Using raw and thermally modified fibrous clay minerals as low concentration NH4+–N adsorbents

Raw and modified fibrous clay minerals palygorskite (Pal) and sepiolite (Sep) were tested for their ability to remove ammonium from ammonium polluted water. Palygorskite and sepiolite underwent thermal treatment at 400°C (T-Pal and T-Sep respectively). Raw and thermally treated samples were characterized using XRD, SEM, BET, FTIR, TGA, zeta potential, and XRF. The techniques verified the effect of thermal treatment on sample structures and the enhancement of negative charge. Both raw and thermally activated materials were applied in batch kinetic experiments, and found to be efficient adsorbents in their raw forms, since Pal and Sep achieved 60 and 80% NH4+–N removal respectively within 20 min of contact for initial NH4+–N concentration of 4 mg/L. Similar removal rates were gained for other concentrations representative of contaminated aquifers that were examined, ranging from 1 to 8 mg/L. Results for the modified T-Pal and T-Sep minerals showed up to 20% higher removal rate. Saturation tests indicated the positive effect of thermal treatment on the minerals since T-Pal and T-Sep removal efficiency reached 85% and remained stable for 24 h. However, competitive ions in real water samples can influence the NH4+–N removal efficiency of the examined samples. At almost all the examined samples, the nonlinear Freundlich isotherm and linear pseudo-second kinetic models showed better fitted all examined samples thus indicating heterogeneous chemisorption.


Introduction
Ammonium (NH 4 + ) is an inorganic pollutant mostly found in wastewater discharges, landfill leachates, industrial sewage, and agricultural areas (Böhlke et al. 2006). It is commonly found in aquifers contaminated from one of the abovementioned pollution sources because it is a highly mobile contaminant (Böhlke et al. 2006). Ammonium can be extremely toxic since it can be transformed into NO 2 − or NO 3 − via the nitrification process under oxidant conditions and lead to eutrophication in aquatic ecosystems (Rozïc et al. 2000). Moreover, when ingested in quantities exceeding 100 mg/kg of body weight per day, it can cause the formation of ammonium salts, such as ammonium chloride, that are toxic to human health (WHO 2003). The European Union has set the permitted limit of NH 4 + in groundwater as 0.5 mg/L (WHO 2017). In general, the NH 4 + -N concentration in aquifers is < 0.2 mg/L; however, when anaerobic conditions dominate, concentrations of 1-5 mg/L can be reached (Voudouris et al. 2013;Rusydi et al. 2020).
A variety of methods has been applied for NH 4 + removal from water, such as biological treatment (Yang et al. 2019;Zeng et al. 2020), air stripping (Gui and Li 2019), and physicochemical techniques, like sorption mechanism (Hou et al. 2016;Vu et al. 2017). Zeolite has been widely used for NH 4 + retention since it is a low-cost, widely available, and Responsible Editor: Tito Roberto Cadaval Jr environmental-friendly material (Fu et al. 2020). Nevertheless, limited research has investigated the efficiency of clay minerals for NH 4 + removal (Alshameri et al. 2018) despite them being as abundant and low-cost as other industrial minerals.
Clay minerals consist of continuous silicon-oxygen tetrahedral and aluminum-oxygen octahedral sheets. Substitutions of Al +3 for Si +4 , and Mg +2 for Al +3 or other lower charge cations in tetrahedral and octahedral sheets respectively, are responsible for the negative charge of clay minerals (Brigatti et al. 2013). At the interlayer of the minerals, there are cations balancing the negative charge, which can be exchanged with other cations in solutions. The size and the valence of ions play a key role at the substitutions. In general, the smaller size ions are preferable because they can approach a surface closer, so a small monovalent overcomes a large divalent ion in a competitive case (Valiskó et al. 2007). This ability is denoted as the cation exchange capacity (CEC) of clay minerals (Brigatti et al. 2013). The high CEC and the permanent negative charge that clay minerals present render them promising adsorbents for various cationic pollutants (Lazaratou et al. 2020a).
Palygorskite (Pal) and sepiolite (Sep) are the only clay minerals classified as belonging to the fibrous group with a ribbon-like structure (Galán 1996). Both Pal and Sep are 2:1 clay minerals with high porosity, high specific surface area (SSA), and numerous inner nano-tunnels. Despite the fact that the fibrous clay minerals present relatively low CEC (Galán 1996), the aforementioned properties ensure both these clay minerals have high adsorption capacities. Moreover, Pal and Sep deposits exist worldwide, most of which are exploited for a variety of industrial applications (Murray et al. 2011), rendering their study in environmental applications as adsorbents crucial, since they could be an alternative and cost-effective choice, especially for the areas where the relative deposits exist (Kastritis et al. 2003;Yeniyol 2012).
Thermal treatment, or calcination, has been applied to Pal and Sep to improve their adsorption capacities since at temperatures above 100°C the water in their intercrystalline tunnels is removed causing modifications in their pore structure and SSA (Chen et al. 2011b). At temperatures of 350-1000°C, dehydroxylation takes place, which may reveal negatively charged sites and result in additional protonated surfaces (Zuo et al. 2017;Lazaratou et al. 2020b). This physical modification outweighs other modification techniques in terms of operational simplicity and cost.
So far, many studies have been conducted for cationic pollutant removal by palygorskite or sepiolite samples Wang et al. 2019). However, in limited references, these minerals were applied for ammonium removal (Balci and Dinçel 2002;Alshameri et al. 2018;Gianni et al. 2021). Recently, Alshameri et al. (2018) and Gianni et al. (2021) evaluated the NH 4 + removal efficiency (R%) of raw palygorskite and sepiolite, as well as raw and Na-, Ca-, and H-enriched palygorskite respectively. Raw palygorskite and sepiolite reached almost 50% and 40% of 10 mg NH 4 + /L respectively (Alshameri et al. 2018), whereas Na-palygorskite achieved almost 99% removal of 1 mg NH 4 + /L (Gianni et al.2021). Nevertheless, both studies reported that with initial concentration increase, the R% is decreasing, while Ca-and H-treated samples reached approximately 30% removal independently the initial NH 4 + concentration. Also, chemical treatment (salt enrichment or acid treatment) of clay minerals may be insufficient and despite methods being low-cost, it is difficult to scale up for industrial purposes.
This study evaluates the effectiveness of raw and thermally treated palygorskite and sepiolite minerals as costeffective NH 4 + -N adsorbents. Experiments focused on the removal of low concentrations of NH 4 + -N that are commonly found in contaminated aquifers, as few studies were concentrated on NH 4 + -N removal from groundwater. Moreover, thermal treatment has not been previously applied to fibrous clay minerals intended for ammonium removal. To determine the optimum operational conditions, all samples were mineralogically characterized with XRD, SEM, BET, N 2 sorption-desorption isotherms, FTIR, TGA, zeta potential, and XRF methods, while batch kinetic experiments were conducted with varying adsorbent dosages, initial NH 4 + -N concentrations, contact time, and pH values. Saturation tests were carried out to examine the potential reusability of the adsorbents, as well as the adsorption under real water sample conditions. Isotherms and kinetic models were applied to determine the nature of the adsorption process.

