Simultaneous removal of ammonium ions and sulfamethoxazole by ozone regenerated high silica zeolites

.


Introduction
The presence of organic micropollutants in industrial and municipal wastewater has become a major concern (Rogowska et al., 2019;Kamaz et al., 2019 ).These emerging pollutants include pharmaceuticals, personal care products, industrial chemicals, pesticides, fire retardants, etc.The increasing concentration of these compounds in wastewater requires that wastewater treatment facilities introduce an additional removal technology, although discharge guidelines for many of these compounds do not exist.Various studies have shown the presence of these pollutants in effluents from sewage treatment (up to 10 μg/L) and in surface water with average values up to 1 μg/L, and maximum values up to 10 μg/L ( Sathiskumar et al., 2020 ;Liwarska-Bizukojc et al., 2018 ;Zhou et al., 2019 ).Given that most existing wastewater treatment plants (WWTPs) are not designed for removal of these compounds, they end up in WWTP effluent, while, in practice, water authorities aim at reducing or eliminating the adverse effects of trace organic compounds on the aquatic environment and on human health ( Loos et al., 2013 ;Guillossou et al., 2019 ).
Additionally, ammonium is a key component in most domestic WWTPs.Anticipated new treatment targets to avoid water pollution require a large number of WWTPs to meet discharge consents between 1 and 3 mg NH 4 + /L ( WHO, 1996 ;Milieurecht 2017).The conventional process for ammonium removal consists of aerobic nitrification and anoxic denitrification, consuming high amounts of energy ( van der Hoek et al., 2018 ).A possible solution to move towards energy consumption reduction and efficient ammonium removal could lie in ion exchange ( Amini 2017 ).A variety of materials can be used in such an ion exchange process, including polymeric ion exchange resins, as well as natural and synthetic zeolites.Especially natural zeolites, including Clinoptilolite, MOR and Chabazite, with Clinoptilolite being the most commonly used, tend to be effective in ammonium removal ( Langwaldt, 2008 ;Widiastuti et al., 2011 ;Wang et al., 2010 ).The purity, chemical composition, crystal size, porosity and pore diameter of these zeolites vary, which influences their efficacy for wastewater treatment in terms of their capacity and selectivity towards ammonium ( Moshoeshoe et al., 2017 ).
On the other hand, in recent years, synthetic, high silica, zeolites have been evaluated as alternative adsorbents for activated carbon, for organic micropollutant removal from water.A synthetic zeolite is made by modification of clay and other aluminium bearing minerals, and shows promising operating efficiencies, comparable to the natural zeolites ( Ham et al., 2018 ).The capacity and selectivity of the various zeolites are influenced by a combination of the internal pore sizes of the lattice structure, the ratio of Si/Al and the distribution of Al within the lattice frame ( Jiang et al., 2018 ).MOR was selected as a high silica zeolite, amongst others, due to its commercial availability and proven removal of wide range of different or ganic micropollutants ( Jiang, 2019 ).The removal of micropollutants (in this case pharmaceuticals) on MOR might benefit from the wide range of pores sizes, i.e. 2.6 Å-7.0 Å, from two types of channels with opening sizes 2.6 Å * 5.7 Å and 6.5 Å * 7.0 Å ( Jiang, 2019 ).In addition, MOR has been, next to clinoptilolite, reported as a zeolite for ammonium removal (Chen et al., 2018).The possibility of regenerating MOR zeolite by ozone is reported ( Wang, 2020 ).Therefore, MOR could have potential adsorption characteristics for micropollutants, rather than clinoptilolite, with a simultaneous decrease of ammonium concentrations in water." The above mentioned processes are, in principle, reversible so that, after the zeolite has been exhausted, it could be regenerated by treating the zeolite with e.g. an advanced oxidation technology such as ozonation ( Zhang et al., 2014 ), being economically and environmentally competitive when compared to the large amounts of salts required for the conventional regeneration of exhausted zeolites ( Deng et al., 2014 ).Additionally regeneration is done on-site making it more attractive than granular activated carbon (GAC), for which reactivation plants are normally used ( Clements et al., 2004 ).
Due to the co-existence of multiple contaminants in water sources, including ammonium and micropollutants as indicated above, technologies for their simultaneous removal are desirable.The rationale behind promoting simultaneous removal of micropollutants and ammonium, is to promote so-called "one-step" treatment methods, replacing complex and multi-step methods of already existing conventional processes mentioned in the paragraph above.In the current study, we developed an innovative and sustainable polishing concept that can remove multiple contaminants in a single process and could potentially be incorporated in existing wastewater treatment plants.We demonstrated the potential of this treatment strategy using laboratory-scale experiments.The innovation thus lies in 1) simultaneous removal of organic micropollutants and ammonium from water by zeolite granules and 2) regeneration of granules with ozone.Accordingly, the current paper provides the comparison of a synthetic (MOR) and natural zeolite (Zeolita) in a series of batch experiments in ultra-pure water to evaluate the performance and assess the impact of initial ammonium concentration on ammonium and sulfamethoxazole (SMX) removal efficacy.Sulphonamide antibiotic SMX was chosen as a model compound.Nowadays, the sulfonamide group is present in many drugs as antibiotics, antimalarics, diuretics, hypoglycemics, antiinflammatories, antihypertensives, antitumor compounds, antithyroid compounds, etc., making SMX good representative drug for broad range of antibiotics ( Zhou et al., 2018 ;Thiebault, 2020 ).Additionally, SMX was chosen as a model compound due to its resistance to remediation and natural attenuation and due to its extensive use in both human and veterinary medicines ( Jiang, 2019 ;Bueno et al., 2012 ;Loos et al., 2013 ).This antibiotic has been classified as emerging contaminant, and has been found in water resources at concentration levels from ng/L to μg/L ( Bueno et al., 2012 ;Al Aukidy et al., 2012 ;Loos et al., 2013 ).SMX has been detected in the concentration range from 1 to 150 ng/L in WWTP effluents and along Dutch rivers ( Zhou et al., 2018 ;Sabri et al., 2020 ).

