1.90	Si/Al	4.80–5.40 (−)
MgO	0.60–1.20	Grain sizes	1–2.5 mm
Na2O	0.20–1.30		8–16 mm
TiO2	0.10–0.30		16–32 mm

, the chemical composition and general characteristics of the zeolite are shown. In order to remove particulate matter, salts, and excess adsorbed cations, the zeolites were washed with ultrapure water until the electrical conductivity was less than 10 µS/cm (water appeared clear by visual inspection). The zeolites were subsequently dried at 80 °C for 24 h before beginning the experiment.

2.2. Batch-Set-Up

Batch experiments to investigate the impact of various process parameters on the NH₄⁺ sorption capacity of zeolites were carried out in triplicate at 22 °C. Matrices used included: ultrapure water; tap water; surface water; groundwater; and bank filtrate. The initial NH₄⁺ concentration of 0.71 mM (12.8 mg/L; 10 mgN/L) for all experiments—except the investigations into ammonium load—was obtained using a 0.03 M NH₄Cl (analytical grade) stock solution. For investigating the influence of contact time, three zeolite grain sizes (1–2.5 mm, 8–16 mm, 16–32 mm) were used. All other experiments were conducted exclusively using zeolite granules with a grain size of 8–16 mm. To avoid microbial contamination, the water matrices used, as well as the batch set-ups, were autoclaved prior to experiments. The batch set-ups were performed using the bottle point method and were carried out in borosilicate flasks (Borosilicate 3.3 glass; VWR International GmbH; Dresden, Germany) with 100 g of zeolites per liter of water [44]. Flasks were stirred at 50 rpm (SM-30 orbital shaker; Edmund Bühler GmbH; Bodelshausen, Germany), and sampling was performed at defined intervals until equilibrium was reached. The samples were filtered through a 0.45 µm PET filter (CHROMAFIL^® Xtra PET-45/25; Macherey-Nagel GmbH and Co., KG; Düren, Germany), before determining the amount of residual NH₄⁺.

To calculate the NH₄⁺ sorption capacity (q_t(NH₄⁺)) and the normalized sorption capacity in equilibrium (normalized q_eq) of zeolites, and to determine the sorption efficiency E_Sorption (%), the following equations were used:

$${\mathrm{q}}_{\mathrm{t}}\left({{\mathrm{NH}}_4}^{+}\right)=\frac{{\mathrm{C}}_0 - {\mathrm{C}}_{\mathrm{t}}}{{\mathrm{m}}_{\mathrm{Z}}}·\mathrm{V}$$

(1)

$$\mathrm{normalized}·{\mathrm{q}}_{\mathrm{eq}}\left({{\mathrm{NH}}_4}^{+}\right)=\frac{{\mathrm{q}}_{\mathrm{Experiment}}}{{\mathrm{q}}_{\mathrm{Control}}}$$

(2)

$${\mathrm{E}}_{\mathrm{Sorption}}(\%)=(1-\frac{{\mathrm{q}}_{\mathrm{Experiment}}}{{\mathrm{q}}_{\mathrm{Control}}})·100\%$$

(3)

The sorption capacity q_t(NH₄⁺) (mgNH₄⁺/g_Z) describes the time-dependent amount of adsorbed NH₄⁺ ions per unit weight of zeolite (g_Z). C₀ and C_t are the initial and time-de-pendent NH₄⁺ concentrations (mg/L) in solution. m_Z (g_Z) and V (L) are the adsorbent mass and the volume of treated water, respectively. q_Experiment and q_Control are the sorption capacities of a specific experiment and of an ultrapure water control, respectively. By using an ultrapure water control, a mostly non-competitive water matrix should be provided. q_Experiment was normalized against q_Control to get the normalized q_eq(NH₄⁺) (−). Before and after an experiment, the pH, electrical conductivity, and water temperature were recorded.

2.2.1. Influence of Contact Time and Ammonium Load

The influence of contact time was investigated under time-dependent sampling from 5 min to 20,160 min (14 days). To analyze the sorption kinetics in more detail, the experimental data were fitted against pseudo-first order reaction, pseudo-second order reaction, Elovich, and intra-particle diffusion kinetic models using the following equations [45]:

$$\mathrm{Pseudo}\text{-}\mathrm{first} \mathrm{order} \mathrm{reaction}: \ln \frac{\left({\mathrm{q}}_{\mathrm{eq}} - {\mathrm{q}}_{\mathrm{t}}\right)}{{\mathrm{q}}_{\mathrm{eq}}}=-{\mathrm{k}}_1·\mathrm{t}$$

(4)

$$\mathrm{Pseudo}\text{-}\sec \mathrm{ond} \mathrm{order} \mathrm{reaction}: \raisebox{1ex}{$\mathrm{t}$}\!\left/ \!\raisebox{-1ex}{${\mathrm{q}}_{\mathrm{t}} $}\right.=\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{${\mathrm{k}}_2\ast {{\mathrm{q}}_{\mathrm{eq}}}^2$}\right.+\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{${\mathrm{q}}_{\mathrm{eq}}$}\right.$$

(5)

$$\mathrm{Elovich}: {\mathrm{q}}_{\mathrm{t}}=\left(\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$\mathrm{b}$}\right.\right)·\ln \left(\mathrm{a}·\mathrm{b}\right)+\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$\mathrm{b}$}\right.·\ln \left(\mathrm{t}\right)$$

(6)

$$\mathrm{Intra}\text{-}\mathrm{particle} \mathrm{diffusion}: {{\mathrm{q}}_{\mathrm{t}}={\mathrm{k}}_{\mathrm{i}}\ast \mathrm{t}}^{0.5}+\mathrm{C}$$

(7)

where q_eq (mgNH₄⁺/g_Z) is the amount of adsorbed NH₄⁺ ions per weight unit of zeolite at equilibrium: k₁ (1/min) and k₂ (g/mg·min) are the pseudo-first and second order rate constants and t is the contact time (min); a is the initial sorption rate (mg/g·min) and b describes the extent of surface coverage and activation energy (g/mg); k_i is the intra-particle diffusion rate constant (mg/min^0.5·g) and C is a constant related to the thickness of the boundary layer.