Samples preparation and thermal treatment
The Pal sample was supplied from Geohellas S.A. (Ventzia basin Grevena, Greece) while Sep was collected from Solomos village (Korinthos, Greece). The raw samples were washed with distilled water and dried at 40°C for 48 h then sieved until powder diffraction was obtained (< 50 μm). For the thermal treatment, 20 g of each powdered sample was treated at 400°C in a controlled muffled oven for 2 h and then cooled at room temperature in a desiccator. The temperature of 400°C was selected since it is the minimum where both coordinated water loss and dehydroxylation take place within the samples, thus, ensuring effective structural changes and cost effectiveness (Perraki and Orfanoudaki 2008;Chen et al. 2011b).

Characterization methods
Before any characterization method, the raw and thermally treated Pal and Sep samples were dried at 50°C for 24 h to maintain the representative conditions applied to the samples before adsorption batch tests. X-ray diffraction (XRD) patterns were obtained for the samples in a 2θ range of 2° to 60° and at a scanning rate of 2°/min, using XRD Bruker D8 advance diffractometer, with Ni-filtered CuK α radiation (λ = 1.5418 Ǻ). For the semi-quantitative analysis of the clay mineral composition, the Area method was used according to Brindley and Brown (1980). Their typical morphological characteristics were verified with scanning electron microscopy (SEM), using a SEM LEO SUPRA 35VP. The N 2 adsorption-desorption isothermal tests were carried out at 77K on the samples that were previously degassed at 100°C for 3 h, using Micromeritics Tristar 3000 analyzer which is equipped with a SmartPrep degasser. From these isotherms, the Brunauner-Emmet-Teller (BET) surface area, pore size, and pore distribution were determined for all the samples. Fourier transform infrared spectroscopy (FTIR) spectra were obtained using FTIR spectrophotometer Spectrum RXI (Perkin Elmer) at room temperature. The samples were prepared by mixing 0.1 mg of Pal, Sep, T-Pal, and T-Sep with KBr, and then were pressed until pellets were formed. The spectra were collected over 12 scans in the wavenumbers range from 400 to 4000 cm −1 , and were analyzed using Spectrum v5.3.1 software. Thermogravimetric analysis (TGA) was conducted on a Perkin Elmer Simultaneous Thermal Analyzer STA6000 controlled by Pyris Manager Software, using nitrogen as a purging gas. For each run, a weighted amount of sample powder (≅10 mg) was loaded on a ceramic pan and heated from 40 to 600°C, at a heating rate of 10°C/ min. Before each measurement, all samples were subjected to isothermal heating at 40°C for 30 min to eliminate the adsorbed water on their surface from their exposure to ambient atmosphere, simulating the drying process carried out before each case of their characterization. The X-ray florescence (XRF) measurements of the major (SiO 2 , Al 2 O 3 , CaO, MgO, MnO, Fe 2 O 3 , K 2 O, Na 2 O, P 2 O 5 , TiO 2 ) elements were performed. An amount of 1.8 g of dried ground sample was mixed with 0.2 g of wax (acting as a binder) and was pressed on a base of boric acid to a circular powder pellet of 3.2 cm in diameter. Analyses were performed with a RIGAKU ZSX PRIMUS II spectrometer, which was equipped with a Rhanode running at 4kW, for major and trace element analysis. The spectrometer was equipped with the diffracting crystals: LIF (200), LIF (220), PET, Ge, RX-25, RX-61, RX-40, and RX-75. The samples' zeta potential was determined by a Zetasizer, Nano ZS (Malvern, UK). For the measurements, dilute suspensions of various pH values (4-11) and standard ionic strength (0.01M KNO 3 ) were used. The zeta potential was reported as the mean of two measurements, and each measurement was the sum of 14 correlograms and fitting procedures.

Batch kinetic experiments
A series of batch kinetic experiments was conducted for raw Pal,raw Sep,0.8,1.6,and 4.0 g in 200 ml solution), initial NH 4 + -N concentrations (1, 2, 4, 6, and 8 mg/L), and pH values (2, 3, 5, 7, 9, and 11). Solution pH was adjusted using either H 2 SO 4 or NaOH for acidic and basic values, respectively. The batch tests were conducted in 200-ml conical flasks, while for the pH experiment, the flasks were covered with parafilm. The air volume of the flasks is less than 10 ml, so the ammonium removal by evaporation for pH > 9 is minimized and could be ignored (Vu et al. 2017). The standard NH 4 + -N solutions were prepared by dilution of NH 4 Cl in deionized water, at standard ionic strength I= 0.1 M using KClO 4 . The adsorption process was carried out using the jar tester VELP Scientifica JLT6 at 210 rpm. Samples were collected at different time intervals (2.5, 5, 10, 15, 20, 30, and 40 min) and were centrifuged at 5500 rpm for 3 min. The supernatant was filtrated through Whatman filters (0.45 μm) to remove the finest suspended particles. The final removal efficiency was determined according to Eq. 1: where C 0 is the initial NH 4 + -N concentration, and C e is the NH 4 + -N concentration after adsorption in equilibrium.

Saturation test
For the saturation test, the adsorbents were left in contact with 4 mg/L NH 4 + -N for 24 h under constant stirring. At the end of 24 h, samples were selected and centrifuged, and the supernatant was filtered through Whatman filters (0.45 μm). NH 4 + -N concentration of the filtered supernatant was measured according to the analytical methods described in the "Analytical methods" section. The solid phase of each adsorbent was selected and added back at the beaker. Following, fresh 4 mg/L NH 4 + -N was added for a new 24-h adsorption cycle, and new samples were selected. The procedure was repeated until each adsorbent was saturated and no further NH 4 + -N could be adsorbed.