Zeolite materials and zeolite characterisation
For the experiments two zeolites were used: a natural Clinoptilolite NZ (Zeolita, Spain) and a synthetic Mordenite (MOR, 690 HOA, Si/Al~240, Tosoh Corporation, Tokyo, Japan).MOR powder zeolite had a surface area (BET) around 450 m 2 /g with a crystal size of 0.1 × 0.5 μm ( Jiang, 2019 ).The number in parentheses refers to the Si/Al ratio of the zeolite.Preliminary tests on SMX removal by four different types of synthetic zeolites (MOR, FAU, BEA and MFI) indicated that the best ammonium removal was achieved by MOR zeolite ( Jiang et al., 2018 ), and, therefore, MOR was used during the experiments.
Both granular and powdered zeolites can be used in liquidphase applications ( Jiang et al., 2018 ).However, in this study the granular form was chosen since it can be used in continuous processes and can be regenerated, whereas the powdered form is generally used in batch processes, without regeneration.Additionally, the granulation process transforms the fine powders into dust-free granules that are easy to compress.The granules had a spherical shape, for content uniformity and an adequate moisture and hardness to prevent breaking (determined previously by Elshof, 2019 ), dust formation during process and loss of material.Zeolite granules were made using zeolite powders combined with bentonite (Honeywell Fluka TM , USA), which functions as a binder, in the percentage ratio of 85:15 (zeolite/bentonite).Pre-treatment of the zeolites was done by washing it with Mili-Q water.Afterwards the samples were dried for a day in an oven at 105 °C to remove moisture.Finally, the granules were calcinated in the oven for 3 h at 600 °C for NZ and for MOR at 850 °C.
In order to compare the crystal structure and the mineralogy of the zeolite, X-Ray (XRD, Bruker-AXS D5005 diffractometer, using filtered Cu K α radiation, with accelerating voltage 45 kV, current 30 mA and scanned at 2 θ from 4 to 50 °) diffraction analyses were performed.The chemical composition of the zeolite samples was determined by using X-Ray Fluorescence (XRF), using a Panalytical Axios Max WD-XRF spectrometer, and data evaluation was done with SuperQ5.0i/Omniansoftware ( Boatemaa et al., 2018 ).

