Zeolite as a Potential Medium for Ammonium Recovery and Second Cheese Whey Treatment

The efficiency of natural zeolite to remove ammonium from artificial wastewater (ammonium aqueous solutions) and to treat second cheese whey was examined, aiming to recover nitrogen nutrients that can be used for further applications, such as slow-release fertilizers. Sorption experiments were performed using artificial wastewater and zeolite of different granulometries (i.e., 0.71–1.0, 1.8–2.0, 2.0–2.8, 2.8–4.0, and 4.0–5.0 mm). The granulometry of the zeolite had no significant effect on its ability to absorb ammonium. Nevertheless, smaller particles (0.71–1.0 mm) exhibited quicker NH4-N adsorption rates of up to 93.0% in the first 10 min. Maximum ammonium removal efficiency by the zeolite was achieved at ammonium concentrations ranging from 10 to 80 mg/L. Kinetic experiments revealed that chemisorption is the mechanism behind the adsorption process of ammonium on zeolite, while the Freundlich isotherm model fitted the experimental data well. Column sorption experiments under batch operating mode were performed using artificial wastewater and second cheese whey. Column experiments with artificial wastewater showed high NH4-N removal rates (over 96% in the first 120 min) for all granulometries and initial NH4-N concentrations tested (200 and 5000 mg/L). Column experiments with second cheese whey revealed that natural zeolite can remove significant organic loads (up to 40%, 14.53 mg COD/g of zeolite) and NH4-N (about 99%). For PO4-P, the zeolite appeared to saturate after day 1 of the experiments at a removal capacity of 0.15 mg P/g of zeolite. Desorption experiments with water resulted in low NH4-N and PO4-P desorption rates indicating that the zeolite could be used as a substrate for slow nitrogen release in soils.


Introduction
Agro-industrial development is considered to be major contributor of environmental pollution [1]. Agro-industries (e.g., dairies, olive mills, wineries) produce significant quantities of wastewaters characterized by high organic loads and nutrients [2]. Therefore, treatment of these wastewaters is necessary before their discharge.

Artificial Wastewater and Secondary Cheese Whey
Artificial wastewater (AWW) was prepared by drying commercially available ammonium chloride at 104 • C for 24 h and then diluting it with deionized water (DW) to concentrations ranging from 1 to 5000 mg/L. The secondary cheese whey (SCW) used in this study was obtained from the factory of Papathanasiou S.A. (Agrinio, Greece), transported to the laboratory in plastic bottles and then stored at −20 • C until use at −20 • C to prevent any change in its composition. The COD concentration of the SCW was 28.500 mg/L, total Kjeldahl nitrogen (TKN) was 243 mg/L, NH 4 + was 45.27 mg/L, PO 4 3− was 196.69 mg/L, and its pH was recorded as 4.2.

Zeolite Characterization
The natural zeolite used in this research originated from Imerys Minerals Bulgaria AD (Kardzhali, Bulgaria) and was imported into Greece by Zeolife TM Company (Orestiada, Greece). Specifically, it is a hydrated aluminosilicate mineral of volcanic origin, with a clinoptilolite content of 85%, a moisture content of 10%-12%, while the remaining 3%-5% comprises impurities of feldspar, micas, and clays free of fibers and quartz. The chemical composition of the dry clinoptilolite is presented in Table 2. According to the technical data sheet provided, the minimum cation exchange capacity value of this specific zeolite is 150 meq/100 g. Prior to all experiments the zeolite was sieved to obtain different granulometries (i.e., 0.71-1.0, 1.8-2.0, 2.0-2.8, 2.8-4.0, and 4.0-5.0 mm), washed to remove dust, and finally dried.

Zeolite Characterization
Batch tests were carried out in 250 mL borosilicate bottles filled with 200 mL of AWW at low (5 mg/L) and high (200 mg/L) concentration, and 10 g of zeolite of different particle sizes (i.e., 0.71-1.0, 1.8-2.0, 2.0-2.8, 2.8-4.0, and 4.0-5.0 mm). Bottles were stirred at 200 rpm and samples were taken at regular intervals until equilibrium was achieved. In order to generate isotherm models (see Section 2.4) various concentrations of AWW were tested, at a range of 1 to 200 mg/LNH 4 + -N for the above mentioned granulometries. To determine the maximum amount of ammonium adsorbed per mass of zeolite at equilibrium (q e ), additional batch adsorption experiments were conducted for zeolite of grain size 4-5 mm, using higher ammonium feed concentrations (i.e., 500, 1000, 2000, 3000, 4000, and 5000 mg/L). Desorption was tested by placing the used zeolite sample (approximately 10 g) into a volumetric flask containing 200 mL of deionized water. The flasks were stirred at 200 rpm and samples were taken at regular intervals until equilibrium was reached [49].

