Potential Application of Chilean Natural Zeolite as a Support Medium in Treatment Wetlands for Removing Ammonium and Phosphate from Wastewater

: This study aims to evaluate the sorption characteristics of NH + 4 -N and PO 3 − 4 -P onto the surface of natural zeolites coming from Chile and their potential application in the subsurface-ﬂow treatment wetlands for wastewater treatment in rural areas. For this purpose, adsorption experiments onto the zeolite were developed in batch assays. The e ﬀ ects of the adsorbent quantity (20 g and 50 g) and particle size (0.2–1.0 mm; 1.5–3.0 mm, and 5.0–8.0 mm) were evaluated in terms of adsorption capacity at di ﬀ erent NH 4 + -N and PO 4 − 3 -P concentrations. Then, the obtained laboratory results were adjusted to theoretical models: Saturation-growth-rate and Langmuir. The saturation adsorption of NH 4 + -N on the zeolite increases at the same time that the initial concentration increases for the same zeolite quantity; however, the saturation values were similar between the di ﬀ erent zeolite sizes tested. For PO 4 − 3 -P, the adsorption did not have a direct relationship with the initial concentration nor zeolite quantity and better results were only achieved for zeolite sizes of 1.5–3.0 mm. Regarding the Langmuir model, sizes of 1.5–3.0 mm had the best adsorption characteristics, with the maximum adsorption capacity of up to 1.58 mg / g for NH 4 + -N and up to 0.08 mg / g for PO 4 − 3 -P. Therefore, a new material—a natural zeolite from the Maule Region of Chile—is described as a potential support medium for treatment wetlands. hand, the adsorptions of NH 4 + -N and PO 4 − 3 -P were described for the Langmuir isotherm and the results showed that the commercial zeolite size of 1.5–3.0 mm was the size that had better adsorption behavior for NH 4 + -N and PO 4 − 3 -P, with maximum adsorption capacities up to 1.58 mg / g and 0.08 mg / g, respectively, as well as the greatest potential for the removal of these elements contained in wastewater. These results show that these two compounds


Introduction
Subsurface-flow treatment wetlands (SSF-TWs) are an alternative to traditional activated sludge systems for the treatment of wastewater in rural areas [1,2]. When SSF-TWs are used in wastewater Regarding physical characteristics, natural zeolites have a Brunauer-Emmett-Teller (BET) surface area of 446 to 480 m 2 /g and a cation exchange capacity (CEC) of 86.82 to 112.88 cmol/kg.

Batch Assays for the Study of Adsorption Characteristics
Adsorption experiments onto the zeolite were developed in batch assays [10,35]. The effects of adsorbent quantity (zeolite quantity) and particle size were evaluated in terms of adsorption capacity [36,37]. Two different adsorbent quantities were studied-20 g and 50 g-in agreement with Andrés et al. [10]. Three zeolite sizes were tested: a) 0.2-1.0 mm, b) 1.5-3.0 mm, and c) 5.0-8.0 mm. The batch assays are summarized in Table 2. These experiments were carried out in random order. The selection of zeolite size was based on commercial availability and on the fact that natural zeolite sizes, ranging from 0.5 to 10 mm, have previously been employed in some SSF-TWs for wastewater treatment [38][39][40]. Table 2. Experimental design used in NH + 4 -N and PO 3− 4 -P assays for the determination of adsorption characteristics. In trials, the initial NH + 4 -N concentrations were evaluated with solutions containing 25, 45, 65, and 85 mg/L (Table 1). In the case of PO 3− 4 -P, the initial concentrations were evaluated with solutions containing 1, 5, 10, and 15 mg/L (Table 1). These initial concentrations are typically found in domestic wastewaters, including those produced in rural areas in Chile [15,18,26]. Each aliquot was prepared in an Erlenmeyer acid-washed flask that could contain 200 ml of distilled water, with an NH + 4 -N concentration prepared from an NH 4 Cl and PO 3− 4 -P concentration using KH 2 PO 4 stock solutions to reach the desired concentrations, and with the corresponding zeolite weight (20g or 50g). All the samples were tested in duplicates and shaken in the Orbital Laboratory Stirrer at 20 ± 0.1 • C and 100 rpm. Independent trials for NH + 4 and PO 3− 4 were conducted to evaluate adsorption behavior separately. The trials were developed separately since, according to Karapinar [41] synergic effects (positive or negative) do not occur with the simultaneous removal of ammonium and phosphorus by natural zeolites. Two controls were used. At the beginning, microbiological activity was checked with a control using a sterilized zeolite and concentrations of 85 mg/L and 15 mg/L for NH + 4 -N and PO 3− 4 -P, respectively. This check was developed to control the microbial activity in batch assays because only the physic-chemical effect was desired for the study. The zeolite was sterilized in an autoclave (20 min at 105 • C) [42][43][44]. Afterwards, the desorption effect was analyzed with a blank, using the zeolite, a 0 mg/L concentration of NH + 4 -N, and a 0 mg/L concentration of PO 3− 4 -P. The NH + 4 -N concentration in the liquid was monitored every 30 min until saturation or for a maximum 8-h period. The PO 3− 4 -P concentration in the liquid was monitored at different times during a maximum 216-h period (0, 4,8,24,28,32,48,52,56,72,76,80,96,100,104,120,168, and 216 h).