Batch study for SMX and ammonium removal using zeolita and MOR
Equilibrium adsorption was determined in batch experiments, using 1 L flasks for each of the zeolites.Experiments were conducted in demineralised water (RiOs Reverse Osmosis System).Various studies have shown the presence of SMX in effluents from sewage treatment (up to 10 μg/L) and in surface water with average values up to 1 μg/L, and maximum values up to 10 μg/L ( Sathiskumar et al., 2020 ;Liwarska-Bizukojc et al., 2018 ;Zhou et al., 2019 ).In the SMX adsorption experiments, the initial concentrations of SMX were 2 μg/L and 50 mg/L.The latter was chosen to study the changes in competitive adsorption ( Jiang et al., 2018 ).The MOR concentration was varied between 0.01 g/L and 1 g/L, obtained from previous research ( Jiang et al., 2018 ).A blank measurement (0 mg/L adsorbent dose) was included in each of the isotherm series.The solutions with granules were mixed using a magnetic stirrer.After 10 d for the 50 mg/L initial concentration and after 4 weeks for 2 μg/L, samples were filtered through a 0.20 μm membrane filter and analysed.Moreover, to study the effect of initial ammonium concentration and adsorption isotherm, batch experiments were conducted by varying the initial ammonium of 10 -100 mg/L with Zeolita/liquid ratio 2 g/100 ml at 72 h contact time and at a temperature of 20 °C.Ammonium-containing water was prepared from ammonium chloride (Sigma Aldrich).
To determine the above contact times, to reach equilibrium, kinetic studies of ammonium and SMX removal by Zeolita and MOR from aqueous solution were carried out by agitating 1 L of ammonium or SMX solution of known initial concentration with 1 g of MOR and Zeolita for SMX and 20 g of Zeolita or MOR for ammonium at room temperature (20 °C).A quantity of 1 mL of uniform dispersion was pipetted out, using a 5 mL-syringe at different time intervals.The collected samples were then filtered (0.2 μm syringe filter, Whatman SPARTAN TM ), and the concentration in the solution was analysed.
The pH of all solutions was measured by using a pH metre (Inolab R Multi 9420 IDS, WTW, Germany) and was adjusted by using 0.1 M NaOH or 0.1 M NaCl solutions to get pH 7.The ammonium concentration was determined by cell tests (Hach R LCK Cuvette Tests, LCK302, LCK303 and LCK304), using a spectrophotometer (Hach R LANGE model DR3900, the Netherlands) and confirmed by Ion chromatography (IC,883 Basic IC plus,Metrohm,Switzerland). For this method a cation exchange C6 column (Metrohm, Switzerland) was used, with 3 mM HNO 3 (Sigma-Aldrich, Germany) as the eluent.The samples were placed in an autosampler (919 IC Autosampler plus, Metrohm, Switzerland) and measured with a flowrate of 0.9 mL/min.
For SMX analyses 1 ml of sample was acquired from the sample bottle.SMX in water samples of the experiments with initial concentration of 50 mg/L were determined with a HPLC system (Shimadzu Corporation, Japan), equipped with reversed phase C18 column; 2.6 μm, 100 × 4.6 mm (Kinetex Core-Shell C18 column, Phenomenex R , USA). 10 μL of sample was injected into the system and the elution was conducted with a mixture of acetonitrile and ultrapure water (60:40, v/v) at a flow rate of 0.6 ml/min.The column oven was set at 30 °C.The eluted compound was monitored with UV detector at a wavelength of 215 nm.
For the samples of the experiments with low initial SMX concentration (2 μg/L), the Waters Acquity UPLC Plus system was used, with an ACQUITY UPLC R BEH C18 (1.7 μm particle size, 2.1 × 50 mm, Waters Ireland) analytical column.The column temperature was 40 °C.The injection volume and flow rate, pumped by an ACQUITY UPLC IClass Plus (Waters/USA), were 10 μl and 0.35 ml/min, respectively.A gradient mobile phase was used: water containing 0.1% (v/v) formic acid (mobile phase A) and acetonitrile containing 0.1% (v/v) formic acid (mobile phase B).The system was coupled with tandem mass spectrometry (MS).MS was conducted on a Xevo TQ-S micro tandem quadrupole mass spectrophotometer (Waters/USA), equipped with electrospray ionization in positive mode (ESI + ), detecting two fragments (quantifier & qualifier, verified against the Metlin or Massbank MS/MS databases) of each analyte and deuterated internal standards (Torronto Research Chemicals/Canada), with quantification by 10-point calibration at levels from 0.0025 μg/L to 10 μg/L.The MassLynx V4.2 software was used for data acquisition and quantitative analysis.