Ammonium Adsorption Kinetics
The kinetics of ammonium adsorption by the zeolite was mathematically analyzed by applying pseudo-first order [50] and pseudo-second order [51] models according to the following nonlinearized equations: where q e (mg/g) and q t (mg/g) denote the amount of ammonium adsorbed per mass of zeolite at equilibrium and at any time t (min), respectively, and k 1 (1/min) and k 2 (g/mg min) are the rate constants for the pseudo-first order and pseudo-second order models.

Ammonium Isothermal Curves
The relationship between the ammonium concentration in solution and the various zeolite granulometries was studied using two adsorption isotherm models. The experimental equilibrium data were analyzed by applying isotherms with two parameter equations, Freundlich and Langmuir, that are frequently used in the literature [52,53].
The Freundlich isotherm describes heterogeneous adsorption systems [54] and the non-linearized form of this isotherm is expressed by the following equation: where q e (mg/g) is the amount of ammonium adsorbed on the zeolite at equilibrium, Ce (mg/L) is the ammonium concentration at equilibrium, k F (mg/g)/(mg/L) n is the Freundlich constant, and n (dimensionless) is the Freundlich intensity parameter which indicates the magnitude of the adsorption driving force or the surface heterogeneity.
The Langmuir isotherm has been commonly used for the removal of a wide variety of compounds from waters using different adsorbents. It assumes monolayer coverage of the adsorbate over a homogeneous adsorbent surface [55]. The non-linearized model of Langmuir is described as: where q e (mg/g) is the amount of ammonium adsorbed on the zeolite at equilibrium, q m (mg/g) is the maximum monolayer adsorption capacity of the zeolite, C e (mg/L) denotes the equilibrium concentrations of ammonium, and k L (L/mg) is a constant related to the affinity between the adsorbent and adsorbate. The linearized model of Langmuir is described by the following equation: The Langmuir isotherm model expressed extensively in terms of the separation factor or equilibrium parameter (R L ) [56] can be given as: where C 0 (mg/L) is the initial maximum concentration of the analyzed ammonium. The nature of the adsorption can be estimated through the isotherm profile according to the values of R L and the Freundlich exponent (n), being irreversible (R L = 0), favorable (0 < R L < 1), linear (R L > 1), or unfavorable (R L > 1) [57]. The Langmuir theory can be applied to homogeneous adsorption where each adsorbed species involves the same sorption activation energy.
Fitting of the kinetic and isotherm equations to the experimental data was evaluated using the coefficient of determination (R 2 ).

Adsorption and Desorption Tests in Laboratory-Scale Columns Using AWW and SCW
Adsorption experiments were carried out in laboratory-scale columns. The columns used were Plexiglas tubes, 40.0 cm high and 4.0 cm internal diameter, equipped with four sample valves placed at different heights. The columns were first operated in batch mode and were filled with 500 g zeolite of various granulometries (as described in Section 2.2) and fed with 380 mL of either AWW at concentrations of 200 and 5000 mg/L or SCW diluted with tap water to different dilution ratios (i.e., 100%, 75%, and 50%). Samples were collected at regular intervals from the lower sample valve.
To achieve complete zeolite saturation, the columns were fed with an aqueous solution of ammonium chloride of concentration 5000 mg/L. Subsequently, the columns were fed with 380 mL of deionized water to study the desorption capacity of the zeolite [49]. Samples were taken at regular time intervals until equilibrium was achieved.
Chemical oxygen demand (COD) was determined in filtered samples (0.45 µm, Whatman) using the "closed reflux" method following Standard Methods [59]. A HANNA Reactor C98000 was applied to measure digestion and absorbance was recorded using a Multiparameter Bench Photometer (HANNA C99, Hanna Instruments, Woonsocket, RI, USA). Total Kjeldahl nitrogen (TKN) was measured using the titrimetric method according to Standard Methods [59]. Finally, pH and conductivity were measured with a multiparameter meter (HANNA HI5521, Hanna Instruments, Woonsocket, RI, USA). All the reagents used in this study were of analytical grade.