Adsorption Modelling
The removal of PO 3− 4 -P and NH + 4 -N by zeolite adsorption can be considered to be a saturation-growth-rate equation as a function of adsorption time, like the one proposed by the Langmuir adsorption kinetic model [10,48]. The saturation-growth-rate equation is a nonlinear model that was fitted with the experimental data in order to determine the adsorption rate for only one batch assay. This equation is as follows: where t is the time (h), m is the amount adsorbed on a defined amount of adsorbent (mg/g), m max is the maximum amount adsorbed on a defined amount of adsorbent (mg/g), and t 1/2 is the time for m to be half m max . Furthermore, the Langmuir adsorption kinetic model was also used to fit the experimental data from the batch test because, unlike the saturation-growth-rate model, the Langmuir model can cluster different batch assays. The Langmuir model has previously been used to evaluate the adsorption process onto different materials [49][50][51]: where q is the mass of adsorbate per mass unit of adsorbent at equilibrium (mg/g), C e is the equilibrium adsorbate concentration (mg/L), Q is the maximum mass adsorbed at saturation conditions per mass unit of adsorbent (mg/g) (maximum adsorption capacity), and b is the empirical constant with units of inverse concentration (L/mg) [52]. Furthermore, b can be considered to be an energy adsorption indicator onto the adsorbent. In the Langmuir model, Q and b can be estimated by regression.

Statistical Analysis
Statistical analyses were used to evaluate measured and calculated parameters for natural zeolite as the adsorbent material. First, the mean and standard deviation of triplicate results were determined. Second, the degree of intensity or effectiveness that independent variables have in explaining dependent variables in the multiple applied regressions, i.e., the saturation-growth-rate equation and Langmuir model, was determined by R 2 and adjusted R 2 (R 2 adj ): where n is the experiment number, k is the independent variable number and R 2 is the coefficient of determination. Third, the data were subjected to a normality test (i.e., a Shapiro-Wilk Test) to determine the statistical tests for comparison. Fourth, to compare (1) the zeolite's adsorption capacity and the initial concentration, (2) the zeolite's adsorption capacity and the zeolite sizes, and (3) the zeolite's adsorption capacity and the zeolite weight, the following tests were performed, (a) data with normal distribution, one-way ANOVA test and (b) data without normal distribution, the Kruskall-Wallis test. All statistical tests were performed using InfoStat © with a significance level α = 0.05. Figure 1 shows the results of NH + 4 -N adsorption onto Chilean natural zeolite for the batch experiments. The saturation adsorption of NH + 4 -N to the zeolite varied between 0.05 and 0.7 mg/g. In addition, the saturation adsorption of NH + 4 -N increased at the same time that the initial concentration in the solution increased for the same zeolite weight but the saturation values were similar among the different zeolite sizes ( Figure 1). This result shows the positive relationship between two variables (p < 0.05)-the zeolite's adsorption capacity and the initial concentration-but it does not show a relationship between the tested zeolite sizes (p > 0.05). This characteristic is important for the use of Chilean natural zeolite as a support medium in treatment wetlands and the possibility of using its various sizes was tested in this work (0.2-1.0 mm, 1.5-3.0 mm, and 5.0-8.0 mm). In addition, when the zeolite quantity increased from 20 g to 50 g (2.5 times), the adsorbed NH + 4 -N onto zeolite increased between 1.5 and 3.0 times at the same initial concentration and zeolite size.  Figure 2 shows the results of PO -P adsorption onto the tested zeolite. The saturation adsorption of PO -P varied from 0.001 to 0.04 mg/g. In comparison to the adsorption of NH -N, the PO -P did not show a direct relationship with the initial concentration (p > 0.05). In the case of zeolite size, values above 0.04 mg/g were only achieved for the PO -P initial concentrations above  Figure 2 shows the results of PO 3− 4 -P adsorption onto the tested zeolite. The saturation adsorption of PO 3− 4 -P varied from 0.001 to 0.04 mg/g. In comparison to the adsorption of NH + 4 -N, the PO 3− 4 -P did not show a direct relationship with the initial concentration (p > 0.05). In the case of zeolite size, values above 0.04 mg/g were only achieved for the PO 3− 4 -P initial concentrations above 10 mg/L, when the medium zeolite size (1.5-3.0 mm) and 20 g of adsorbent (zeolite) were employed. The results suggest that there is no relationship between the zeolite size and the saturation adsorption of PO 3− 4 -P (p > 0.05). Therefore, any of the zeolite sizes employed in this study could be used as a support medium in treatment wetlands. Furthermore, when the zeolite amount was increased from 20 g to 50 g (2.5 times), there was no increased effect of the saturation adsorption of PO 3− 4 -P. In comparison to results for the NH + 4 -N adsorption, this is another difference with the PO 3− 4 -P adsorption onto natural zeolite. Additionally, for adsorption capacity, Table 4 shows the NH -N and PO -P removal percentage measured in the batch assays. Table 4 shows the NH -N removal efficiency above 60%, indifferent to the initial concentration, zeolite size, and zeolite quantity. However, when zeolite sizes are between 1.5 and 3.0 mm, better removal is seen. In the case of PO -P, Table 4 shows that the removal efficiency was above 35%, regardless of the initial concentration, zeolite size, and zeolite quantity. NH -N and PO -P removal was always above 70% for zeolite size between 1.5 and 3.0 mm. Additionally, for adsorption capacity, Table 4 shows the NH + 4 -N and PO 3− 4 -P removal percentage measured in the batch assays. Table 4 shows the NH + 4 -N removal efficiency above 60%, indifferent to the initial concentration, zeolite size, and zeolite quantity. However, when zeolite sizes are between 1.5 and 3.0 mm, better removal is seen. In the case of PO 3− 4 -P, Table 4 shows that the removal efficiency was above 35%, regardless of the initial concentration, zeolite size, and zeolite quantity. NH + 4 -N and PO 3− 4 -P removal was always above 70% for zeolite size between 1.5 and 3.0 mm.