Batch study on a real water system
In these batch series, the optimal adsorbents' dosages were applied in 200 ml of University of Patras tap water. ΝΗ 4 + -Ν concentration was artificially added to the real water sample in accordance with the other batch tests. Samples were • 100 then treated as described above and NH 4 + -N was measured according to the analytical methods described in "Analytical methods" section. The physicochemical characteristics of the water sample are described in Table 1.

Analytical methods
NH 4 + -N concentrations were measured using a UV-VIS spectrophotometer Hach Lange DR 5000 at 625 nm according to the modified salicylate method (Verdouw et al. 1978). Each sample was reacted with 6% sodium hypochlorite solution and salicylate/catalyst solution (sodium salicylate 10%, sodium nitroferricyanide 0.04% and sodium hydroxide 0.5%). ΝΗ 4 + -Ν concentrations were measured after 10-min color development. All experiments were conducted in duplicate.

Isotherm models
Data from the adsorption experiments were fitted in Langmuir and Freundlich isotherms to determine adsorbent surface sites-adsorbate ion relationship (Aydın Temel and Kuleyin 2016) according to the following equations: where q e is the amount of exchanged ions (mg/g), C 0 and C e are the initial and equilibrium NH 4 + -N concentrations in solution (mg /L), respectively, V is the solution volume (L), and m is the adsorbent weight (g), where constant K L is the C e /q e ratio vs. C e variation, where q max is used, expressing the NH 4 + -N maximum uptake. The fit to the Langmuir linear (Equation 3) and (2) (3) C * e q * e = 1 q * ma * x K * L + 1 q * max c * e (4) q * e= q * m K * L C * e 1+K * L C * e non-linear isotherm (Equation 4) indicates the nature of the monolayer adsorption (Aydın Temel and Kuleyin 2016).
The Freundlich isotherm expresses heterogeneous adsorption surfaces with unequal active sites and energies of adsorption (Yagub et al. 2014) and can be expressed in linear (Equation 5) and non-linear (Equation 6) as: where q e is the number of exchanged ions (mg/g), C e is the equilibrium NH 4 + -N concentrations in solution (mg/L), K F is the adsorbent capacity, and n is the Freundlich constant. When 1/n is 0 < 1/n < 1, adsorption is considered favorable, when 1/n = 1 adsorption is linear and irreversible, and when 1/n > 1 adsorption is a chemical process and unfavorable. The value of 1/n < 1 indicates the adsorption process is physical (Aydın Temel and Kuleyin 2016).

Adsorption kinetic models
The adsorption rate of NH 4 + -N on Pal, Sep, T-Pal, and T-Sep can be estimated via kinetic model application, as well as the optimal adsorption mechanism and adsorbent surficial characteristics can be approached (Karri et al. 2017). Specifically, the pseudo-first-order kinetic model focused on pollutant adsorption mechanisms and the pseudo-secondorder kinetic model was applied to predict the chemisorption of NH 4 + -N onto the tested adsorbents. The linearized forms of the kinetic models are expressed by Eqs. 7 and 8, respectively. and the non-linearized at the equations respectively where q t is the amount of adsorbed pollutant at time t (mg/g), and k 1 (1/min) and k 2 (g/mg min) are the rate constants of NH 4 + -N adsorption for the pseudo-first and second-order kinetic models, respectively (Shahwan 2014). The k 1 value can be determined from the slope of the linear plot of ln(q e −q t ) vs t, and k 2 from the intercept of the linear plot of t/q t vs t. Although both models indicate the adsorption mechanism through time, neither considers the diffusion where q is the adsorbate amount at time t and k id is the intraparticle diffusion constant (mg/g min −1/2 ). The k id parameter can be calculated from the slope of the q t vs t 0.5 linear plot, while I is the intercept of the vertical axis. If the Weber-Morris plot is linear, I=0 and intraparticle diffusion is the rate limiting step, but when I >0 two steps take place, firstly the film, followed by the intraparticle diffusion as the rate limiting steps (Svilović et al. 2010).

Determination of error functions and coefficients
The parameters of each nonlinear isotherm or kinetic model are resulted by linearizing the non-linear model expressions and least squares fit (Genethliou et al. 2021). In order to quantify the deviation and measure the uncertainty in error distribution, the coefficient of determination (R 2 ), the sum of squares of errors, and the root means square were determined as it is expressed in Eqs. 12-14 respectively.
where i is the experimental run, q etheor is the theoretically evaluated equilibrium values from the model (mg/g), and q emeas (mg/g) is the experimental equilibrium data (mg/g).

Results and discussion
Adsorbent characteristics and proposed NH 4 + -N removal mechanism

XRD
The XRD pattern of palygorskite sample (Pal) was characterized by Pal reflections at 2θ° values 8.3°, 20°, 27°, and 34°, rendering palygorskite the dominant mineralogical phase (96%) at Pal sample. The 2θ° reflection at 6° indicated the predicted occurrence of saponite as impurity (4%), since this mineral coexists with Pal in the Ventzia basin deposit (Fig. 1a). Differences were shown at Pal reflections after thermal treatment (T-Pal) (Fig. 1b), where the first characteristic Pal reflection intensity at 8.3° was sharply decreased, in contrast to the extended reflection of saponite. Similar results were observed by Yan et al. (2012), who recorded a decreased Pal reflection at 8.3°, while a new reflection at 30° was formed by dehydration and structural rearrangement. After NH 4 + -N adsorption, both Pal (Pal NH 4 + -N) and T-Pal (T-Pal NH 4 + -N) samples preserved the same mineral phases that were described above, since the existed main palygorskite reflections and saponite impurity were observed ( Fig. 1a,b). However, at Pal NH 4 + -N and T-Pal NH 4 + -N samples, the reflections attributed to palygorskite are decreased, as well as a slight peak shifting was observed at the basal reflections of 20° and 35° (Fig. 1a, b). The intensity change of the reflections highlights the surficial interaction between adsorbent-adsorbate (Papoulis et al. 2019), whereas the peak shifting can be attributed either to the ion exchange that took place or to the surficial bonding that may affect the crystallinity. Moreover, even the small presence of saponite may influence the surficial interactions with NH 4 + , since saponite consists of multiple OH − groups on its surface (Zhou et al. 2019).
At the XRD pattern of the sepiolite sample (Sep), the reflections of all other minerals were absent apart from an impurity of calcite at the reflection at 30° in content of 2% (Fig. 1c). The dominant sepiolite's presence was verified by the typical Sep reflections at 7.2°, 20°, and 35°, rendering sepiolite the main mineralogical phase (98%). The impact of thermal treatment (T-Sep) on Sep structure is revealed, since two new reflections appear at 7.3° and 11.04° (Fig. 1d) due to the formation of sepiolite anhydrite that occurs at temperatures above 350°C (Perraki and Orfanoudaki 2008). After NH 4 + -N adsorption, Sep (Sep NH 4 + -N) and T-Sep (T-Sep NH 4 + -N) samples presented opposed behavior from Pal NH 4 + -N and T-Pal NH 4 + -N samples, since there was no shifting occurred, while the main reflection at 7.2° in both samples was steeply increased (Fig. 1c, d). Potentially, the Sep and T-Sep purity led to enhanced interactions of the samples' pores with NH 4 + , contributing to the sharply increased reflections (Marler et al. 1996). This verifies the ammonium interaction with the inside of Sep or T-Sep structure without altering the crystal structure of the samples. Similar behavior was reported in the study of Alshameri et al. (2018) for vermiculite as ammonium adsorbent. Nevertheless, no mineralogical alterations were observed.