Effect of Particle Size
Initially, the effect of zeolite particle size on its adsorbent capacity was tested for AWW of concentration 5 mg NH 4 + -N mg/L. NH 4 + -N is mainly removed with zeolite through ion exchange activity, which follows the order NH 4 + > Na + > Ca 2+ [40]. According to the results, no great differences were recorded between the granulometries tested ( Figure 1). For each grain size tested, ammonium removal by zeolite occurred rapidly within the first 60 min of contact time (removals ranged from 85.6%-98.8% depending on the granulometry). The removal rate of ammonium was initially fast then gradually decreased until equilibrium state which was achieved at 1440 min. Nevertheless, the finest zeolite particles (0.71-1.0 mm) demonstrated quicker NH 4 + -N adsorption, especially during the first 60 min (98.8%). This is attributed to their higher specific surface area which increases external surface adsorption, as also reported in other relative studies [29,32,[60][61][62]. According to Wang et al. [62], grain size affects the amount of seawater NH 4 + -N adsorbed on zeolite, with smaller sized granules giving better results. However, Leyva-Ramos et al. [63] reported that decreasing the grain size does not significantly increase the adsorption capacity of zeolite. Zeolite is a porous material and interacts with the solution with both external and internal (pores and channels) surface area. By reducing particle size the external area increases considerably but the same does not apply for the internal area of the zeolite. Moreover, its exchange capacity is dependent on the cationic sites which are located inside the pores of the zeolite and the channels and cavities of the zeolite crystalline structure and not on its external area [64,65].
Water 2019, 11, x FOR PEER REVIEW 9 of 23 applied to measure digestion and absorbance was recorded using a Multiparameter Bench Photometer (HANNA C99, Hanna Instruments, Woonsocket, RI, USA). Total Kjeldahl nitrogen (TKN) was measured using the titrimetric method according to Standard Methods [59]. Finally, pH and conductivity were measured with a multiparameter meter (HANNA HI5521, Hanna Instruments, Woonsocket, RI, USA). All the reagents used in this study were of analytical grade.

Effect of Particle Size
Initially, the effect of zeolite particle size on its adsorbent capacity was tested for AWW of concentration 5 mg NH4 + -N mg/L. NH4 + -N is mainly removed with zeolite through ion exchange activity, which follows the order NH4 + > Na + > Ca 2+ [40]. According to the results, no great differences were recorded between the granulometries tested ( Figure 1). For each grain size tested, ammonium removal by zeolite occurred rapidly within the first 60 min of contact time (removals ranged from 85.6%-98.8% depending on the granulometry). The removal rate of ammonium was initially fast then gradually decreased until equilibrium state which was achieved at 1440 min. Nevertheless, the finest zeolite particles (0.71-1.0 mm) demonstrated quicker NH4 + -N adsorption, especially during the first 60 min (98.8%). This is attributed to their higher specific surface area which increases external surface adsorption, as also reported in other relative studies [29,32,[60][61][62]. According to Wang et al. [62], grain size affects the amount of seawater NH4 + -N adsorbed on zeolite, with smaller sized granules giving better results. However, Leyva-Ramos et al. [63] reported that decreasing the grain size does not significantly increase the adsorption capacity of zeolite. Zeolite is a porous material and interacts with the solution with both external and internal (pores and channels) surface area. By reducing particle size the external area increases considerably but the same does not apply for the internal area of the zeolite. Moreover, its exchange capacity is dependent on the cationic sites which are located inside the pores of the zeolite and the channels and cavities of the zeolite crystalline structure and not on its external area [64,65].
Taking into consideration the above and that the present study aimed to apply the zeolite in adsorption columns, a grain size of 4-5 mm was selected for later experiments. Zeolite grains smaller than 4 mm may cause clogging when used as substrate in columns, trickling filters and constructed wetlands.  Taking into consideration the above and that the present study aimed to apply the zeolite in adsorption columns, a grain size of 4-5 mm was selected for later experiments. Zeolite grains smaller than 4 mm may cause clogging when used as substrate in columns, trickling filters and constructed wetlands.

Effect of Contact Time
Adsorption of ammonium on zeolite was studied at various time intervals for low (5 mg/L) and high (200 mg/L) initial concentrations of AWW, and the results obtained are presented in Figure 2. The amount of ammonium retained by the zeolite increased gradually until adsorbent saturation while equilibrium was reached within 180 min for the low initial AWW concentration and within 300 min for the high initial AWW concentration. The removal efficiency of ammonium ions by the zeolite was initially fast, with up to 85% being achieved within the first 60 min for AWW of 5 mg/L. The rate then slowed considerably with increased contact time and remained almost constant before reaching equilibrium. After 2880 min ammonium removal had reached up to 99% for both initial AWW concentrations tested. These results are similar to those recorded in other relevant studies [66,67]. Using batch experiments, Martins et al. [66] reported that ammonium removal by natural zeolite (originating from Cuba) occurred within the first 180 min of contact time. Removal rates were initially fast then gradually decreased with increased contact time. Alshameri et al. [67] observed that natural zeolite from China (Hulaodu) exceeded 78% removal rate at 300 min but then plateaued. This behavior may be attributed to the rapid utilization of the zeolite's available adsorbing sites followed by fast diffusion into its pores and channels until equilibrium is achieved [29,68]. Adsorption of ammonium on zeolite was studied at various time intervals for low (5 mg/L) and high (200 mg/L) initial concentrations of AWW, and the results obtained are presented in Figure 2. The amount of ammonium retained by the zeolite increased gradually until adsorbent saturation while equilibrium was reached within 180 min for the low initial AWW concentration and within 300 min for the high initial AWW concentration. The removal efficiency of ammonium ions by the zeolite was initially fast, with up to 85% being achieved within the first 60 min for AWW of 5 mg/L. The rate then slowed considerably with increased contact time and remained almost constant before reaching equilibrium. After 2880 min ammonium removal had reached up to 99% for both initial AWW concentrations tested. These results are similar to those recorded in other relevant studies [66,67]. Using batch experiments, Martins et al. [66] reported that ammonium removal by natural zeolite (originating from Cuba) occurred within the first 180 min of contact time. Removal rates were initially fast then gradually decreased with increased contact time. Alshameri et al. [67] observed that natural zeolite from China (Hulaodu) exceeded 78% removal rate at 300 min but then plateaued. This behavior may be attributed to the rapid utilization of the zeolite's available adsorbing sites followed by fast diffusion into its pores and channels until equilibrium is achieved [29,68].