Adsorption isotherm and data analysis
The amount of contaminant adsorbed from the aqueous solution was expressed as equilibrium loading per mass of the zeolite (q e ) Where, C o is the initial pollutant concentration (mg/L), C e is the equilibrium concentration (mg/L), V is the batch volume (L) and m is the zeolite mass.Equilibrium data for each of the zeolites was then compared to Freundlich and Langmuir isotherm models ( Ayawei et al., 2017 ).
In the Langmuir model it is assumed that adsorption is monolayer and that adsorption energy is constant.The Langmuir adsorption isotherm can be expressed as: where q e (mg/g) is the amount of compound adsorbed per unit mass of adsorbent, C e (mg/L) is the equilibrium concentration of the compound, q max (mg/g) and K L (L/mg) are Langmuir constants related to adsorption capacity and rate of adsorption ( Ayawei et al., 2017 ).
On the other hand, the Freundlich model assumes that the surface energy of the adsorbent is heterogeneous and that first sites with stronger binding will be occupied.
The Freundlich adsorption isotherm is shown as: where K F (mg/g (L/mg) 1/n ) is the adsorption capacity.The n value (0-1) is factor that indicates adsorption intensity ( Ayawei et al., 2017 ).

Zeolite regeneration
Oxidation by ozone was found to be robust, scalable and cost effective ( Ateia et al., 2018 ) for the regeneration of zeolites.Ma et al. (2019) already studied the effective adsorption and regeneration of SMX loaded high-silica zeolites.Therefore, the specific objectives of regeneration in this study were: (a) optimal pH by regeneration of granules loaded with ammonium by ozone; (b) evaluation of regenerated zeolites over several adsorption/regeneration cycles.

a) Oxidation of ammonium by ozone
To perceive under which conditions zeolite granules, saturated with ammonium, can be regenerated, experiments were conducted by dissolving ammonium in buffers with different pH values (7, 8 and 9) and subsequently oxidizing the dissolved ammonium by ozone.Boric acid-sodium borate mixtures were employed to maintain a constant pH.Before the ozonation test, the initial pH was measured using a pH metre (Inolab R Multi 9420 IDS, WTW, Germany).The reaction solution was sampled every 5 min to analyse the changes in pH values and concentrations of ammonium.Concentrations of ammonium were measured by spectrophotometry methods and ion chromatography (IC, 883 Basic IC plus, Metrohm, Switzerland).