Additionally, the experimental data obtained was fitted to Langmuir and Freundlich adsorption models with an initial ammonium load of 1 to 5000 mgN/L. The Langmuir model (Equation (8)) supposes sorption at specific homogeneous sites within the adsorbent and assumes there is no interaction between adsorbate molecules. On the other hand, the Freundlich isotherm (Equation (9)) describes non-ideal sorption on heterogeneous surfaces and multi-layer sorption [44,46,47].

$$\mathrm{Langmuir} \mathrm{model}: {\mathrm{q}}_{\mathrm{eq}}=\frac{{\mathrm{q}}_{\max }\ast {\mathrm{K}}_{\mathrm{L}}\ast {\mathrm{c}}_{\mathrm{eq}}}{1+{\mathrm{K}}_{\mathrm{L}}\ast {\mathrm{c}}_{\mathrm{eq}}}$$

(8)

$$\mathrm{Freundlich} \mathrm{model}: {{\mathrm{q}}_{\mathrm{eq}}={\mathrm{K}}_{\mathrm{F}}\ast {\mathrm{C}}_{\mathrm{eq}}}^{1/\mathrm{n}}$$

(9)

q_max describes the maximum monomolecular sorption capacity (mg/g) and the constant K_L (L/mg) the affinity between adsorptive and adsorbents. K_F (L/g) characterizes the adsorption strength and n (−) determines the curve of the sorption isotherm.

2.2.2. Influence of pH

The influence of the solutions’ pH on the removal of NH₄⁺ using a natural zeolite as an adsorbent was evaluated under the conditions specified in

Table 2. Composition of the used 0.1 M buffer systems.

Buffer	Chemical A	Chemical B	A (mol/L)	B (mol/L)
pH 5	C2H4O2	C2H3NaO2	0.03	0.07
pH 6	Na2HPO4	NaH2PO4·H2O	0.01	0.09
pH 7	Na2HPO4	NaH2PO4·H2O	0.06	0.04
pH 8	Na2B4O7·10H2O	HCl	0.1	x
pH 9	NaH2PO4·H2O	HCl	0.1	x

^X adjusted by 0.1 M HCl.

. A 0.1 M HCl solution was used to adjust the pH for pH 8 and 9. The buffer systems for pH 8 and 9 contained the greatest molar amount of sodium (due to the chemicals used in Table 2), which affects the ionic strength of the solution. To balance the background ionic strength, a 0.1 M NaCl solution was added equally to the other buffer systems. Finally, the 0.1 M buffer systems were diluted with ultrapure water to achieve a buffer concentration of 0.01 mM for the experiment. To ensure buffer stability, the pH was measured regularly.

2.2.3. Point of Zero Charge Determination

The point of zero charge (PZC) was determined by both the mass titration and pH drift methods. Differing from [48], a 0.03 M NaCl electrolyte solution as well as a reduced shaking speed of 50 U/min were used. By varying the mass concentration of zeolites between 0 and 112 g/L in the electrolyte solution, the change in pH was recorded over 24 h. The final pH values are plotted against the mass of zeolite per unit volume. The pH_PZC is the pH that was achieved when the addition of further zeolite did not result in further changes in the pH. The pH drift method was performed by exposing the zeolites to the electrolyte solution and adjusting the pH between pH 2 to 9 by adding 0.1 M or 0.01 M HCl and NaOH solutions. After 24 h, the pH was measured again. The pH_PZC is the pH value that no longer changes.

2.2.4. Influence of Competing Cations and Anions

Alkali metals and alkaline earth metals are ubiquitous cations in natural water matrices. To estimate the competing effect of Na⁺, K⁺, Mg²⁺ and Ca²⁺ ions on the NH₄⁺ sorption capacity, concentrations of 0.7, 3.5, and 7 mM of each competing ion were investigated. To classify the possible effect of anions, the influence of NO₂⁻, NO₃⁻, SO₄²⁻, and HPO₄²⁻/H₂PO₄⁻, each at a molar concentration of 3.5 mM, was investigated. Stock solutions with concentrations of 0.1 M NaCl, KCl, MgCl₂·6H₂O, CaCl₂·6H₂O, NaNO₂, NaNO₃, NaH₂PO₄, and Na₂SO₄ were diluted to obtain appropriate anion concentrations. It was also necessary to balance the background ionic strength using a 10 mM Na⁺ solution for the anion experiments (due to the chemicals used). All experiments with competing ions were normalized to a control that contained NH₄⁺ in pure water.