Effect of Initial Concentration
The effect of initial ammonium concentration on zeolite's removal efficiency was studied and the results are shown in Figure 3. The experimental data indicated that the maximum removal efficiency of the zeolite was achieved with ammonium concentrations ranging from 10 to 80 mg/L, while increasing the initial ammonium concentration decreased its removal percentage ( Figure 3). Similar results were observed by Sarioglu [69] who reported that maximum removal efficiency was achieved at concentrations of 8.8 to 40 mg/L, leading to the conclusion that initial ammonium concentration plays a key role in the adsorption mechanism of ammonium on zeolite. Huang et al. [29] also observed that the efficiency of a zeolite (Chende zeolite from China) to remove ammonium (initial NH4 + concentration was 80 mg/L) decreased when the concentration of ammonium in solution was increased. Alshameri et al. [67] also reported that maximum ammonium removal was achieved at initial concentrations ranging from 10 to 50 mg/L. These results can be expected for porous materials such as zeolite. Increasing the initial ammonium concentration leads to an increase of the mass transfer driving force and consequently the rate at which ammonium ions pass from the bulk solution to the surface of the zeolite. Above a critical concentration, where the driving force is high,

Effect of Initial Concentration
The effect of initial ammonium concentration on zeolite's removal efficiency was studied and the results are shown in Figure 3. The experimental data indicated that the maximum removal efficiency of the zeolite was achieved with ammonium concentrations ranging from 10 to 80 mg/L, while increasing the initial ammonium concentration decreased its removal percentage (Figure 3). Similar results were observed by Sarioglu [69] who reported that maximum removal efficiency was achieved at concentrations of 8.8 to 40 mg/L, leading to the conclusion that initial ammonium concentration plays a key role in the adsorption mechanism of ammonium on zeolite. Huang et al. [29] also observed that the efficiency of a zeolite (Chende zeolite from China) to remove ammonium (initial NH 4 + concentration was 80 mg/L) decreased when the concentration of ammonium in solution was increased. Alshameri et al. [67] also reported that maximum ammonium removal was achieved at initial concentrations ranging from 10 to 50 mg/L. These results can be expected for porous materials such as zeolite.
Increasing the initial ammonium concentration leads to an increase of the mass transfer driving force and consequently the rate at which ammonium ions pass from the bulk solution to the surface of the zeolite. Above a critical concentration, where the driving force is high, the ammonium ions migrate from the zeolite's surface to its core (pores and channels) and when all the ion exchange processes are completed equilibrium state is achieved [70,71].

Kinetic Models
To comprehend the mechanism involved during the adsorption of ammonium on zeolite and the potential controlling steps, two kinetic models were applied and evaluated, namely, pseudo-first order and pseudo-second order. The best fit model was based on both the linear regression correlation coefficient (R 2 ) and the calculated q e values. According to the results obtained (Table 3), low R 2 values and notable variances between the experimental and theoretical uptakes reveal clearly the poor fitting of the pseudo-first order model. On the contrary, the pseudo-second model shows better fitting, for all the initial tested ammonium concentration, which is in agreement with previous research [29,72,73]. A fitting to the pseudo-second order reveals that chemisorptions dominate the adsorption process of ammonium on zeolite, and that this process involves three stages: (i) diffusion from the liquid state to the liquid-solid interface, (ii) movement of ammonium ions toward the solid surface, and (iii) movement of ammonium ions to the pores and channels of the zeolite [74]. the ammonium ions migrate from the zeolite's surface to its core (pores and channels) and when all the ion exchange processes are completed equilibrium state is achieved [70,71].