Adsorption Modelling
From the results presented in Figures 1 and 2, Table 5 shows the results for the adjustment to the saturation-growth-rate model proposed by Equation (1). R 2 and R 2 adj are similar as the data number considered in the curve fit increases. In Table 5, the time for t max/2 in the NH + 4 -N adsorption process had variability, with values from 0.2 h to 7.7 h. Furthermore, the t max/2 was reduced when zeolite quantity increased from 20 g to 50 g for all zeolite sizes evaluated in this work (Table 5). In the case of PO 3− 4 -P, the time for t max/2 in the PO 3− 4 -P adsorption process had huge variability, similar to the NH + 4 -N adsorption process, with values from 5.1 h to 78.2 h. Similar results were achieved for different zeolite sizes evaluated in this work, when zeolite quantity increased from 20 g to 50 g. The time needed for maximum adsorption and removal of compounds of interest from aqueous solutions is an important parameter, because this information would be relevant for determining the recommendations for hydraulic retention time when this material is used as a support medium in treatment wetlands. The m max values were similar for NH + 4 -N adsorption between different zeolite sizes and quantities but better results for PO 3− 4 -P adsorption were achieved for zeolite sizes of 1.5-3.0 mm, with values up to 0.14 mg/g when zeolite quantity was 20 g and up to 0.06 mg/g when zeolite quantity was 50 g. These maximum values for the medium size, 1.5-3.0 mm, represent up to three times the m max values of other zeolite sizes. Thus, this result explains that the best removal efficiencies are achieved by the medium zeolite size of 1.5-3.0 mm, as shown in Table 4.  Figure 3 shows the adsorption isotherm for particle size and zeolite amount from the batch assays of the two analyzed compounds: NH 4 + -N and PO 4 −3 -P. In the case of NH 4 + -N, Figure 3 shows a typical evolution of the sorption isotherm when 50 g of zeolite was used. For PO 4 −3 -P, the sorption isotherm was in the linear phase and the beginning of the saturation phase for the two quantities of zeolite and the different sizes. The result for PO 4 −3 -P shows the potential of the material to be used for treatment of wastewater with concentrations above 15 mg/L and, considering the typical concentrations found in domestic wastewaters, it suggests that it can be used to treat this kind of wastewater. Table 6 shows the Langmuir adsorption kinetic parameters, calculated by regression using Equation (2) and the coefficient of determination (R 2 and R 2 adj ), which are based on data from Figure 3. According to results presented in Table 6, all analytical isotherms showed R 2 above 0.9. Despite that, when the zeolite amount of 50 g and particle size of 5.0-8.0 mm were adjusted to the Langmuir model, the maximum adsorption capacity, Q, had a negative value. This result can be explained because the NH 4 + -N concentrations evaluated in the batch assays (up to 85 mg/L), for these conditions of quantity and size, were apparently at the start of the linear phase (b×Ce <<< 1) and not in the saturation phase; thus, the mass of the adsorbate per mass unit of adsorbent at the equilibrium is described by q = Q*b*Ce. This equation has infinite solutions, one of these being a negative slope, as shown in Table 6.
concentrations found in domestic wastewaters, it suggests that it can be used to treat this kind of wastewater.  Table 6 shows the Langmuir adsorption kinetic parameters, calculated by regression using Equation 2 and the coefficient of determination (R 2 and R 2 adj), which are based on data from Figure 3. According to results presented in Table 6, all analytical isotherms showed R 2 above 0.9. Despite that, when the zeolite amount of 50 g and particle size of 5.0-8.0 mm were adjusted to the Langmuir model, the maximum adsorption capacity, Q, had a negative value. This result can be explained because the NH4 + -N concentrations evaluated in the batch assays (up to 85 mg/L), for these conditions of quantity and size, were apparently at the start of the linear phase (b×Ce <<< 1) and not in the saturation phase; thus, the mass of the adsorbate per mass unit of adsorbent at the equilibrium is described by q = Q*b*Ce. This equation has infinite solutions, one of these being a negative slope, as shown in Table  6.     Table 6 shows that the medium zeolite size (1.5-3.0 mm) had higher values for the maximum adsorption capacity of ammonium for the two quantities evaluated (20 g and 50 g), above 0.95 mg/g and up to 1.58 mg/g. Similarly, for phosphate, the medium size showed higher values when 50 g were used, above 0.04 mg/g; however, it showed the second highest removal when 20 g were used, with removal above 0.08 mg/g. These results explain why the best removal results were achieved for this size of zeolite, as shown in Table 4.

Discussion
The behavior of the adsorption process of NH 4 + -N and PO 4 −3 -P could be described for Chilean natural zeolites through two adsorption models: Saturation-growth-rate and Langmuir (Table 5, Figure 3, Table 6). The results are similar to the results reported for other materials studied to be used as support mediums in treatment wetlands [9,32,53]. In addition, the results of this study confirm that Chilean natural zeolites can be used for both N and P removal from wastewater (including those produced in rural areas) and that their potential application as a support medium for treatment wetlands is similar to that of other natural zeolites previously used for this purpose [30,40,54].