2.2.5. Single and Multiple Regeneration

Zeolites have to be regenerated after a certain operating period. In this study, the zeolites were loaded with 10 mgN/L NH₄⁺ in a first step and thereafter regenerated using 0.01 M and 0.1 M K⁺/Na⁺ solutions. Following the initial loading step, the zeolites were washed with ultrapure water until the electrical conductivity was less than 10 µS/cm and dried at 80 °C for 24 h to remove residual NH₄⁺ from the surface and pores. After adding the regeneration solutions, time-dependent samples were taken and analyzed until equilibrium was reached. For multiple loading experiments, the zeolites were washed until the electrical conductivity was less than 10 µS/cm and dried at 80 °C for 24 h before another loading step was conducted. The cation concentrations before and after a loading step were also measured to identify any possible modifications of the zeolites’ chemical structure over time. The desorption efficiency can be calculated by the following Equation (10):

$${\mathrm{E}}_{\mathrm{Desorption}}(\%)=\left(1-\left(\frac{{\mathrm{q}}_{\mathrm{Z}}·{\mathrm{m}}_{\mathrm{Z}}-{\mathrm{c}}_{\mathrm{t}}·{\mathrm{V}}_{\mathrm{Reg}}}{{\mathrm{q}}_{\mathrm{Z}}-{\mathrm{m}}_{\mathrm{Z}}}\right)\right)\ast 100\%$$

(10)

where E_Desorption is the desorption efficiency expressed as a percentage. q_Z (mg NH₄⁺/g_Z) and m_Z (g) describe the loading capacity of the previous loading step and the mass of zeolites used, respectively. The time-dependent NH₄⁺ concentration (mg/L) and the volume of the regeneration solution (L) is described by c_t and V_Reg.

2.2.6. Influence of Natural Matrices

For the investigation of real water matrices, Elbe river water, bank filtrate, groundwater, and a 1:10 mix of Elbe river water and tap water were analyzed. The NH₄⁺ concentrations of the autoclaved water samples were measured and equal NH₄⁺ levels were achieved by spiking with a 0.03 M NH₄Cl stock solution. To estimate the effect of cations and the dissolved organic carbon (DOC) content on the loading capacity, the levels of both were measured before and after the experiment.

2.3. Analytical Methods

The standard parameters of pH, electrical conductivity, and temperature were recorded using a multimeter and sensors (Sentix^®41 pH electrode, TetraCon^®325 conductivity cell, Multi 340i multimeter; Xylem Analytics Germany Sales GmbH and Co., KG; Weilheilm, Germany). NH₄⁺ was measured photometrically (Varian Cary^® 50 UV-Vis Spectrophotomer; Agilent Technologies Corporation; Santa Clara, CA, USA) according to DIN 38 4606-E5-1. Two chromatography devices were used to measure the amount of cations (930 Compact IC; Methrom AG; Herisau, Switzerland) and anions (DionexTM ICS-6000; Fisher Scientific GmbH; Schwerte, Germany). DOC was determined by an analyzer (TOC-VCPN Analyzer; Shimazu Corporation, Kyoto, Japan).

3. Results and Discussion

3.1. Comparison of Different Zeolite Grain Sizes

3.1.1. Kinetic Evaluation

Initially, ammonium sorption onto zeolites was investigated at various time intervals for a 10 mgN/L NH₄⁺ starting concentration. The influence of different zeolite grain sizes on the time-dependent sorption capacity of the zeolite q_t(NH₄⁺) are shown in Figure 1. Additionally, the initial rates (Figure 1a) and final stages (Figure 1b) are highlighted.

NH₄⁺ removal from the matrix increased with exposure time until q_eq was reached, at the state of equilibrium and for all grain sizes, on day 14. During the first 5 min, and for each grain size tested, the rate of NH₄⁺ uptake by the zeolites was at its maximum. Similar results were reported by Taddeo et al. [25] and Kotoulas et al. [39], who found the maximum NH₄⁺ uptake rate occurred in the first 5–10 min. q_t(NH₄⁺) for the different grain sizes diverged progressively after 40 min (Figure 1a) and decreased considerably after 3 d (or 4320 min; Figure 1b) for all grain sizes. Until the sorption equilibrium was reached for all grain sizes, the finest zeolite particle exhibited a slightly higher q_t(NH₄⁺) than the larger grain sizes. Remarkably, q_t between the 1–2.5 mm and 8–16 mm grain sizes did not differ significantly (p < 0.02) beyond day 3. A steady q_t was reached after 6 days for small grain sizes (1–2.5 mm and 8–16 mm), while the largest size (16–32 mm) required up to 9 days.

The NH₄⁺ uptake rates determined can be ascribed to differences in the structures of the different zeolite grain sizes. As porous materials, zeolites are characterized by their external and internal surface areas, which interact with the surrounding solution and define the uptake rate for a given adsorptive—here, NH₄⁺. Adsorption is defined by the bulk transport and film diffusion rates during the initial loading stages (Figure 1a), which was almost the same for all grain sizes over the first 20 min.

The finest grain size presumably favors external surface sorption due to its higher specific surface area, which explains the faster sorption rate during the first day. Similarly, the decreased rate of q_t after day 6 can be explained by a reduction in the driving force and internal diffusion rate, due a reduced concentration gradient. For larger grain sizes, the greater internal diffusion rate after day 6 serves to equalize NH₄⁺ sorption rates in the long run (Figure 1b).

Compared to the findings of Taddeo et al. [25] (stirring speed: 150 rpm; t_eq: 2 h; zeolite particle size: 0.2–2.0 mm), the equilibration time required in our study was quite long, especially for the 16–32 mm grain size. Nevertheless, operating at a slower stirring speed (50 rpm) was necessary to avoid zeolite grinding that would result in falsified diffusion conditions (grinding leading to a reduction in particle size and increased surface area).

From a practical point of view, the sorption kinetics of the 16–32 mm grain size seem not to be sufficiently fast to recommend the use of the largest zeolite granules tested as a single component in flow-through fixed-bed columns. However, the comparable kinetics of the 8–16 mm and 1–2.5 mm grain sizes highlights the promise of granular natural zeolites (8–16 mm) for use in drinking water treatment and the potential to reduce milling costs, by at least a factor of two, compared to finer particulate and powder-like zeolites.