Kinetic Models
To comprehend the mechanism involved during the adsorption of ammonium on zeolite and the potential controlling steps, two kinetic models were applied and evaluated, namely, pseudo-first order and pseudo-second order. The best fit model was based on both the linear regression correlation coefficient (R 2 ) and the calculated qe values. According to the results obtained (Table 3), low R 2 values and notable variances between the experimental and theoretical uptakes reveal clearly the poor fitting of the pseudo-first order model. On the contrary, the pseudo-second model shows better fitting, for all the initial tested ammonium concentration, which is in agreement with previous research [29,72,73]. A fitting to the pseudo-second order reveals that chemisorptions dominate the adsorption process of ammonium on zeolite, and that this process involves three stages: (i) diffusion from the liquid state to the liquid-solid interface, (ii) movement of ammonium ions toward the solid surface, and (iii) movement of ammonium ions to the pores and channels of the zeolite [74].  Table 3. Fitting kinetics parameters of adsorption of ammonium onto zeolite according to pseudofirst order and pseudo-second order models at various N-NH4 + concentrations (zeolite granulometry: 4.0-5.0 mm).

Equilibrium Isotherms
The two most popular isothermal models, Langmuir and Freundlich (see Section 2.4.2) were employed to describe the equilibrium adsorption isotherms. The estimated parameters and error function values for the applied adsorption isotherms are shown in Table 4. The Freundlich isotherm model shows higher relative R 2 values, thus indicating that the Freundlich model fits the experimental data better than the Langmuir model. A 1/n value of less than 1 indicates that NH 4 + -N adsorbance on zeolite is favorable.
However, according to the literature, the optimal adsorption isotherm for NH 4 + adsorption on zeolite with different characteristics can vary. Many studies suggest that the Freundlich isothermal model fits better in describing NH 4 + -N adsorption on zeolite (natural or modified) [15,23,24,29,48,75], while others propose that the Langmuir model has the better fitting [24,30,37,76].

Desorption Batch Tests
Natural zeolite is particularly useful for controlled agricultural soil fertilization as it provides the slow release of nutrients, such as N and P, which help ensure high crop yields. For this reason, desorption experiments were performed to test its suitability for direct use as a soil additive. The desorption tests were conducted using deionized water and the results showed that ammonium release was remarkably slow and final desorption equilibrium was achieved after 7200 min, at which time only 1.59% of the absorbed ammonium had been released. These results may be explained by the characteristics of the aqueous solution used in the desorption experiments [41] since the solution lacks any ion exchange capacity to release and replace ammonium anions. According to the literature, a strong ion exchange solution is needed to regenerate used zeolite, such as sodium hydroxide that can increase NH 4 + -N desorption rates by up to 99% [42].