Of the different zeolite sizes evaluated, the medium size of 1.5-3.0 mm was the size that performed best, achieving an NH 4 + -N removal above 85% and a PO 4 −3 -P removal above 70% ( Table 4). The removal efficiencies achieved with this size show that adsorption can be improved for both NH 4 + -N and PO 4 −3 -P removal by at least 40% and 30%, respectively, in comparison to traditional support mediums, such as sand and gravel, which have a more limited capacity [5,21]. In addition, the highest maximum adsorption capacity of NH 4 + -N was achieved for this zeolite size (1.5-3.0 mm), with values up to 1.6 mg/g. However, the regular values reported for the NH 4 + -N maximum adsorption capacity varied between 2.7 and 31 mg/g [22]. According to this, the NH 4 + -N maximum adsorption capacity of this study is around 60% of the minimum value reported in the literature. However, the result achieved in this work (1.58 mg/g) shows that Chilean natural zeolite has the potential to remove ammonium from wastewater (Table 4). For PO 4 −3 -P, the medium zeolite size (1.5-3.0 mm) showed the highest maximum adsorption capacity, with values up to 0.08 mg/g. The potential capacity is three times better than the results reported by Andrés et al. [10], with 0.026 mg/g for another Chilean natural zeolite and 1.0-2.0 mm size. The difference could be justified by the Ca content and zeolite size and, therefore, the surface area. The Ca content was described by Del Bubba et al. [55] as an important factor that influences the P adsorption capacity in natural materials employed as adsorbents. In this work, the reported Ca content was 3.42%, while in Andrés et al. [10] it was 2.5%. With respect to size, the P adsorbed is expected to increase as the adsorbent surface increases [56,57], as is the case when the zeolite size decreases because there is a larger contact surface available. However, Moharami and Jalali [32] showed that, when the adsorbent quantity and surface are greater, there is a very fast adsorption onto the adsorbent surface and the consequence is a lower adsorbate concentration in the solution. The result is that some adsorption sites on the adsorbent surface remain unsaturated. Therefore, the adsorbate concentration in the solution drops to a lower value, which makes it possible that the amount of adsorbate per unit weight of adsorbent can be reduced when the adsorbent quantity and surface are increased.
According to Table 3, the t max/2 for the 1.5-3.0 mm zeolite size varies from 0.1 to 2 h for NH 4 + -N removal and from 6.0 h up to 80 h (approximately 3 d) for PO 4 −3 -P removal. This result can be seen in the need for longer hydraulic retention time (HRT) in treatment wetlands to reach 50% of the maximum adsorption capacity of the tested zeolite. In the case of treatment wetlands with horizontal subsurface-flow (HSSF-TWs), the HRT varies from 1.5 to 7 d [58]; thus, residence time considered in the present study is included. In the case of treatment wetlands with vertical subsurface-flow (VSSF-TWs), the HRT is short and under one day [59]; thus, NH 4 + -N removal can be achieved but it might not be enough for PO 4 −3 -P removal. However, the HRT in VSSF-TWs can be increased by saturating the bottom of the bed [60]. The lifespan of the zeolite (up to saturation) as a support medium in a SSF-TW for wastewater treatment is an important aspect that can be assessed by determining the adsorption maximum capacity achieved by adsorption experiments. In wastewater, one person produces around 2 g P/PE-d and 12 gN/PE-d [61]. Therefore, according to the study, 12.5 g of the Chilean natural zeolite would be necessary for P and N removal. Considering that the design recommendation for HSSF-TWs include 5 m 2 /PE and depth of 0.5 m [62], then 2.5 m 3 of volume or 5500 kg of zeolite (density of zeolite, 2200 kg/m 3 , Engler and Rubio [29]) would be necessary for the support medium. This quantity of HSSF-TWs would be useful for 611 d (almost 2 years). This can be considered a low lifespan. However, Andrés et al. [10] showed that a material such as the natural zeolite can extend the lifespan of the SSF-TW up to 2 times through the regeneration of adsorption sites as a result of the plants' uptake of nutrients and aeration strategies. Thus, the lifespan of the Chilean zeolites studied indicates that they have the potential to be used for longer periods of time than those calculated only by the adsorption process. However, further studies will help determine the duration of this lifespan extension. In a complementary way, Andrés et al. [10] showed that maximum adsorption capacity for P removal was increased by ten times when tested by batch loading the assays, when zeolite was the