3.1.2. Application of Different Reaction Kinetic Models

For a further investigation of the sorption kinetics, the experimental data shown in Figure 1 were fitted to three different reaction kinetic models (see also Figure S1). For all grain sizes, the kinetics were more accurately fit to pseudo-second order reaction kinetic plots (R² = 1 for 1–2.5 mm, R² = 0.999 for 8–16 mm, R² = 0.998 for 16–32 mm). This outcome suggests that (1) the rate-limiting step is chemical sorption; and (2) the sorption rate is dependent on the sorption capacity of the zeolite and not on the concentration of NH₄⁺ in the solution. Initially, all sorption sites are empty, and the NH₄⁺ gradient is high. However, the zeolite becomes more and more saturated, reducing the NH₄⁺ concentration gradient and the rate of removal (from solution). This outcome is comparable with Tang et al. [49] and Mazloomi and Jalali [50]. However, using simplified reaction kinetic models and judging their suitability based only on the correlation coefficient (R²) can lead to an overestimation of how well the data fits the model. For example, the linearization of non-linear functions can result in false correlations by reducing the error factor with log or square root transformation [45]. On the other hand, it is well known that intraparticle diffusion is typically the rate-limiting step in porous materials. Furthermore, the first- and second-order reaction kinetic models are considered empirical due to certain limitations, such as disregarding (1) the intraparticle or film diffusion that occurs; or (2) other boundary conditions. Their application is quite common for practical studies, for example, in cases (1) with adsorbents with lower porosity; (2) where film diffusion is negligible; and (3) where rather slow surface reactions occur. First- and second-order reaction kinetic models are more commonly used because they are easier to implement than more complex and precise diffusion models [44]. Due to the reasons listed above, applying exact diffusion models (film, surface, pore and intraparticle diffusion), rather than the simplified models used in many studies, is recommended [44].

3.1.3. Zeolite Sorption Capacities under Equilibrium Conditions

As demonstrated in Figure 1, q_eq(NH₄⁺) reached a state of equilibrium after 14 days. There were negligible differences in the zeolite sorption capacities with 0.130, 0.128, and 0.127 mgNH₄⁺/g_Z, corresponding to NH₄⁺ sorption efficiencies of 98.4%, 97.8%, and 97.4%, for the 1–2.5 mm, 8–16 mm, and 16–32 mm grain sizes, respectively. Other studies observed similar sorption efficiencies for finer zeolite grain sizes [23,44,51,52,53,54]. This experiment showed that the total sorption capacity is most likely characterized by the exchangeable sites located in the zeolite’s defined inner structure and not on its external surface, hence the similar removal efficiencies. Therefore, increasing the grain size did not significantly decrease q_eq(NH₄⁺) for granular zeolites in this set-up. Furthermore, a possible temperature effect can be excluded on the basis of similar removal efficiencies being observed between 10 °C and 40 °C in Figure S2.

3.2. Application of Isothermal Models

To classify the sorption mechanisms that underlie ammonium removal by granular zeolites, our experimental equilibrium data was fitted to the most common isothermal models, Langmuir (Figure 2a) and Freundlich (Figure 2b), in Figure 2.

By comparing the linear plots of both sorption models it can be stated that the Langmuir isothermal model shows a better correlation—with an R² value of 0.992 compared to R² = 0.975 for the Freundlich model. A linear progression can be identified with subsequent cumulative saturation at higher concentrations.

Table 3. Experimental isotherm constants compared to other studies.

Zeolite	Grain Size (mm)	Si/Al (−)	c0(NH4+) (mg/L)	Langmuir			Freundlich
				KL (L/mg)	qmax (mg/L)	R2 (−)	KF (L/g)	n (−)	R2 (−)
CLP85+ (a)	8–16	4.8–5.4	1–6890	0.004	21.3	0.992	1.21	0.33	0.975
Slovakia (b)	<0.2	5.6	10–5000	0.006	33	0.99	1.84	0.36	0.97
China (c)	0.8–1.43	3.38	10–4000	0.009	14.3	0.993	0.98	0.36	0.973

^(a) this study, ^(b) [26], ^(c) [55].

compares the sorption properties of zeolites from different locations used in this and other studies.

The underlying sorption mechanism cannot simply be described as solely homogeneous or heterogeneous, nor can it be ascribed entirely to chemisorption or physisorption. Several studies propose the Langmuir isothermal model for a better correlation [26,54,55,56], while others described the Freundlich model as more convenient [25,39,57,58]. However, as Lin et al. and Chen et al. described, it is most likely that ion exchange and adsorption coexist while removing NH₄⁺ using clinoptilolite-like zeolites [55,59]. In their studies, the proportion of adsorption increased with an increase in the initial NH₄⁺ concentration (clearly at an initial NH₄⁺ concentration of 1000 mgN/L). According to both studies mentioned, the underlying sorption mechanism at 10 mgN/L NH₄⁺ (the concentration that we selected as being relevant for drinking water treatment) can be attributed to ion exchange. This assumption was strengthened by balancing the initial and end point ion equivalents of NH₄⁺, Na⁺, K⁺, Mg²⁺, and Ca²⁺ ions with an R² = 0.997 (see Figure S3).

The varying maximum sorption capacities q_max for zeolites from different origins, shown in Table 3, are notable. Theoretically, if the number of silicon atoms (Si) replaced by aluminum atoms (Al) increases, it enhances the number of potential adsorption sites for exchangeable cations—resulting in a higher adsorption capacity. However, the opposite is actually observed when studying the Si/Al-ratios presented in Table 3. Here, replacing Si with Al leads to distortions of linkages (Si-O-Al, Si-O-Si). These distortions are the result of alterations in: (1) bond angles and lengths; (2) distances between linkage groups; (3) electronic charge strength; and (4) density distribution in the zeolite framework [60,61,62,63]. Zeolites with low Si/Al-ratios tend to possess negatively charged sites that are closer together, resulting in stronger electrostatic attraction and greater selectivity [64]. Conversely, this leads to shorter distances between hydrated NH₄⁺ ions bonded via attractive electrostatic forces to the zeolite framework. Therefore, low Si/Al-ratios can favor electrostatic repulsion, which might explain the lower q_max at lower Si/Al-ratios observed in Table 3 [65].