SEM
SEM images revealed the morphology of the Pal and Sep samples before and after thermal treatment. The fibrous morphology of Pal and Sep were verified (Fig. 2a, 2b respectively), with fiber lengths ranging from 250 nm to 1 μm. The effect of thermal treatment at 400°C was significant in both samples (Fig. 2c, d). Specifically, T-Pal fibers (Fig. 2c) were strongly agglomerated and their length decreased after thermal treatment as a result of the total loss of coordinated water molecules (Bu et al. 2011;Xavier et al. 2016). T-Sep also presented similar characteristics (Fig. 2d) with shorter fibers and agglomeration due to the loss of water molecules (Perraki and Orfanoudaki 2008;Miura et al. 2012).

BET surface area and pore size distribution
The N 2 adsorption-desorption isotherms for raw and thermally treated clay minerals are presented in Fig. 3. Based on the IUPAC classification, the isotherms were classified as Type IV with H 3 -type hysteresis loop, indicating the dominance of mesoporosity but also the presence of micropores (Sing et al. 1984). The overlap at P/P 0 < 0.4 verified the microporosity, while the limited loop at 0.7 < P/P 0 < 0.9 is typical of low degree mesoporosity (Cases et al. 1991;Wang et al. 2016), which may be attributed to aggregates or capillary condensation of typical H 3 -type loop and isotherm Type IV, respectively (Sing et al. 1984). After thermal treatment at 400°C, the loop at 0.6 < P/P 0 < 0.9 steepened. This was especially noticeable in T-Sep, since Sep is more fragile than palygorskite due to its Mg-rich composition and the size of its structural microchannels (Myriam et al. 1998). Up to 350°C, the loss of Sep first coordinated water causes reversible inner channel folding, but from 400 o C the folding is irreversible, with Sep preserving its structure (Myriam et al. 1998). The SSA values, average pore diameters or widths, and the total pore volumes of the Pal, Sep, T-Sep, and T-Pal minerals were determined as well ( Table 2). The thermally treated samples showed reduction in SSA as the zeolitic water, and part of the coordinated water or magnesiumcoordinated water respectively is removed during treatment (Perraki and Orfanoudaki 2008;Chen et al. 2011a). This removal of Mg +2 within the T-Sep structure, as well  as the zeolitic and coordinated water, may cause the reduction in SSA, as was also reported at the studies of Balci (1999) and Miura et al. (2012), whereas the pore volume was slightly decreased, potentially due to the rearrangements after Mg 2+ collapse. However, the average pore width was increased, which could be attributed to the structural rearrangement, as well as the secondary slight occurrence of microporosity resulted from the dehydration of silanol groups which may also explain the more intense loop observed after calcination in Fig. 3 and Fig. 4 (Balci 1999). T-Pal SSA was also decreased, due to the water loss below surface, as well as to the dihydroxylation of silanol groups after the removal of coordinated water molecules at the octahedral sheets (Xavier et al. 2016), as well as the pore size and volume of T-Pal due to the collapse of Pal nano tunnels and the condensation of silanol groups as can be observed in Fig. 4 (Chen et al. 2011a). Moreover, in Figure 4, it can be slightly observed the secondary microporous occurrence at both T-samples, highlighting the dominance of mesoporous which can characterize the main provider of adsorption sites.

FTIR
The FTIR spectra of raw and thermally modified Pal and Sep before and after NH 4 + -N adsorption can be observed in Fig. 4 a-d. Pal sample (Fig. 5a) presented typical bands of Si-O stretching, slightly shifted, at 469 cm −1 , 1024 cm −1 , 1170 cm −1 , and 1655 cm −1 (Madejová et al. 2017), whereas the band at 1170 cm −1 can be Si-O bond that connects two inverse SiO 4 of palygorskite structure, and the 1200-cm −1 band is typical of the ribbon structure (Blanco et al. 1989;Yan et al. 2012). Saponite impurity can be verified based on the band at 650 cm −1 (Lainé et al. 2017). At the range of 3610 to ~3200 cm −1 , characteristic bands of OH − from zeolitic water groups coordinated with structural Mg are shown (Xavier et al. 2016). The FTIR data verified XRD results that thermal treatment did not influence the mineralogical phase of palygorskite, since the same bands with the raw sample are maintained (Fig. 5b). Moreover, the vibration of the band at 1170 cm −1 in Fig. 4 b is steeply decreased after thermal treatment and a new band at 882 cm −1 appeared, similar to the observation of Yan et al. (2012) for the analogous band at 1196 cm −1 and at 885 cm −1 respectively. Vibration decrease and shifting at ~3400 to ~3600 cm −1 that was followed can be attributed to a partial loss of bound water after the thermal treatment (Fig. 5b). The total loss of water could be achieved at temperature higher than 700°C (Xavier et al. 2016). After NH 4 + -N adsorption on Pal (Fig. 5a), an intense band at 1380 cm −1 can be attributed to the newly formed N-H bond, rendering the ammonium removal on Pal by chemisorption (He et al. 2016).
The FTIR spectra of Sep sample (Fig. 5c) presented bands at 463 cm −1 , 683cm −1 , 830 cm −1 , 889cm −1 , and 1020 cm −1 that are referred to Si-O-Si bonds, whereas the bands at 1659 cm −1 and 3419 cm −1 can be attributed to zeolitic water molecules. In addition, the bands at 441cm −1 and 3568 cm −1 are representative of Si-O-Mg and surficial Mg-OH bonds of sepiolite respectively (Perraki and Orfanoudaki 2008). After thermal treatment at 400°C, the Mg-bond bands are steeply decreased (Fig. 5d), inducing significant shifting at Si-O-Si band from 463 to 474 cm −1 . These results come in agreement with the BET analysis where Sep structural rearrangement and Mg removal were referred. After NH 4 + -N adsorption, a slight shifting was observed at the band areas 1260-1390 cm −1 and 616-730 cm −1 , whereas band differences occurred at the area of 1380-1390 cm −1 . However, there is not an intense new band, typical of N-H bonding, because the bands of the sepiolite mineral may overlap these bands. The slight occurred band shifting could be attributed to interactions taking place on the surface which, however, are not as intense as it was observed at Pal and T-Pal samples.