a) Adsorption-Regeneration Cycles
Adsorption-regeneration cycles were conducted to examine the regeneration efficiency of the zeolite granules after use.The adsorption cycle consisted of loading the zeolite granules with ammonium.Adsorption in a batch reactor was followed by a regeneration cycle.After regeneration, another loading cycle was executed.
Therefore, granules, loaded with ammonium, were added into the 200 mL glass-reactor after the adjustment of pH for the regeneration process.Then ozone was introduced in the solution through the glass diffuser.The zeolite granules were stirred by a magnetic stirrer.Ozonation was performed for 2 h, whereas ozone concentration in the gas phase was 90 g/m 3 with flow rate of Moreover, removal of the compound during ozonation can be attributed to either oxidation by ozone and/or formed hydroxyl radicals ( •OH).However here •OH radical contribution to the regeneration of the granules at pH 9 is assumed to be very small since it was found that •OH formation accelerates at pH 10 ( Zhang et al., 2015 ;Elovitz et al., 2008 ).The regeneration process was followed by rinsing the granules with demineralised water and drying them in an oven at 60 °C.Then the regenerated zeolites were, again, exposed to batch adsorption in a fresh ammonium solution.The procedure was repeated three times (3 cycles) for different initial ammonium concentrations.After each regeneration cycle, the adsorption capacity of the regenerated granules was measured.All adsorption and regeneration experiments were conducted in duplicates.An air stripping experiment was set as a blank experiment to exclude the stripping removal of ammonia by ozone.The removal efficiency ( R ) was calculated by Eq. ( 2) as follows: R = where C 0 is the initial concentration of ammonia (mg/L) and C is the concentration (mg/L) at reaction time t (min).

Characteristics of natural and synthetic zeolite
Chemical characteristics of the natural, Zeolita, and the synthetic zeolite, MOR, were analysed using XRD and XRF and are presented in Table 1 , showing that both Zeolita and MOR mainly consist of silica and alumina.However, the hydrophilic/hydrophobic features, controlled by varying the Si/Al ratio, can influence the behaviour of zeolites towards polar/non-polar reactants and products in adsorption processes ( Sarti et al., 2017 ).This is confirmed by the observation that MOR, with Si/Al~240, was an effective adsorbent for the removal of SMX ( Figs. 4 and 5 ).Table 1  that both zeolites contain exchangeable cations such as Fe 3 + , Ca 2 + , Mg 2 + , K + and Na + , potentially being able to be exchanged with other cations in the solution, such as, in this case, ammonium.

also reveals
The zeolites were additionally characterised by XRD to determine the mineral structure, and the results are given in Fig. 1 .It was revealed that Zeolita comprised, besides clinoptilolite, muscovite and nontronite ( Fig. 1 a), and the MOR of mordenite ( Fig. 1 b), which was to be expected.In case of natural zeolites used in other research, the XRD analyses have shown that the dominant mineral component is clinoptilolite ( Langwaldt, 2008 ).However, the content of clinoptilolite in zeolitic material can differ from 25% to 70-75% ( Franus et al., 2011 ).Besides clinoptilolite in natural zeolites small amount of opal, quartz, montmorillonite or heulandite can be found ( Langwaldt, 2008 ).When comparing MOR zolites, in general, the overall x-ray investigation of natural or modified MOR zeolite samples in the literature mainly consist of mordenite which agrees with the findings in Fig. 1 b) ( Sakizci et al., 2015 ;Hussain et al., 2019 ).