These results support the findings of Section 3.1.3: the grain size of the natural zeolite under investigation does not affect the loading capacity of NH₄⁺ ions—compare the results for larger granules to the finer zeolite grain sizes in Table 3. An increase in the initial NH₄⁺ concentration, from 1 to 5000 mgN/L, reduced the sorption efficiency for the 8–16 mm zeolite grain size from 99% to 30% due to the limited sorption capacity (see also Figure S4). Remarkably, the sorption efficiency only declined from 99% to 95% over the range of concentrations we deemed relevant to drinking water (1–110 mgN/L NH₄⁺), further strengthening the case for granular natural zeolites.

In conclusion to this section, three factors should be considered when applying granular natural zeolites to the treatment of drinking water: (1) where the zeolites were mined—due to the different chemical compositions of natural zeolites; (2) finding a reasonable compromise of the Si/Al-ratio to achieve optimal NH₄⁺-removal efficiencies; and (3) the initial NH₄⁺ concentration.

Alternative Adsorbents for Ammonium Removal

Table 4. Alternative adsorbents used for NH₄⁺ removal, summarized from Huang et al. [34] and Gupta et al. [23].

Alternative Adsorbent	Product Name	Grain Size (mm)	c0(NH4+) (mg/L)	Capacity (mgNH4+/g)	Removal Efficiency (%)	Reference
Clinoptilolite	CLP85+	8–16	2–6890	0.02–20.29	99–30	This study
Other zeolites	Mesolite	NA x	50	55	95	[67]
	Synthetic zeolite	NA x	18	13	70	[68]
Polymeric ion exchanger	Purolite C 150 H	NA x	32–193	36	>65	[69]
	Dowex 50w-x8 and Purolite MN500	~0.18	32–257	51	NA x	[70]
Carbon-based adsorbents	Combination of PAC xx and PZ xxx	0.50–0.85	23	NAx	49	[71]
	Rice husk biochar	0.25–0.50	322–1800	51	84	[72]
Industrial wastes	Fly ash (thermally activated)	<0.80	67	7	84	[73]
	Red mud	~0.15	6–644	22	>50	[74]
Nanoparticles	Fe3O4	NA x	103–180	171	NA x	[66]
Biosorbents	Aerobic granules	NA x	300	24	NA x	[75]
	Municipal sludge	NA x	300	NA x	89	[76]

^x NA: not available; ^xx PAC: powdered activated carbon; ^xxx PZ: powdered zeolite.

shows a wide range of alternative adsorbents developed and used for NH₄⁺ removal and compares the results from Section 3.2.

The different boundary conditions of (1) initial NH₄⁺ concentration; (2) particle size; (3) adsorbent dosage; (4) chemical and physical adsorbent properties; (5) water matrix; and (6) water pH make a direct comparison difficult. Nevertheless, the used zeolite (clinoptilolite) shows comparable removal efficiencies (99% to 30%) but a lower sorption capacity (0.02–20.29 mgNH₄⁺/g_Z) than most of the demonstrated alternative adsorbents.

Other adsorbents show better sorption capacities; however, these adsorbents should meet certain criteria for their cost-effective use: (1) high removal capacity; (2) abundance; and (3) low cost.

For example, Fe₃O₄ nanoparticles have a far higher sorption capacity (171 mgNH₄⁺/g) [66] than the 8–16 mm granules (20.29 mgNH₄⁺/g) investigated in this study. However, it can be assumed that these nanoparticles are not abundant and low cost. Due to their fabrication and conditioning (chemically, thermally), the alternative adsorbents shown will not meet the criteria of (2) abundance and (3) low cost (as can be expected of the listed biosorbents).

3.3. Influence of pH

pH is an important parameter affecting (1) the net charge of an adsorbent; and (2) the dissociation equilibrium between the protonated NH₄⁺ and its conjugated base NH_3(aq). To identify the pH for maximal E_Sorption, the sorption of NH₄⁺ was analyzed between pH 5 and 9 at 24 h and 14 d time points (Figure 3). The E_Sorption for all samples was normalized against a control containing just NH₄⁺ (10 mgN/L) in ultrapure water without an additional buffer. To equalize the background ionic strength, NaCl was added to samples at pH 5, 6, and 7. To exclude the competing effect of Na⁺ ions, and to ensure that the influence of pH can explain different sorption efficiencies, a separate sample (control+NaCl) containing 10 mgN/L NH₄⁺ with an equal amount of Na⁺ ions in ultrapure water without an additional buffer was used.

Maximal E_Sorption, at 98.7%, was observed at pH 7 in the state of equilibrium after 14 days. The samples at pH 5 and pH 6 achieved E_Sorption values of 97.4% and 97.3%, respectively. Further pH increases reduced the sorption efficiency: 94.9% and 92.2% at pH 8 and pH 9, respectively. Therefore, the optimal pH range for NH₄⁺ adsorption, using the granular natural zeolite investigated, is located between pH 5 and pH 7. This range is well suited to the relevant pH range for drinking water applications. These results are also consistent with those reported in other studies [25,49,77].