TGA
TGA curves of the raw and thermally treated clay mineral samples are shown in Figure 6. Regarding the raw clay minerals and, at first, the Pal sample, three distinct weight loss steps can be observed in the TGA curve for the studied range of temperatures (12% total weight loss). The first step is determined at about 60°C and has to do with the elimination of the interparticle adsorbed water; the second step is found at about 170°C and is assigned to the thermal dehydration of the sample, while an extra weight loss step appears at a higher temperature region (340-420°C) (Frost and Ding 2003). Two considerations can be used for this step according to the literature. The first one is attributed to the Al, and the second one (more likely) to irreversible dehydration of residual bound water. On the other hand, regarding the Sep sample, three weight loss steps may also be detected for the studied range of temperatures (8% total weight loss). The first step takes place at about 50°C and is attributed to the loss of remaining adsorbed water on the sample surface; the second step is determined just above 240°C and corresponds to the loss of hydration water, while a third step could also be considered at the temperature region 450-500°C and has to do with the loss of coordination water (Frost and Ding 2003). Regarding the thermally treated clay minerals samples, in these cases, the weight loss was found quite reduced compared to the corresponding raw samples (total weight loss was about 8% for the T-Pal sample and 4% for the T-Sep sample). This was an expected observation since these samples had already undergone a thermal treatment at 400°C. TGA analysis was performed until 600°C and in both samples the adsorbed water molecules were degraded until 400°C, while a small amount of the structural water remained in the samples. The partial loss of adsorbed water at 400°C was also verified at FTIR spectra.

Zeta potential
The zeta potential distribution (Fig. 7) comes in agreement with previous experimental studies (Alshameri et al. 2018;Abrougui et al. 2019). Specifically, the zeta potential of raw and thermally modified samples was negative at the whole pH range examined (4-11).This can be attributed to the isomorphous substitutions of Al 3+ or Mg 2+ with Si 4+ on octahedral sheets of minerals, as well as to the potential exchange of monovalent ions with divalent ones, which affect minerals' electrochemistry of solid/liquid interface (Alshameri et al. 2018;Abrougui et al. 2019) . That resulted in more negatively charged interface with pH increase, rendering all the examined materials to be promising adsorbents for cations, especially at pH values > 5. Moreover, the thermal treatment had impact on the negative charge of the materials as well, since T-Pal and T-Sep samples presented more negative values than the raw samples.

XRF analysis
XRF analysis (Table 3) revealed that the chemical composition of the samples comprises mainly SiO 2 (62-65%) and MgO (17-25%), while Pal and T-Pal samples were also rich in Fe (< 9%) due to their deposit lithologies (Kastritis et al. 2003). The main oxides (Si, Al, Mg, and Fe) were observed to remain more or less constant following thermal treatment, although a slight increase was seen in T-Sep and SiO 2 and MgO decreased slightly in T-Pal. This may be attributed either the sample heterogeneity or structural rearrangements that took place after thermal treatment. Oxide percentages remained stable after NH 4 + -N adsorption thus indicating that all the presented ions are exchangeable with NH 4 + without strong preferences to one ion, verifying that ion exchange is an essential mechanism at ammonium removal. Moreover, due to the low concentration of NH 4 + , the potential %mass that could be ion-exchanged may be below the detection limits of the XRF equipment.

Thermal treatment effect and proposed NH 4 + -N removal mechanism on raw and thermally modified Pal and Sep
In order to better understand the insight of NH 4 + -N removal mechanism onto Pal, Sep, T-Pal, and T-Sep, the observations from characterization methods were evaluated for the adsorption process. So far, it was reported that ion exchange between Na + , Ca 2+ , and K + with NH 4 + is the primary mechanism for ammonium removal (He et al. 2016;Alshameri et al. 2018). However, the XRF results did not indicate significant reduction at %mass of the exchangeable cations of the adsorbents after NH 4 + -N adsorption (Table 3). This implied either that all the cations are exchangeable with ammonium as expressed in Eq. 15, or that ion exchange partially contributes to ammonium removal on raw and thermally modified fibrous clay minerals.  On the other hand, the XRD patterns and FTIR spectra of Pal and T-Pal verified the surficial interactions of NH 4 + with negatively charged active sites of Pal and T-Pal (Figs. 1,  4), because of the decreased intensity of the XRD reflections of Pal and T-Pal after NH 4 + -N adsorption, highlighting the ammonium bonding on the surface. The FTIR spectra verified the results obtained from XRD, since a new band was created at 1380 cm −1 for the Pal and T-Pal samples after NH 4 + -N adsorption, which is referred to surficial N-H bonding. Moreover, the BET analysis indicated that the SSA of the Pal sample was decreased after thermal treatment (T-Pal); nevertheless, according to the zeta potential distribution, T-Pal was more negatively charged than Pal at pH range 4-11. Potentially, the rearrangement that thermal treatment emerged and was observed (Fig. 3) gradually increased the number of exchangeable cations (Chen et al. 2011b), making T-Pal a promising adsorbent for cations, such as ammonium, by enhancing its ion exchange capacity in combination with its surficial interactions.
On the contrary, Sep and T-Sep sample characteristics were differentiated from Pal and T-Pal. Precisely, the XRD reflections of Sep and T-Sep samples were increased after NH 4 + -N adsorption, revealing that potentially the interactions between sepiolite samples and ammonium are not mainly surficial. The FTIR spectra also verified this case, (15) * N * a + , * K + , * C * a 2+ since no notable shiftings or new bands occurred after NH 4 + -N adsorption. Moreover, T-Sep presented an increase in pore diameter (10.4 cm 3 /g) compared to Sep (9.3 cm 3 /g respectively), which can be attributed to the inner channels breaking due to the removal of structural Mg. The Mg-O-Si and Mg-OH bonding proved to be broken at the FTIR spectra, as was mentioned in the "FTIR" section. All these structural changes in combination with the negative charge at a wide range of pH render the inner space and surface of Sep playing a key role at NH 4 + -N adsorption, while T-Sep adsorption capacity seems to be enhanced, due to the more negatively charged surface than Sep and the increase of the pore characteristics as well.