Ammonium removal by zeolita and MOR
Zeolita and MOR ( Table 2 and Fig. 2 ) were tested on ionexchange capacity, focusing on ammonium removal.The ammonium adsorption was found to increase with increasing initial ammonium concentration from 10 -100 mg/L at a constant zeolite amount of 20 g/L.Based on Fig. 2 , it can be observed that ammonium adsorption capacity on Zeolita was between 0.5 -3 mg/g and on MOR between 0.4 -2 mg/g.The solid points belong to experimental data while the lines are calculated according to Equations (2) (dashed line) and (3) (solid line).
The adsorption process was evaluated by fitting the data to two adsorption isotherms; the Langmuir and Freundlich.The isotherm Eqs. ( 2) and ( 3) and the experimental adsorption equilibrium data were used to determine the Langmuir and Freundlich parameters, for ammonium both for Zeolita and MOR, and for SMX only for MOR.Both the Langmuir and Freundlich isotherm models equally best fitted the experimental adsorption data with high coefficient of determination R 2 .The Langmuir parameters ( K L and q max ) and Freundlich parameters ( K F and 1/ n ) for the adsorption isotherms are given in Table 2 .However, in view of the higher R 2 value, listed in Table 2 , the Freundlich model showed to be the best fitting model for MOR.The conformity to the Freundlich model is a result of the heterogeneous nature of the zeolite surface and supports the earlier multi-linear nature of ammonium adsorption ( Wasielewski et al., 2018 ;Kotoulas et al., 2019 ).
On the other hand, it was found that the Langmuir model yielded a better fit to the data compared to the Freundlich when using Zeolita.In this case it is assumed that adsorption happened at specific homogeneous sites within the zeolite, and that there was no interaction between adsorbate molecules.Langmuir parameters with other natural zeolites from various ori- gins, Widiastuti et al. (2011) has reported that a q max value of 6.032 mg/g at initial ammonium concentration of 5 −120 mg/L using an Australian zeolite.However, q max values of 0.608-6 mg/g at initial ammonium concentration of 10-80 mg/L ( Demir et al., 2002 ) and 6.65 mg/g ( Rahmani et al., 2004 ) have also been reported.In our case the maximum achieved adsorption capacity of 4.081 mg/g, was less compared to Australian zeolite.Lower adsorption capacity can be attributed to the variation in their mineralogical compositions and morphology ( Selvam et al., 2018 ;Widiastuti et al., 2011 ).Although the results show a large variation with regard to the maximum ammonium adsorption capacity at the varied concentrations, it can nevertheless be found in the literature that in the range from 10 to 80 mg/l maximum removal efficiency of zeolite can be achieved, while increasing the initial ammonium concentration decreased its removal percentage ( Wasielewski et al., 2018 ;Kotoulas et al., 2019 ).
For the Freundlich model similar values of K F and 1/n ranged from 0.522 to 0,607 have been reported for ammonium removal using natural zeolites ( Demir et al., 2002 ;Widiastuti et al., 2011 ).For all initial ammonium concentrations, the 1/ n parameter ranges from 0 to 1. On this basis, it can be assumed that adsorption conditions of ammonium on zeolites were favourable.
The adsorption capacities of ammonium by MOR were similar to those of Zeolita at low initial ammonium concentration (10-40 mg/L).The measured data showed a larger difference in adsorption efficiencies when higher initial ammonium concentrations were present (60-100 mg/L).Similar outcome was found when comparing Romanian zeolite tuff and synthetic zeolites 5A and 13X ( Wang et al., 2010 ).