Adsorption and Desorption Experiments on Laboratory Scale Column Using AWW
Following the batch adsorption experiments, column experiments were carried out to treat AWW. Previous column adsorption experiments (using zeolitic adsorbents) have shown that fixed-bed columns are more efficient than fluidized bed columns [22,77]. Therefore, fixed-bed column experiments were performed here to study the removal of ammonium from AWW and then treat SCW. Figure 4 presents the results of NH 4 + -N adsorption obtained in column kinetic experiments (under batch operation) for an initial ammonium concentration of 200 mg/L using natural zeolite of various granulometries. Each experimental column contained 500 g of zeolite and was filled with 380 mL of NH 4 + -N aqueous solution. For all granulometries tested the solution covered the zeolite entirely.
It was observed that ammonium ion concentrations decreased more sharply in the experiments with smaller-sized fractions. This result was expected because total surface area increases as particle size decreases. Specifically, the times required to meet the discharge standard of 5 mg/L (discharge limit for total nitrogen) were: <5 min for granulometries of 0.71-1.0 mm, 20 min for 1.8-2.0 mm, 30 min for 2.0-2.8 mm, and >60 min for 2.8-4.0 mm and 4.0-5.0 mm. For all the particle sizes tested, the zeolite was able to absorb almost all available NH 4 + -N (above 99%) in the first 120 min, while 1440 min were required to reach equilibrium (99.7% of the NH 4 + -N was absorbed). However, similar results were not observed when the initial concentration of NH 4 + -N was 5000 mg/L ( Figure 5). At this concentration, only zeolite with 0.71-1.0 mm sized particles was able to remove over 99% of the NH 4 + -N (99.9%, 3.8 mg/g) in 120 min. All the other granulometries established equilibrium much later, at 1440 min, mg/g falls within the range of the recorded ammonium capacity for natural zeolites (2.7-22.9 mg/g) [60]. Following the same procedure as the batch sorption tests, experiments were then performed to investigate the degree of desorption of sorbed ammonium from the zeolite using deionized water. These experiments aimed to study the slow release of nitrogen [49] and were performed for all granulometries and an initial NH4 + -N concentration of 5000 mg/L. Specifically, 380 mL of deionized water was added into each column. The results are presented in Table 5 and are in line with those of the adsorption experiments (Table 4), as finer-grained zeolite (i.e., 0.70-1.0 mm and 1.8-2.0 mm) desorbed higher amounts of NH4 + -N (over 4% of the absorbed NH4 + -N) than coarser zeolite (i.e., 2.0-2.8, 2.8-4.0, and 4.0-5.0 mm) (between 2.77% and 3.8% of the absorbed NH4 + -N). Finally, the NH4 + -N desorption equilibrium was reached after 11,520 min (192 h). The rates of NH4 + -N desorption recorded in the column experiments were also low compared with those of previous related studies in which tap or deionized water was used for desorption [49,78]. Specifically, Cyrus and Reddy [49] showed that after 250 h of desorption with water, a significant amount of nitrogen remained in the column (>20%) and was available for desorption using only 0.1 N HCl, thus confirming that zeolite could be a suitable medium for slow nitrogen release in soils. Hedstrom and Amofah [78] also studied NH4 + -N desorption and found that only 23% of NH4 + -N was desorbed by tap water, while desorption was more pronounced in saturated conditions.   Following the same procedure as the batch sorption tests, experiments were then performed to investigate the degree of desorption of sorbed ammonium from the zeolite using deionized water. These experiments aimed to study the slow release of nitrogen [49] and were performed for all granulometries and an initial NH 4 + -N concentration of 5000 mg/L. Specifically, 380 mL of deionized water was added into each column. The results are presented in Table 5 and are in line with those of the adsorption experiments (Table 4), as finer-grained zeolite (i.e., 0.70-1.0 mm and 1.8-2.0 mm) desorbed higher amounts of NH 4 + -N (over 4% of the absorbed NH 4 + -N) than coarser zeolite (i.e., 2.0-2.8, 2.8-4.0, and 4.0-5.0 mm) (between 2.77% and 3.8% of the absorbed NH 4 + -N). Finally, the NH 4 + -N desorption equilibrium was reached after 11,520 min (192 h). The rates of NH 4 + -N desorption recorded in the column experiments were also low compared with those of previous related studies in which tap or deionized water was used for desorption [49,78]. Specifically, Cyrus and Reddy [49] showed that after 250 h of desorption with water, a significant amount of nitrogen remained in the column (>20%) and was available for desorption using only 0.1 N HCl, thus confirming that zeolite could be a suitable medium for slow nitrogen release in soils. Hedstrom and Amofah [78] also studied NH 4 + -N desorption and found that only 23% of NH 4 + -N was desorbed by tap water, while desorption was more pronounced in saturated conditions.