The differences in E_Sorption can be explained by the protonation of NH₄⁺ ions. In aqueous environments, ammonium can be found as protonated NH₄⁺ or deprotonated NH_3(aq). As a function of temperature and salinity, and at pH values above 7, the NH₄⁺/NH_3(aq)-equilibrium shifts towards NH_3(aq) due to the deprotonation of NH₄⁺. The pKa of ammonium is 9.24; at pHs above this value, the predominant species is NH_3(aq) that can no longer be exchanged at the zeolites’ surface (when pH > pKa). Additionally, a significant difference in the adsorption kinetics between 24 h and 14 d, especially for pH 8 and pH 9, can be explained by the NH₄⁺/NH_3(aq)-equilibrium. Apparently, the driving force for diffusion at pH 8 and pH 9 is limited due to a higher proportion of non-exchangeable NH_3(aq). A pH lower than 5 was not investigated because it can lead to the collapse, or dissolving, of natural zeolites, as reported by Zhang et al. [78], Murayama et al. [79], and Leyva-Ramos et al. [80]. Furthermore, a pH value lower than 5 or higher than 9 is not relevant for most drinking water applications.

3.3.1. Point of Zero Charge

The adsorbent-dependent point of zero charge influences the net charge of adsorbents and their adsorption behavior with regard to several cations and anions. Figure 4 shows the determination of the point of zero charge by the pH drift method between pH 2 and 9 (Figure 4a) and the mass titration method with zeolite/solution ratios of w/v = 0–11% (Figure 4b).

Both determination methods indicate that the pH_PZC is located between pH 6.24 and pH 6.47. Within this pH range, the ratio between positive and negative adsorption sites is almost equal. Furthermore, the interaction between the zeolite cations or anions can be estimated. Taking this into account, the zeolite used this feature to adsorb a certain amount of anions at pH < pH_PZC because the surface is weakly positive charged.

The present study demonstrated the pH_PZC of a granular natural zeolite with a larger grain size for the first time. However, because the composition of natural zeolites can differ according to where they were mined from, it can be assumed that the pH_PZC can also vary—and several studies contain data that supports this [81,82,83,84]. Vaičiukynienė et al., used the pH drift method to determine the pH_PZC in the range of 5.0–5.4 [84]. Abatal et al., identified the pH_PZC around 6.2 of a natural zeolite from Mexico [81]. The pH_PZC of a Serbian (Vranska Banja) natural zeolite was in the range of pH 6.5 and 7.5 [83]. Ngeno et al. reported a pH_PZC around 7.5 for a natural zeolite from Kenya [82]. Finally, the water matrix under investigation, with its specific ionic strength, can also influence the location of the pH_PZC [85,86].

3.4. Competition from Cations and Anions

3.4.1. Cations

Alkali metals and alkaline earth metals are ubiquitous in surface and groundwater matrices and have the potential to compete with NH₄⁺ ions for the sorption sites found in natural zeolites, as Figure 5 shows.

Considering the standard deviations, the lowest concentrations of competing cations (of 0.7 mmol/L) had no significant effect on the normalized q_eq(NH₄⁺). At higher concentrations, the normalized q_eq(NH₄⁺) declined, especially for sodium and most notably for potassium. The largest effect observed, in the selected concentration range, occurred for potassium at a concentration of 7 mmol/L. Such a concentration of potassium reduced the zeolite’s normalized q_eq(NH₄⁺) by 8.2%. The normalized q_eq(NH₄⁺) in the presence of Mg²⁺ and Ca²⁺ appeared to plateau at 3.5 mmol/L, with no further effects at higher concentrations. In conclusion, the sorption of NH₄⁺ was most affected by the presence of K⁺, followed by Na⁺, Mg²⁺, and Ca²⁺. Thus, a selectivity order can be assembled as follows: K⁺ > Na⁺ > Mg²⁺ > Ca²⁺, which is generally in agreement with previous reported studies (although these can differ with respect to Mg²⁺ and Ca²⁺) [43,50].

The differing competing effects of the cations investigated can be interpreted by considering their physical characteristics and the dimensions of the framework of natural zeolites—shown in

Table 5. Channel dimensions and ring types of natural zeolites.

Channel Dimension (nm)		Ring Type
Inglezakis and Zorpas [87]	Margeta et al. [88]	Number	Arrangement
0.75·0.31	0.72·0.44	10	Horizontal
0.47·0.28	0.55·0.40	8	Vertical
0.46·0.36	0.47·0.41	8	Horizontal

and

Table 6. Physical characteristics of the investigated cations.

Cation	Charge Density (C/mm3)	Ionic Radius (nm)	Hydrated Radius (nm)
NH4+	12	0.148 (a)	0.331 (c)
K+	14	0.138 (b)	0.331 (c)
Na+	37	0.102 (b)	0.358 (c)
Ca2+	76	0.100 (b)	0.412 (c)
Mg2+	205	0.072 (b)	0.428 (c)

^(a) [89], ^(b) [90], ^(c) [91].

.

Three ring types with different channel dimensions make up the framework of natural zeolites. These Si/AlO₄ ring types provide different potential sorption/exchange sites for hydrated cations. Molecular sieving effects occur, which enable cations with smaller hydrated radii to diffuse into the pores and channels of natural zeolites. These sieving effects are determined by the hydrated radii—shown in Table 6—and the channel dimensions—shown in Table 5. Horizontal ten-membered rings define one channel that Na⁺ and Ca²⁺ can occupy and Mg²⁺ can also be exchanged in the center of this channel. Na⁺ and Ca²⁺ can both be exchanged in the horizontal eight-membered rings. However, only K⁺ and NH₄⁺, with identical hydrated radii, can diffuse into, and be exchanged in, vertical and horizontal eight-membered rings. Therefore, the dominance of K⁺ in the cation competition experiments can be explained by its ability to be exchanged within all ring types, much like NH₄⁺. In addition to the fact that diffusion limits Mg²⁺ and Ca²⁺ to specific adsorption/exchangeable sites, an electrostatic repulsion effect can also occur due to their comparatively higher charge density. Furthermore, it is assumed that divalent cations, such as Mg²⁺ and Ca²⁺, can saturate potential exchange sites on the zeolites’ outer surface much faster than monovalent cations (such as K⁺ and Na⁺) can. Potential CaCO₃ precipitation can be excluded in this study since the pH was clearly below the saturation limit for all molar concentrations investigated.