Effect of adsorbent dosage and initial concentration
The effect of Pal and Sep dosage on NH 4 + -N removal was examined for various initial concentrations of the pollutant. It was obtained that higher adsorbent dosages lead to higher NH 4 + -N removal for all the initial NH 4 + -N concentrations examined for both examined dosages of Pal and Sep (Fig. 8a,  b). These results were attributed to the high specific surface area of both clay minerals which provides numerous, readily available active sites for NH 4 + -N to be adsorbed (Alshameri et al. 2018). Specifically, the dosage of 4 g Pal or Sep adsorbent in 200 ml solution was found to be the most effective, especially for the removal of 1 or 2 mg NH 4 + -N /L below the permitted limit for drinking water (< 0.5 mg/L). The kinetic behavior of the highest examined adsorbent dosage (4 g) for all the initial NH 4 + -N concentrations examined is shown in Figure 9.
Both Pal and Sep presented maximum NH 4 + -N removal efficiencies of 60-80% within the first 20 min for all NH 4 + -N concentrations tested. Specifically, the highest removal efficiency (76%) of Pal was observed for lower initial NH 4 + -N concentrations (1 and 2 mg/L), in contrast with Sep that presented 75-80% removal when NH 4 + -N ranged from 4 to 6 mg/L, although its removal capacity decreased for 8 mg NH 4 + -N/L. It is possible that the larger fibers of Sep may enhance Sep adsorption capacity. Potentially, NH 4 + at low concentrations can contact with the most of Pal active sites, whereas Pal smaller basal space inhibits the excessed ammonium ions to entry in the interlayer space, leading to decreased NH 4 + -N removal at higher concentrations (Rytwo et al. 2000;Aydın Temel and Kuleyin 2016). On the other hand, Sep removal efficiency is higher for increased ammonium concentrations for constant 4 g of dosage. This may be attributed to the increased ratio of ammonium ions per L, available to interact with Sep active sites into the solution, leading to enhanced adsorption capacity. However, since the NH 4 + -N removal efficiency of Sep for 4-6 mg NH 4 + -N/L had no significant variation, probably the adsorption capacity of the specific dosage is achieved, and there were not available active sites to interact with 8 mg NH 4 + -N/L. To determine the effect of T-Pal and T-Sep dosage on the removal of NH 4 + -N of various concentrations, the batch experiments were performed by applying the optimum dose of each raw mineral (determined as 4.0 g) for 1, 2, 4, 6, and 8 mg/L NH 4 + -N removal. The removal efficiencies of both Pal and Sep increased by 10-20% after thermal treatment, but also retained the tendency presented as raw materials, concerning the NH 4 + -N concentration increase (Fig. 10a, b).
The efficiency increase can be attributed to the enhancement of negative charge that thermal treatment emerged according to the zeta potential results that can be attributed to water loss that alters and possibly slightly increases the interlayer space of both adsorbents (Zadaka-Amir et al. 2013). Chen et al. (2011a) reported that thermal treatment can reveal more exchangeable ions at palygorskite structure, increasing the ion exchange and adsorption capacity of the sample. This has also been reported in the "XRF analysis" section with XRF analysis. Specifically, T-Pal produced 85% removal of 2 mg/L NH 4 + -N compared to 70-75% removal by Pal due to potential ion exchange capacity increase after thermal treatment. The highest removal efficiencies of T-Sep was observed for 4 and 6 mg/L NH 4 + -N concentrations, since the pore diameter was increased as the BET analysis reinsured, whereas thermal treatment did not notably enhance the mineral's removal efficiency for 1 or 2 mg NH 4 + -N/L. Similar tendency was reported at the study of Balci and Dinçel (2002), where sepiolite ammonium removal efficiency was increased with initial concentration increase, until the saturation capacity of the used sepiolite was reached. Additionally, the removal rate was decreased at low initial ammonium concentrations at the study of Balci and Dinçel (2002), rendering the lower ammonium concentrations more time-dependent compared to the higher ones. This fact is also expressed in the present study at T-Sep for 1 and 2 mg NH 4 + -N/L, as well as at T-Pal for 1 mg NH 4 + -N/L. It is noteworthy that following thermal treatment, T-Sep was able to successfully lower a 4 mg initial NH 4 + -N/L concentration to below the permitted EU limit. The enhanced removal ability of T-Pal was apparent; however, final concentrations of ammonium were 0.1 mg/L above the permitted EU limits. Increases in adsorption, despite being limited, can be attributed to water loss that alters and possibly

Effect of pH
To examine the impact of pH on NH 4 + -N removal, the following conditions were applied: 4 g of adsorbent and 4 mg/L initial NH 4 + -N concentration were examined with pH values ranging from 2 to 11. These conditions were deemed suitable as lower ammonium concentrations (1-2 mg/L) are effectively removed to levels below the legislated limit (< 0.5 mg/L), and higher concentrations (6-8 mg/L) remained above this value after adsorption. The experimental results showed that the removal efficiency of all the adsorbents correlates positively to pH increase (Fig. 11), since the adsorption procedure is enhanced up to the case of pH 9, where it exceeds 85% for the thermally treated mineral samples. This efficiency can be explained by the state of NH 4 + ions in water as at low pH values they are present in the NH 4 + form and at pH values above 8.5 they are present as ammonia (NH 3aq ). At very low pH values, the H + ions compete strongly with NH 4 + for available surface adsorption sites and interaction between adsorbent and adsorbate is inhibited. Similar results were also recorded by Vu et al. (2017) who used biochar. It is likely that at pH values above 9, most of the NH 4 + is transformed to NH 3(aq) and the electrostatic attraction to clay minerals decreases (Vu et al. 2017;Pan et al. 2019). The experimental results come in agreement with the zeta potential distribution (Fig. 7). The negatively charged interface at higher pH values interpreted the low removal efficiency of all the samples at pH range 2-5. Moreover, at pH 6, the zeta potential decreases more steeply than the lower pH values while at pH 7 and pH 8 the zeta potential remained almost stable. This may explain the reason why at pH 7 the NH 4 + -N removal efficiency is not the highest, like it was noted in the study of Alshameri et al. (2018), highlighting the samples' origin dependence on interfacial properties. Moreover, the fact that at pH > 9 the zeta potential becomes more negative is not equalized with enhanced ammonium removal, potentially due to ammonium alteration to NH 3(aq) as it was abovementionedPlease check if the minimal edits to the sentences "Moreover, despite the fact..." and "According to Table 1, the water consistency..." retained the sentences' intended meaning