Sorption Column Studies Using SCW
Column sorption experiments under batch operation were also performed using pre-treated SCW. The pre-treatment step included a biological trickling filter and the COD concentration of the SCW used was 29,000 mg/L. A series of experiments was carried out in columns using SCW at different dilution ratios (100%, 75%, and 50%) to determine the feasibility of the adsorption process using the specific zeolite. The three different columns used for this experiment were filled with zeolite of the coarser grain (4.0-5.0 mm) to prevent pore clogging. The first column was fed with undiluted SCW, the second was fed with a solution of 75% SCW and 25% tap water, and the third was fed with a solution of 50% SCW and 50% tap water. The columns were fed with fresh wastewater every day (380 mL on day one, 320 mL on day two, and 310 mL on day three) as the water was absorbed by the zeolite, and samples were taken at specific time intervals to determine the kinetics of COD and NH 4 + -N removal. The total duration of the experiments was three days. Figure 6 presents the time series data for COD removal in the three columns. A short, three-day, experimental duration was chosen to avoid the formation of biofilm that could interfere with the zeolite's adsorption ability. All three columns presented the same pattern concerning COD removal, as maximum removal was observed in day one of the experiments (i.e., 36.89 for 100% SCW (7.7 mg COD/g absorbed); 24.33% for 75% SCW (3.78 mg COD/g absorbed); 23.14% for 50% SCW (2.52 mg COD/g absorbed). COD removal rates were observed to decrease in the following two days (Day two: 19.29% for 100% SCW (3.52 mg COD/g absorbed); 20% for 75% SCW (2.88 mg COD/g absorbed); 19.31% for 50% SCW (1.79 mg COD/g absorbed); Day three: 17.63 for 100% SCW (3.31mg COD/g absorbed); 17.11% for 75% SCW (2.38 mg COD/g absorbed); 15.86% for 50% SCW (1.43 mg COD/g absorbed)). The total adsorption capacity of the zeolite recorded in these experiments was 14.53, 9.04, and 5.74 mg COD/g of zeolite for 100%, 75%, and 50% SCW, respectively. Although COD removal decreased with time, the zeolite was probably not fully saturated as it continued to remove organic matter. Nevertheless, the decrease in COD removal does indicate partial zeolite saturation.
COD/g absorbed). COD removal rates were observed to decrease in the following two days (Day two: 19.29% for 100% SCW (3.52 mg COD/g absorbed); 20% for 75% SCW (2.88 mg COD/g absorbed); 19.31% for 50% SCW (1.79 mg COD/g absorbed); Day three: 17.63 for 100% SCW (3.31mg COD/g absorbed); 17.11% for 75% SCW (2.38 mg COD/g absorbed); 15.86% for 50% SCW (1.43 mg COD/g absorbed)). The total adsorption capacity of the zeolite recorded in these experiments was 14.53, 9.04, and 5.74 mg COD/g of zeolite for 100%, 75%, and 50% SCW, respectively. Although COD removal decreased with time, the zeolite was probably not fully saturated as it continued to remove organic matter. Nevertheless, the decrease in COD removal does indicate partial zeolite saturation. The ability of zeolites to absorb organic matter has been previously reported in various studies [17,18,39]. Kolakovic et al. [17] reported that COD reduction from dairy wastewater using organozeolite and a filter column ranged from 30% to 50% when a mean initial COD concentration of 2676 mg/L was applied. Nevertheless, the exact mechanism of COD removal was not examined by Kolakovic et al. [17] and was attributed to the zeolite's ability to absorb both organic and inorganic substances. Samkutty and Gough [18] reported higher COD removal (76%) when applying dairy processing wastewater with low initial COD concentrations (between 713 and 1410 mg/L), however provided no data on the zeolite used. High COD removal (71.8%) was also recorded by Guo et al. [48], however this research studied the treatment of swine wastewater using a composite of MgOmodified zeolite and a bioflocculant rather than a natural zeolite. Lakdawala and Patel [44] studied the use of natural zeolite for the removal of both COD and BOD from sugar industry wastewater and reported COD removal rates of up to 26.6%. These values are similar to those presented in this study. However, direct comparison of the above results with previous studies is not currently possible, since almost all adsorption experiments with zeolite were performed in batch systems (flasks) or in columns under continuous operation and not in columns under batch operation.
To investigate the possible removal of COD by biodegradation, a void column was loaded only with SCW and was operated under the same experimental conditions as the other three columns. At the end of day one COD concentration had decreased by about 8% (due to the indigenous The ability of zeolites to absorb organic matter has been previously reported in various studies [17,18,39]. Kolakovic et al. [17] reported that COD reduction from dairy wastewater using organo-zeolite and a filter column ranged from 30% to 50% when a mean initial COD concentration of 2676 mg/L was applied. Nevertheless, the exact mechanism of COD removal was not examined by Kolakovic et al. [17] and was attributed to the zeolite's ability to absorb both organic and inorganic substances. Samkutty and Gough [18] reported higher COD removal (76%) when applying dairy processing wastewater with low initial COD concentrations (between 713 and 1410 mg/L), however provided no data on the zeolite used. High COD removal (71.8%) was also recorded by Guo et al. [48], however this research studied the treatment of swine wastewater using a composite of MgO-modified zeolite and a bioflocculant rather than a natural zeolite. Lakdawala and Patel [44] studied the use of natural zeolite for the removal of both COD and BOD from sugar industry wastewater and reported COD removal rates of up to 26.6%. These values are similar to those presented in this study. However, direct comparison of the above results with previous studies is not currently possible, since almost all adsorption experiments with zeolite were performed in batch systems (flasks) or in columns under continuous operation and not in columns under batch operation.
To investigate the possible removal of COD by biodegradation, a void column was loaded only with SCW and was operated under the same experimental conditions as the other three columns. At the end of day one COD concentration had decreased by about 8% (due to the indigenous microorganisms existing in the SCW), thus confirming the hypothesis that COD in the other three columns was removed mainly through adsorption by the zeolite. Adsorption of PO 4 3− -P was observed to be less than that of NH 4 + -N, as the zeolite appears to become saturated after day one (day one: 77%-79%, day two: 20%-30%, and day three: 7%-17% for all three SCW dilution ratios), this phenomenon was expected as zeolite ability on adsorbing PO 4 3− P is rather limited compared to its ability adsorbing NH 4 + -N [17,34]. PO 4 3− -P is mainly adsorbed on zeolite through an inner-sphere complexation [28]. The maximum adsorption capacity of the zeolite for PO 4 3− -P (days one to three) was 0.15 mg P/g zeolite, and was observed in experiments with raw SCW (100%). Zeolite's low adsorption capacity for PO 4 3− -P was also reported by Drizo et al. [79], who recorded a maximum capacity of 0.46 mg P/g. Lin et al. [33] stated that zeolite's ability for PO 4 3− -P adsorption increases when pH values exceed the value of 9, while for neutral pH values adsorption capacity is limited to 0.30 mg P/g of zeolite. Adsorption capacity for PO 4 3− -P could be increased to 8.2 mg P/g when the zeolite is impregnated with hydrated metal oxides [26], or to 17.2 mg P/g when zeolite is modified with lanthanum oxide [28]. Although modified zeolites show up to 400% improved adsorption capacity for PO 4 3− -P, this value is still low (1.58 mg P/g of zeolite) compared to that for NH 4 + -N [34]. The relatively limited ability of zeolite for PO 4 3− -P removal was also reported by Kolakovic et al. [17], who examined its use in dairy wastewater treatment and reported extremely low phosphorus removal (approximately 20%), even though the zeolite was modified with stearin-dimethyl-benzyl ammonium chloride.
To determine the COD desorption rate, the experimental columns were filled with deionized water (DI) which was replaced every 24 h, while COD was measured at regular intervals. After two days, almost all the absorbed COD was recovered, indicating that organic substances are not firmly bound onto zeolite grains. In contrast to COD, zeolite desorbed part of the absorbed NH 4 + -N (1%) and PO 4 3− -P (23.2%). Therefore, the use of an appropriate quantity of zeolite in soils may promote the slow release of N from zeolite without the desorbed COD affecting plant growth.