These findings indicate that K⁺ concentrations have to be considered when applying granular zeolites to the treatment of drinking water. However, as natural water matrices do not usually contain K⁺ concentrations as high as 7.0 mmol/L (273 mg/L), minor competing effects can generally be assumed (see also Section 3.6).

3.4.2. Anions

A range of anions can be found in surface and groundwaters, possible affecting the sorption of NH₄⁺ by natural zeolites. In the first instance, natural zeolites are known primarily as cation exchangers. Nevertheless, in Section 3.4.1, it was shown, theoretically, that natural zeolites could adsorb anions at protonated functional groups at pHs < pH_PZC. For a general confirmation the effects of NO₂⁻ (161 mg/L), NO₃⁻ (217 mg/L), PO₄³⁻ (332 mg/L), and SO₄²⁻ (336 mg/L), all at 3.5 mmol/L, were investigated. Note that these concentrations are not relevant to drinking water. All E_Sorption values were normalized against a control containing 0.71 mmol/L NH₄⁺ in ultrapure water (Figure 6). To equalize the background ionic strength, NaCl was added to specific anion samples.

To exclude a possible influence of Na⁺ ions and to ensure that the influence of different anions can explain different E_Sorption values, a further sample (control + NaCl) was used. This contained 0.71 mmol/L NH₄⁺ and an equal amount of Na⁺ ions in ultrapure water.

With regard to the normalized q_eq(NH₄⁺) values shown in Figure 6, the principal reduction of 17.4% can be attributed to the competing effect of Na⁺ (used as a counter ion). This effect is visible in the control+NaCl sample. Focusing on the mean values for normalized q_eq(NH₄⁺), a gradual decrease from NO₂⁻ (18.8%), to NO₃⁻ (19.8%), to HPO₄²⁻/H₂PO₄⁻ (21.2%), and to SO₄²⁻ (21.8%) can be observed. However, the overlapping standard deviations mean that a definite reduction order cannot be determined.

For granular natural zeolites, even the unrealistically high concentrations of the investigated anions had no significant effect on the sorption of NH₄⁺. Drinking water sources usually contain far lower concentrations of these anions and thus their influence on the treatment of drinking water by granular natural zeolites is likely to be negligible.

3.5. Regeneration

3.5.1. Single Regeneration

The total sorption capacity of the granular natural zeolites decreased gradually by 19% over three loading steps (see Figure S5), indicating the need for regeneration to ensure the longest possible product life cycle.

Firstly, single regeneration experiments provide a basic understanding of the desorption mechanisms and enable the optimization of regeneration conditions for applications in the field. In Section 3.4.1, Na⁺ and especially K⁺ were identified as the main competitors for NH₄⁺ sorption and therefore were selected for further regeneration studies. To this end, different molar concentrations, ranging from 1 mmol/L to 1000 mmol/L, of Na⁺ and K⁺ ions, were investigated for their ability to regenerate previously loaded zeolites (Figure 7).

It was shown that the regeneration efficiency using K⁺ in a single regeneration step was consistently higher than that achieved with Na⁺ for all molar concentrations tested. Both ions displayed an exponential increase in regeneration efficiency with increasing concentration. From a concentration of 100 mmol/L and upwards, the regeneration efficiencies diverged significantly for K⁺ and Na⁺ solutions. For example, even at a 1000 mmol/L Na⁺ solution, the zeolite cannot be regenerated as effectively (61.8%) as it can be using a K⁺ solution diluted tenfold more (72.6%). The previously loaded zeolites could only be completely regenerated using a 1000 mmol/L K⁺ solution. This outcome reflects the findings of Section 3.4.1 and agrees with the selectivity order and the competition for cation sorption sites in the pore and channel structures.

3.5.2. Multiple Regeneration

For applications in the field, it is necessary to conduct multiple regeneration steps in order to reuse zeolites following the application of a specific amount of NH₄⁺ or after a predetermined operating time. To investigate multiple loading and desorption steps, zeolites were first loaded in the presence of a 0.71 mM NH₄⁺ solution and afterward were regenerated using 100 and 1000 mmol/L K⁺ and Na⁺ solutions (Figure 8). The compositions of the exchanged cations after each loading step are shown in Figure S6.

The results shown in Figure 8 confirm the findings of Section 3.5.1 and illustrate that K⁺ solutions are more effective regenerative reagents than Na⁺ solutions. It was possible to completely regenerate the zeolites twice using a 1000 mmol/L K⁺ solution (d), the sorption capacity remained almost constant. It is interesting to note that a 100 mmol/L K⁺ solution initially regenerated the zeolite to 72.6% capacity but the second and third desorption step restored a full 100% capacity (c). A 1000 mmol/L Na⁺ solution could not fully regenerate the zeolites; however, the desorption efficiency increased from 61.3% to 93.5% over three regeneration cycles (b). Using a 100 mmol/L Na⁺ solution, the zeolites could be regenerated to 22.8% to 43.2% capacity over three cycles (a). Because an incomplete regeneration was achieved with Na⁺ solutions, the adsorption capacity was duly decreased.

To the best of our knowledge, this study is the first deep investigation of the sorption and desorption mechanisms of granular natural zeolites. The findings from Section 3.6 are of great interest when considering the application of granular natural zeolites to the treatment of drinking water, particularly in terms of the extended product life cycle that can be achieved by using K⁺ solutions to regenerate the zeolites. In most cases, the NH₄⁺ concentration in drinking water sources is lower than for wastewater applications. This suggests that an extended loading phase can be employed and that fewer regeneration cycles will be needed. Finally, operational costs can be reduced further still by optimizing the K⁺ concentration necessary for complete zeolite regeneration after the initial loading step.