Saturation test
The potential reusability of Pal, Sep, T-Pal, and T-Sep as NH 4 + -N adsorbents was examined using 4 g of each adsorbent, 4 mg/L ammonium solution, and natural pH solution (5.5). The saturation test verified the effects of the thermal treatment on the fibrous clay minerals (Fig. 12). T-Pal and T-Sep achieved 81% and 85% removal, respectively, within the first 24 h and these rates remained almost constant until the third adsorption cycle (day 3). At 24 h, the ammonium removal efficiency of Pal and Sep samples was already declined. This could be attributed to the fine particle size of the samples in combination with the limited adsorbent dosage and relatively low ammonium initial concentration. The finest the particle size is, the sooner desorption rates can be achieved, especially under intense agitation (Keyes and Silcox 1994). Both untreated minerals achieved 55% removal (0.48 ±0.05 mg/g) in 24 h, but only the removal pace of Sep remained almost stable for two more cycles, whereas Pal adsorption capacity decreased by up to 17% from the second day. According to this saturation test, T-Pal or T-Sep can be characterized as sufficient adsorbents for groundwater treatment that adsorbed 0.64-0.68 ± 0.02 mg/g NH 4 + -N respectively, degrading the NH 4 + -N concentration to the acceptable drinking limits during 24 h. Nevertheless, none of the raw or thermally treated clay minerals tested presented sufficient reusability for treatment of water for NH 4 + -N, but all could be effectively reused for pre-treatment. Pal and Sep both removed 4.28 ±0.02 mg NH 4 + -N/g by the end of the saturation test (day 22), while T-Pal and T-Sep removed 5.95 ±0.04 mg/g NH 4 + -N in the same period. The multiple adsorption cycles were attributed to the primary saturation of the external surfaces of the adsorbents. When this point was reached, the adsorbate enters into the adsorbent's interlayer space and pores (Alshameri et al. 2018). No research is available on NH 4 + -N saturation in Pal and Sep to compare the results.

Effect of operational parameters on removal efficiency-a comparative analysis
Comparing the results of the present study with frequently used NH 4 + adsorbents, presented in Table 4, it is evident that the achieved removal efficiencies are within the literature range. The main operational parameters for ammonium removal are NH 4 + initial concentration, adsorbent dosage, and pH. Table 4 presents the removal efficiencies from batch tests, at room temperature and at natural pH, for various experiments. Concerning the zeolites and other clay minerals, researchers proceeded in using either modified or nonmodified materials. Modification can increase the efficiency of one material (Yin and Kong 2014;Fu et al. 2020), but at the same time, cost and effort of each modification should be considered. Moreover, compared to the most frequently used natural adsorbent for NH 4 + -N, zeolite, the fibrous clay minerals presented similar removal capacity with the study of Fu et al. (2020) or Kotoulas et al. (2019). Nevertheless, Fu et al. (2020) underwent zeolite two different modification methods for the removal of 5 mg/L NH 4 + -N, whereas the double adsorbent dosage from current study was used in the study of Kotoulas et al. (2019) (Table 4). Also, biochars may present high removal efficiencies; however, very high biochar dosages were demanded for these removal rates, significantly increasing the operational cost . Vermiculite and montmorillonite in the study of Alshameri et al. (2018) presented also high adsorption capacities. However, the removal efficiency of any mineral adsorbent strongly depends on the purity, chemical composition, and properties of each sample which vary, as it can be observed from Table 4. Specifically, vermiculite at the study of Wang et al. (2011) was less effective than vermiculite of Alshameri et al. (2018) used, even though almost 10 times more dosage was applied. Moreover, Gianni et al. (2021) used palygorskite with ammonium removal relatively lower than the results presented in this study (Table 4). Summarizing, thermal treatment has not been applied before at the fibrous clay minerals as the main modification method, as well as saturation test for ammonium removal has not been performed. Concerning the data of Table 4, the adsorbents of the present study could be a satisfactory alternative for low ammonium concentration removal.

Adsorption isotherms
The linear and nonlinear forms of the Langmuir and Freundlich isotherms were applied to Pal, Sep, T-Sep, and T-Pal to determine the variation of their removal capacities with increased NH 4 + -N initial concentrations. Both linearized Langmuir and Freundlich isotherms presented a good fit to all the adsorbents examined since the R 2 value in all cases is > 0.95 (Table 5). Nevertheless, the linear form of the Freundlich isotherm expressed better the adsorption of NH 4 + -N on Pal and T-Pal with R 2 values of 0.974 and 0.991, respectively, which renders the adsorption a heterogeneous procedure. The same tendency was also observed for the nonlinear isotherm models, verifying the heterogeneous sorption of NH 4 + on Pal and T-Pal with correlation coefficients 0.969 and 0.989 respectively ( Table 6). The nonlinear models expressed a better fit only at Langmuir isotherm with R 2 0.968 and 0.986 instead of the 0.952 and 0.954 at linearized form for Pal and T-Pal  On the other hand, the adsorption behavior of Sep differentiated slightly after thermal treatment as the Langmuir isotherm for Sep presented a better fit for NH 4 + -N adsorption, but for T-Sep adsorption is expressed better by the Freundlich isotherm with R 2 0.998 (compared to R 2 0.991 for the Langmuir isotherm). This result revealed the preference of almost monolayer NH 4 + -N adsorption on the Sep mineral, which potentially became heterogeneous for T-Sep. This change was probably attributed to the increased number of micropores occurring after calcination, as was verified by the BET analysis. Similarly, the nonlinear models indicated the effect of thermal treatment on NH 4 + -N removal, since the nonlinear Freundlich model fits slightly better on T-Sep instead of the Langmuir's (Table 6), whereas Langmuir isotherm expresses better the interactions on Sep, verifying that the removal of NH 4 + -N was monolayerd at the raw sample and became heterogeneous after thermal treatment. Briefly, based on the isotherm parameters at Tables 5 and 6, there is no dominance between the linear or nonlinear models based on correlation coefficient, since the nonlinear form of Langmuir fits better on Pal, T-Pal, and T-Sep, whereas the nonlinear Freundlich model fits better only at T-Sep sample. However, the statistical errors are higher for the linear forms of isotherm models, rendering the nonlinear ones better to express the adsorption process; in specific, the nonlinear Freundlich isotherm expresses the interactions on Pal, T-Pal, and T-Sep, while the nonlinear Langmuir model fits better on Sep sample.