SMX removal
Adsorption of SMX onto the selected zeolite granules has been carried out at room temperature (20 °C) under different initial SMX concentrations (about 50 mg/L and 2 μg/L, respectively), and with and without the presence of NH 4 + ions (40 mg/L).However, no measurable reduction in SMX concentration in the presence of the natural zeolite clinoptilolite has been reported in literature ( Farías et al., 2003 ) or observed experimentally (data not shown).Figs. 3 and 4 illustrate the adsorption isotherms of SMX on the MOR granules.The adsorption capacity increased with increasing the equilibrium concentration, as expected ( Jiang et al., 2020 ).Removal of SMX by initial SMX concentration of 50 mg/L by MOR zeolite (1 g/L) in the absence of NH 4 + ions was 71.6% (from initial 56.59 mg/L of SMX, 40.52 mg/L was removed from the solution by adsorbent) in agreement with prior work ( Braschi et al., 2010 ).From the results in Table 2 it can be concluded that the experimental data are found to be best fitted by Freundlich equation.Therefore adsorption isotherms were fitted by the Freundlich isotherms in Fig. 3 and 4 .Similar behaviour was obtained by de Sousa (2018) showing that stronger binding sites on the surface are occupied first and that the binding strength decreases with the increasing degree of site occupancy.Additionally, he presented a SMX adsorption isotherm as a characteristic profile of type Scurves, indicative of vertical orientation of adsorbed molecules at the surface ( de Sousa et al., 2018 ).
In addition, isotherms' comparison of single SMX adsorption on MOR and adsorption of SMX in the presence of ammonium on MOR are shown in Fig. 3 and Fig. 4 .Increased SMX adsorption on MOR occurred in presence of NH 4 + ions at concentration of 40 mg/L, where, in this case, the adsorbed amount was 78.3% of the initial amount present in the solution.This rise in adsorption efficiency of SMX probably reflects that the process was assisted by surface complexation amongst SMX, NH 4 + and MOR, as suggested by Bajusz et al. (1998) , who studied the adsorption of N 2 -O 2 mixtures at different P N 2 / P O 2 ratios on a series of NaCaX zeolites with different degrees of Na + replacement.It was shown that, for single adsorbates, the adsorption increased when Na + cations were replaced by Ca 2 + cations.An exchange causes an increase in adsorption, suggesting that Na + is replaced at the sites which are accessible to the molecules.Sato et al. (2003) gave an additional explanation on this behaviour through the observation of structural changes of zeolites during ion exchange.In this research experiments on structural changes of zeolite were conducted (XRD and XRF analysis) with "clean" and loaded zeolite.However, no changes could be detected (data not shown in the manuscript).Therefore, we hypothesize that structural changes of the zeolite occur during the ion exchange and we support it with literature findings and measurements done by Sato et al., 2003, Price et al., 2017, Munthali et al., 2014 ;Azambre et al., 2020 .They found that the zeolitic framework structure deteriorated as evidenced by decreases in the crystallinity and micropore surface area.amongst three zeo-lites, one of them showed the most prominent dealumination and mesopore formation.Price et al. (2017) confirmed the same structural changes occurrence on another type of zeolite after partially exchanging the extra framework Na + ions with monovalent, Li + , K + , Rb + and NH 4 + and divalent, Ca 2 + cations.The results that she obtained from the NMR and FT-IR spectra indicated that the local framework environment was affected and that a loss of some aluminium from the framework had occurred.Dealuminated zeolites contain less Al which corresponds to a higher Si/Al molar ratio.The increased SMX adsorption in the presence of NH 4 + ions, could thus be explained by structural changes of MOR, during ion exchange with NH 4 + , leading to a zeolite with a more favourable Si/Al ratio and mesoporosity, enhancing the accessibility of active sites, for SMX adsorption, as observed in similar studies ( Munthali et al., 2014 ;Azambre et al., 2020 ).
However, the presence of NH 4 + at the surface might also have resulted in a higher positive surface charge, which could have increased the adsorption capacity of SMX, which, at higher pH values, is negatively charged ( Liu et al., 2018 ).
The isotherm model, shown in Fig. 4 and fitted for the data for the removal of SMX (2 μg/L) in the presence of NH 4 + ions, describes an appropriate fit to the adsorption of adsorbate at relatively low concentrations.
Apart from the effect that the presence of ammonium ions had on the adsorption of SMX by the zeolite granules, the adsorption of ammonium (initial concentration of 40 mg/L) in the same mixture was tested.From Table 3 it can be seen that there was hardly any effect of SMX on ammonium adsorption even at high concentrations (about 50 mg/L) of SMX.Accordingly, it is assumed that the effect of SMX at a concentration of 2 μg/L on ammonium removal is absent and therefore was not further tested.