3.6. Influence of Natural Water Matrices

In addition to the fundamental characterization of the sorption behavior of granular natural zeolites in artificial matrices, the investigation of natural water matrices is also of great interest. Therefore, the role of different cations and dissolved organic carbon (DOC), at concentrations typically found in natural water matrices, were investigated in order to evaluate their influence on the normalized q_eq(NH₄⁺) of granular natural zeolites. Here, tap water, Elbe river water, a 1:10 mix of Elbe river water and tap water, bank filtrate, and groundwater were used as matrices.

Figure 1 shows the influence of natural water matrices on the normalized q_eq(NH₄⁺) (Figure 9a) and the initial and final DOC-concentrations after zeolite treatment (Figure 9b).

A slight reduction in the normalized q_eq(NH₄⁺) of zeolites in natural water matrices, compared to the control, was observed (Figure 9a). The normalized q_eq(NH₄⁺) decreased by ~1% in tap water and by up to ~8% in Elbe water and groundwater matrices. In the discussion of Section 3.4.1, it was noted that the normalized q_eq(NH₄⁺) was expected to decrease in natural water matrices due to the increased competition from other cations. Most notably, the relatively high K⁺ concentration in the groundwater matrix was expected to reduce the normalized q_eq(NH₄⁺).

Table 7. Initial cation and DOC concentrations of the investigated natural water matrices.

Water Matrices	Cations (mmol/L)				DOC(mg/L)
	K+	Na+	Mg2+	Ca2+
Ulrapure water (control)	<CR x	<CR x	<CR x	<CR x	0.09
Tap water	0.04	0.38	0.30	0.41	1.77
Bank filtrate	0.25	0.91	1.72	0.78	1.77
Elbe-/tap water (1:10)	0.05	0.46	0.36	0.44	2.39
Groundwater	1.75	1.17	1.13	0.91	5.00
Elbe river water	0.11	0.15	0.94	0.56	6.59

^x below calibration range.

and Table S1 summarizes the initial cation and DOC concentrations of the different matrices used.

The DOC concentration in the control and tap water samples was observed to increase following zeolite treatment. Conversely, the DOC in natural water matrices (with a higher initial DOC) slightly decreased following treatment (Figure 9b). The increasing DOC can be explained by residual organic matter which was released from the washed zeolites and flasks. In general, zeolites can be used to remove DOC in a targeted manner [58] and in more complex applications [92]. It is assumed that the DOC fraction has a competition effect in the present experimental set-up as Elbe water had the highest content.

As stated in Section 3.3.1, the granular natural zeolite investigated had a pH_PZC between 6.24 and 6.47, indicating that the surface is mostly positively charged at pHs < pH_PZC. Since the pHs used in our experiments were above the pH_PZC, it can be assumed that the zeolites’ surfaces are mostly negatively charged and oppose the negative zeta potential of humic compounds in the DOC at pHs > 1.6. Nevertheless, a slight increase in the small number of positively charged sorption sites could be attributed to the removal of DOC that was observed. Furthermore, it is suspected that macromolecular humic compounds functioned to block pores, or as an additional molecular sieve (diffusion resistance) for NH₄⁺, resulting in minor removal efficiencies [44]. However, prior autoclaving in our experiment can exclude the following factors: (1) a subsequent buildup of biofilm on the surface or in the pores; and (2) microbial conversion of NH₄⁺.

This section showed that the normalized q_eq(NH₄⁺) of granular natural zeolites was only marginally reduced by the presence of different cations or the DOC content of natural water matrices.

4. Conclusions

This study is the first time that granular natural zeolites of different grain sizes were used in an investigation of how various process parameters influence ammonium removal efficiency. The conclusions of our study can be summarized as follows:

• Similar sorption kinetics and capacities were exhibited by 1–2.5 mm and 8–16 mm zeolite grain sizes. For this reason, 8–16 mm granular zeolites have the potential for applications in the field, will reduce milling costs, and thus can be used as cost-effective adsorbents;
• The Langmuir sorption model described the experimental data the most accurately, and ion exchange was revealed to be the primary mechanism behind the sorption of NH₄⁺ at concentrations relevant to drinking water treatment;
• A maximum sorption efficiency was identified at pH 7 and the point of zero charge was determined to be in the range of pH 6.24 to pH 6.47. A reduction in zeolite’s efficiency is not expected in the relevant pH range for drinking water applications;
• K⁺ concentrations have to be considered when applying granular zeolites to the treatment of drinking water as K⁺ ions were the most effective competitor for NH₄⁺ sorption to natural zeolites. Accordingly, potassium ions can be used to regenerate NH₄⁺-loaded zeolites cost-effectively;
• Different anions exhibited minor effects on the sorption capacity of granular zeolites and can be neglected in applications in the field;
• The natural water matrices investigated decreased the sorption capacity of granular zeolites by up to 8%. Natural zeolites were also shown to act as adsorbents for organic matter and we showed that the DOC-concentration in a water matrix affects NH₄⁺ sorption efficiency.

In future studies, we will validate the results reported here whilst studying sorption dynamics in a more practically relevant scenario—in-column systems. Nevertheless, this study highlights the potential of granular natural zeolites (8–16 mm) as highly-effective and cost-saving adsorbents for NH₄⁺ removal within the field of drinking water treatment. Furthermore, besides slower sorption kinetics, no disadvantages were uncovered for the largest grain size (16–32 mm) compared to the finer zeolite grain size (1–2.5 mm) tested.