unique support medium in SSF-TWs for wastewater treatment. This increase in maximum adsorption capacity, under real operation, can be explained by the more adsorbent surface available for adsorption as a consequence of more zeolite employed in SSF-TWs. In addition, Drizzo et al. [63] attributed the increase to the age of the system, hydraulic distribution, pretreatment unit, solids, and plant condition. Milliot et al. [38] employed natural zeolites as a layer of support media in VSSF-TWs for treating wastewater, and discussed reductions in the capacity of ammonium removal. The reductions were explained by the effect of several factors such as preferential flows, short hydraulic residence times, and low concentration of influent ammonium (below 10 mg/L). To solve it: two strategies were proposed: (1) increasing zeolite's quantity, and (2) employing a similar layer, reducing zeolite granulometry to increase the adsorbent surface. Regarding the second solution, this work showed that reducing the granulometry did not necessary redound to the increase of the adsorption capacity (Table 5); the kinetics reaction is more important than mass transfer. Therefore, increasing zeolite quantity in the TW seems to be a better solution because more adsorbent surface is available and therefore the adsorption capacity would increase. Another strategy could be the use of an external filter as post treatment, which would imply the use of an extra structure but might result in long-term operation and maintenance savings. This research can be helpful considering the serious operational problem of SSF-TWS: clogging [64]. Clogging is the accumulation of solids (wastewater suspended solids) or the buildup of biofilm (chemical precipitates, plant detritus accumulation and biofilm growth) that results in porosity and hydraulic conductivity and water transport decrease over time [65]. This problem could affect the contact time between zeolite and wastewater; thus, reducing the removal capacity of ammonium and phosphate by the SSF-TWs during the treatment. However, different management strategies to reduce clogging in SSF-TWs are discussed in Nivala et al. [66], where the authors suggest a limit of 250 g/m 2 for cross-sectional Biological Oxygen Demand (BOD) loading. In recognition of the physical reality that most SSF treatment wetland processes are a function of biofilm surface area, Austin et al. [67] propose a method to systematically analyze this type of clogging as a design tool: A Damköhler number (Da) definition based on aggregate specific surface area is used to investigate a method of predicting clogging induced by heterotrophic biofilms growing on treatment media.
Finally, taking into account the previous discussion, the Chilean natural zeolite with a commercial size of 1.5-3.0 mm is recommended to be used as the support media in treatment wetlands to remove N and P from domestic wastewaters, including those produced in rural areas. In addition, after saturation in treatment wetlands, the Chilean natural zeolite has the possibility to be used as a source of nutrients, namely N and P, as has already been proposed by several authors [27,[68][69][70].

Conclusions
A new material, Chilean natural zeolite, for the removal of ammonium and phosphate from wastewaters, including those produced in rural areas, was described in this work. The saturation-growth-rate equation was fit successfully to the experimental data to determine the adsorption rate for each batch assay. On the other hand, the adsorptions of NH 4 + -N and PO 4 −3 -P were described for the Langmuir isotherm and the results showed that the commercial zeolite size of 1.5-3.0 mm was the size that had better adsorption behavior for NH 4 + -N and PO 4 −3 -P, with maximum adsorption capacities up to 1.58 mg/g and 0.08 mg/g, respectively, as well as the greatest potential for the removal of these elements contained in wastewater. These results show that these two compounds can be removed from aqueous solutions using Chilean natural zeolite and the material has the potential to be used for the removal of NH 4 + -N and PO 4 −3 -P from rural wastewater. Therefore, Chilean natural zeolite has the potential to be used as a support medium in treatment wetlands, taking advantage of its adsorbent properties. Finally, 12.5 g of the Chilean natural zeolite would be necessary for the removal of P and N found in wastewater in the amount produced by one person every day. Finally, considering the design recommendation for HSSF-TWs for treating wastewater coming from one inhabitant, the zeolite used in this kind of treatment wetland as a support medium would have a lifespan of around two years considering only the adsorption effect.