Adsorption kinetic models
The relative parameters of all applied linear and nonlinear kinetic models (pseudo-first-order, pseudo-second-order, and intra particle diffusion) are presented in Table 7 and 8 respectively. The correlation coefficients (R 2 > 0.99) showed that the linear pseudo-second-order fits the data better than the other two models or the nonlinear ones (Table 7 and 8). These results were in agreement with FTIR spectra and other studies that examined NH 4 + -N removal from aqueous solutions using aluminosilicate adsorbents and concluded chemisorption as adequately expressing NH 4 + -N removal on clay minerals (Sun et al. 2017;Alshameri et al. 2018). Adsorption capacity at equilibrium (q e ) is a significant parameter of each model's evaluation. From each kinetic model result, a q ecal value is determined from the intercept and the slope of the pseudo-first-and pseudo-second-order plots, respectively, which is further compared to the maximum removal rate at equilibrium (q eexp ). In all cases, the q eexp variables were in full agreement with the q ecal from the linear pseudo-second-order model. So, despite the good fit of nonlinear plots (R 2 > 0.94) with very low statistical errors (Table 8), the q ecal of nonlinear pseudo-first-or pseudo-second-order kinetics is not representative of q eexp for neither of the adsorbents and cannot express the specific adsorption procedure.
The Weber-Morris nonlinear model is parted from two linear segments (Table 8). In the first stage, NH 4 + -N diffuses from the aqueous solution to liquid:solid interface, while in the second stage the adsorbate diffuses from the interface into the adsorbent pores Weber and Morris (1963). In all cases, stage 1 k id > stage 2 k id , thus indicating the rapid diffusion of NH 4 + -N at the solids interface, compared to a very slow rate diffusion in their inner pores (Genethliou et al. 2021).

Real-water application of the adsorbents
The raw and thermally treated samples were applied in the optimal dosage determined at the previous batch experiments (4 g adsorbent), for 4 mg/L NH 4 + -N degradation from ammonium rich tap water. According to Table 1, the water consistency is enriched in potentially competitive cations for NH 4 + , such as Na + , Ca 2+ , and Mg 2+ , and is an extremely hard water (WHO 2011). Under these conditions, Pal and T-Pal removal efficiency for NH 4 + -N was strongly influenced, since it was decreased up to 50% for both samples (Fig. 13). Palygorskite clay mineral has reported in other studies strong ion exchange affinity for Ca 2+ spontaneously (Lazaratou et al. 2020a), which can be inhibiting factor for ammonium removal (Yin and Kong 2014). Sep and T-Sep samples' removal efficiency for NH 4 + -N was impacted as well; however, 25% and 20% decreases were noted (Fig. 13), rendering Sep and T-Sep with stronger affinity -N removal capacity of the examined samples, further investigation must be conducted, since there is limited literature about natural water systems and how their composition or the concentration of the existed ionic species influence interactions between ammonium and clay minerals' interface.

Cost-benefit analysis
The operating cost of the batch tests was estimated, considering the purchase of each adsorbent, and the thermal treatment cost. The cost analysis for Pal and T-Pal samples was based on price information given by Geohellas S.A., whereas those costs were compared to the widely used zeolite, according to USGS (2020) financial information. No accurate cost range was available for sepiolite, while for this study there was no purchase cost since Sep was selected from the authors. Precisely, the highest quality of Pal, T-Pal, and commercial zeolite cost reaches 0.22 €/kg, 0.35 €/kg, and 0.25 €/kg respectively. According to Table 4 and comparing the results of this study with those of Kotoulas et al. (2019), the cost per gram of NH 4 + -N removed was 1.62€, 2.52€, and 2.19€ for Pal, zeolite, and T-pal, respectively. Thus, Pal seems to be antagonistic to common applied adsorption media such as zeolite.

Conclusions
Raw and thermally treated palygorskite (Pal/T-Pal) and sepiolite (Sep/T-Sep) were applied as low-concentration NH 4 + -N adsorbents from aqueous solutions. The extensive characterization of the samples verified the thermal Table 8 Kinetic parameters of nonlinear pseudo-first-order, pseudo-second-order, and intraparticle diffusion kinetic models for NH Pal, Sep, T-Pal, and T-Sep presented different adsorptive properties, especially after thermal treatment, since Pal and T-Pal mostly interact with NH 4 + -N on its surface, conversely to Sep and T-Sep where the inner interactions are dominant. According to a series of batch kinetic experiments, NH 4 + -N removal process was quite rapid and preferable at pH 4-9. Removal rates were adequate to reduce NH 4 + -N concentrations to below the permitted EU limit for drinking water for initial NH 4 + -N concentrations of 1 and 2 mg/L, thus rendering Pal, Sep, T-Pal, and T-Sep suitable materials for the treatment of most contaminated aquifers. T-Sep can also be used to treat groundwaters with initial NH 4 + -N concentration of 4 mg/L. Nevertheless, the presence of competitive ions such as Ca 2+ strongly deteriorate ammonium interaction with Pal and T-Pal, indicating that its application in very hard waters may prohibit NH 4 + -N degradation. On the contrary, Sep and T-Sep may not be strongly influenced from competitive ion presentation but achieved decreased NH 4 + -N removal up to 25%. The saturation test indicated that T-Pal and T-Sep are effective at NH 4 + -N removal for 24 h but cannot be reused for water treatment when solutions are highly contaminated (4 mg NH 4 + -N/L). Nevertheless, Pal, Sep, T-Pal, and T-Sep can be applied as suitable pre-treatment materials for multiple times. Mechanical and kinetic properties of the procedure were determined using the linear and nonlinear forms of Langmuir and Freundlich isotherm models, as well as the linear and nonlinear form of pseudo-first-order, pseudosecond-order, and Weber-Morris kinetic models. All the abovementioned models fit to the data of each tested mineral sample; however, the nonlinear Freundlich isotherm and linear pseudo-second kinetic models presented better fits for all samples, thus indicating the heterogeneous nature of adsorption via strong chemical bonds (chemisorption). Finally, based on the cost benefit analysis, the examined Pal and T-Pal samples were found to be strongly competitive to commercial zeolite adsorbent, concerning their cost per gram of NH 4 + -N that was removed.