Effect of Temperature and Biofilm Growth
During the three-day column experiments, the zeolite did not become saturated, as it was observed to remove most of the nitrogen present. Therefore, additional prolonged column experiments were conducted using initial COD concentrations of 2600 ± 200 mg/L. The experimental set-up was identical to that described above, extending only the experimental duration to six days. Furthermore, three additional columns were operated inside a refrigerator (4 • C) to prevent biofilm formation and inhibit any biological processes. Table 7 presents the mean values and standard deviations of all pollutant concentrations at the start and end of each experimental day. Concerning COD, final concentrations from day 1 to day 3 did not present great differences (COD removal 16%-23%), from those observed during the previous three-day experiments, thus verifying that no significant biofilm growth occurred in the columns in the first three days. However, from day four to six, the columns operating at ambient temperature (approximately 22 • C) achieved higher COD removal rates (28%-36%) than the refrigerated columns (16%-22%), indicating that biodegradation of organic matter occurs from day four onward.
TKN and PO 4 3− -P adsorption by zeolite do not seem to be greatly affected either by temperature or by biofilm formation, as final concentrations of both pollutants did not differ significantly between the refrigerated columns and those at ambient temperature throughout the experiment. It is possible however that the concentrations of these pollutants were so low that any decrease in them could not be detected.

Conclusions
In this study, natural Bulgarian zeolite was investigated for its effectiveness to remove ammonium from synthetic wastewater and to treat second cheese whey. Batch experiments were performed in both flasks and fixed-bed columns to examine the effect of various operating conditions on zeolite performance. The conclusions reached from this study are:

•
Batch experiments revealed that zeolite granulometry had no significant effect on its ability to absorb ammonium, while maximum removal efficiency was achieved at ammonium concentrations ranging from 10 to 80 mg/L.

•
The pseudo-second order model fitted the experimental data thus revealing that chemisorption is the mechanism for the adsorption process of ammonium on zeolite. The Freundlich isotherm model best fit the experimental data. • Experiments in columns using synthetic wastewater with an initial NH 4 + -N concentration of 200 mg/L revealed that for all granulometries tested, the zeolite was able to absorb almost all available NH 4 + -N (over 99%) in the first 120 min. The same was not observed for initial concentrations of 5000 mg/L where only zeolite with grain sizes of 0.71-1.0 mm managed to remove over 99% NH 4 + -N (99.9%) in 120 min. All other zeolite granulometries reached equilibrium much later (24 h, 96%-97% removal of NH 4 + -N).
• Natural zeolite may be used as an alternative substrate for second cheese whey treatment, as significant removal of organic load (up to 40%, 14.53 mg COD/g of zeolite) and NH 4 + -N (about 99%) can be achieved. Concerning PO 4 3− -P, the zeolite appeared to saturate after day one of the experiments at a removal capacity of 0.15 mg P/g of zeolite. Desorption experiments with deionized water in batch and columns presented low desorption rates for NH 4 + -N and PO 4 3− -P, thus indicating that this zeolite could be used as substrate for slow nitrogen release in soils.

•
Prolonged use of zeolite in SCW treatment (after three days) led to the formation of biomass by increasing the percentage of organic load removal.    Funding: This research received no external funding.