Regeneration
To study the effect of initial pH on ammonia removal by ozone oxidation, a series of experiments were conducted with different initial pH values (7, 8 and 9) at an ozone flow rate of 0.7 L/min and an ozone concentration in the gas phase of 90 g/m 3 .The results are shown in Figs. 5 a) and 5 b).
The decrease in the concentration of ammonium in Fig. 5 a) indicates that ammonium was oxidized by ozone at pH 9 according to the reaction ( Singer et al., 1975 ;Khuntia et al., 2012 ): This is in accordance with the alkaline pH range reported to be the most effective for the ozone oxidation of ammonia ( Singer et al., 1975 ;Khuntia et al., 2012 ).Almost no removal of ammonium at initial pH 8 (8% removal) was observed, while the removal efficiency reached 96% at pH 9 after 120 min.It is observed from Fig. 5 b) that the pH value of the solution reduced with reaction time.The decrease in pH was mainly due to continuous production of H + ions during the oxidation reaction ( eq. ( 4) ).Luo et al. (2015) observed the same trend in pH decrease, while they studied the effect of initial pH on ammonia removal by ozone oxidation.They also found that the higher the pH value is, the larger the ammonium removal efficiency.
In order to avoid interference of air stripping of NH 3 ( Mohammed-Nour et al., 2019 ) with the removal of ammonium by oxidation with ozone, the stripping process was tested.The pKa value for ammonia 9.25 indicates that it can be predicted that some electrically neutral ammonia would exist at pH 9 (Danut ė Vai čiukynien ė et al., 2020 ).At pH 9 it was found that no NH 3 was transferred to the atmosphere.This was in accordance with findings from literature where at the higher pH range (pH 10-pH 12) greater proportions of nitrogen in form of gaseous NH 3 are present,   and a larger percentage of NH 3 could be stripped ( Guštin et al., 2011 ;Campos et al., 2013 ).When adsorption equilibrium by MOR granules with NH 4 + was obtained at various initial concentrations, as shown in Fig. 6 , its regeneration was conducted by oxidation using an ozonated water.Fig. 6 shows the results of adsorption/oxidation cycles performed over the ammonium loaded MOR granules.
As can be noticed, after the first regeneration cycle the adsorption capacity dropped for each initial ammonium concentration.The initial adsorption capacity for low initial ammonium concen-tration (10 mg/L) of 0.4235 mg/g was decreased, after three regeneration cycles, by 0.057 mg/g.At the high initial ammonium concentrations (e.g. 100 mg/L) the decrease after first regeneration cycle was higher (from 1.92 mg/g to 1.22 mg/g).However the adsorption capacity after the second and third regeneration cycle recuperated and reached 70% of the initial adsorption capacity.This drop in the adsorption capacity could be related to slow diffusion of ozonated water to inner adsorption sites of the granules, resulting in incomplete regeneration.This is in a good agreement with the study of Zhang et al. (2014) , who concluded that ozone dosages must be sufficiently high to lead to surface modification of zeolites, making the adsorption sites more approachable and available for regeneration and consequently adsorption.

Conclusions
In recent years effort s have been made towards advanced removal of organic micro-pollutants and nutrients from wastewater. .In this study it was demonstrated that NH 4 + adsorption by Zeolita and MOR zeolites, as well as simultaneous adsorption of NH 4 + and SMX by MOR offers an alternative, opening ways to simultaneous removal of NH 4 + ions and SMX (and potentially other micropollutants) from wastewater using synthetic zeolites.Based on the results the following conclusions can be drawn: -Ammonium adsorption capacity on Zeolita granules was found to be between 2.3 -3 mg/g, whereas MOR manifested similar adsorption performance at low initial ammonium concentrations (10-40 mg/L) and lower at high ammonium concentrations (60-100 mg/L).-Presence of ammonium ions enhances SMX adsorption by MOR possibly due to the local structure changes in the zeolite framework upon ion exchange with ammonium.
-Hardly any effect of SMX on ammonium adsorption by MOR was observed, even at high concentrations (50 mg/L) of SMX.-The adsorption capacity of ammonium drastically decreased after the first regeneration cycle by ozonation, probably due to incomplete regeneration.However, after the second and third regeneration cycle adsorption capacity was restored to 86% of the initial capacity at low (10 mg/L) ammonium concentration and to 70% at high (100 mg/L) ammonium concentration.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 5 .
Fig. 5. Ammonium removal by ozonation of NH 4 Cl solutions containing 40 mg/L ammonium at initial pH 7, 8 and 9. a) effect of pH on removal efficiency; b) effect of ozone on ammonium removal and buffer (pH 9).

Table 1
Chemical composition of natural and synthetic zeolite.
0.7 L/min (O 3 concentration in water 25 mg/l).Ozone was produced from pure oxygen with an ozone generator and measured with analysers (Ozone analyzer BMT 964C model, RMG MESSTECH-NIK GmbH, Berlin, Germany), installed before and after the reactor.

Table 2
Langmuir and Freundlich ammonium and SMX adsorption isotherm parameters.

Table 3
Exchange of NH 4 + ions on MOR granule zeolites at 20 °C with the absence and presence of SMX (about 50 mg/L).