Synthesis and characterization of zeolites

Zeolite was synthesized using the method described by ^{Dussan Dussan, Otálora, Chica, Bonilla and Otálora (2012)}. The characterization of the synthesized zeolite was carried out using thermogravimetric analysis (TGA), infrared spectroscopy (IR), x-ray diffraction (XRD) and scanning electron microscopy (SEM). The thermogravimetric analysis was carried out in an air atmosphere at a ramping temperature rate of 10°C.min^-1, from room temperature up to 800 °C, followed by a 10-minute isotherm. The analysis was carried out using Q500 thermogravimetric analyzer (TA Instruments). Fourier transform infrared spectroscopy (FTIR) analysis was conducted using the KBr pellet method in a 4 000 cm^-1 to 400 cm^-1 wavelength range with a Nicolet 6700 spectrometer. XRD analysis of the synthesized zeolite was conducted using a Panalytical X'Pert Pro Multipurpose Diffractometer (MPD), with an accelerating voltage of 45 kV and a 40 mA current intensity, in a 2θ range of 10 to 80°, a step of 0.020º, and a time per step of 8s. The SEM analysis was conducted using a JEOL JSM-6490LV scanning electron microscope. The sample was previously coated with gold using the Denton Vacuum Desk IV coating system. Additionally, a surface chemical analysis was carried out by energy-dispersive x-ray spectroscopy (EDX).

Sampling and characterization of wastewater

A characteristic municipal wastewater discharge of Florencia, Colombia, was used by the present study. The samples were collected according to standard sampling and preservation techniques by the Colombian Institute of Hydrology, Meteorology and Environmental Studies - IDEAM (^{Sánchez, 2005}). The zeolite adsorption response was evaluated with three concentrations of organic matter, measured as chemical oxygen demand (COD), which were characteristic of MWW at different times of the year. The MWW with higher COD values (400 - 500 mg.L^-1) were classified as MWW with high organic matter content, while those with medium and low concentrations were obtained using dilutions to reach concentrations ranging from 200 to 300 mg.L^-1 and 80 to 120 mg.L^-1, respectively. The characteristics of the MWWs were determined using the closed reflux method for COD. The level of ammoniacal nitrogen was determined by the Nessler method, and the total water hardness was determined by titration (calcium and magnesium). Heavy metal content (Fe³⁺, Cu²⁺, Zn²⁺, and Pb²⁺) was evaluated by atomic absorption spectrophotometry (AAS). All tests were performed according to the Standard Methods for the Examination of Water and Wastewater (APHA-AWWA-WEF) (^{APHA-AWWA-WEF, 2012}). These methods were used to analyze the response variables in synthetic water adsorption tests and determine the adsorption capacity of zeolites.

Preliminary adsorption tests - Synthetic water

All adsorption tests were conducted using lab-made solutions, e.g., synthetic water (SW). Each of the ions to be evaluated (Fe³⁺, Cu²⁺, Pb²⁺, Zn²⁺, Ca²⁺, Mg²⁺, and NH₄ ⁺) was dissolved in deionized water. The optimal dosage at which synthetic faujasite-type zeolite reached the highest ion adsorption rate was determined using zeolite doses ranging from 0.01 g.L^-1 to 4 g.L^-1, with increments of 0.25 g.L^-1. For all cases, the response variable was represented by the removal capacity of the ion.

Adsorption tests in synthetic water and municipal wastewater

The ion removal capacity for both SW and MWW was evaluated using assays in reaction vessels, stirring for 1 hour, with 50 mL each of the ammoniacal nitrogen, hardness (Ca²⁺, and Mg²⁺), and heavy metal (Fe³⁺, Cu²⁺, Zn²⁺, and Pb²⁺) solutions used at different concentrations (maximum and minimum for MWW). Zeolite doses were used according to the results obtained in preliminary tests, and a 1-hour adsorption test was conducted at room temperature with a constant stirring speed. Zeolite doses were used according to the results obtained in preliminary tests. The total concentration for each of the ions was calculated using the aforementioned methods. MMW treatability using faujasite-type zeolites was evaluated according to the pollutant removal percentage found for the samples (metals, ammonia nitrogen, hardness, and organic matter as COD).

Variance was analyzed (ANOVA) based on the results of the experimental design. This analysis made it possible to draw conclusions about the influence of each factor and their correlation. The factors considered in the statistical analysis included the maximum adsorption zeolite dose and higher adsorption concentrations. This analysis involved the use of SPSS statistical software.

The results obtained permitted Freundlich and Langmuir isotherms to be determined in order to find the theoretical model that best fit the experimental conditions. A goodness of fit test (χ²) was performed, then the specific surface area was determined. The adsorption process for each of the isotherms was adjusted based on the results obtained from tests conducted on the SW and MWW samples.

Zeolite characterization

The infrared spectrum obtained with the synthesized zeolite (Figure 1A) had three signals in the OH region, as follows: 3700 cm^-1 corresponding to terminal silanols, 3620 cm^-1 for Al-OH-Si groups, and a weak band at 3650 cm^-1 corresponding to OH extra framework (^{Bevilacqua, Montanari, Finocchio, & Busca, 2006}). The signal shown at 3440 cm^-1 corresponds to stretching vibrations of silanol (Si-OH) and O-H (^{Baur, Héroguel, Spring, Luterbacher, & Kiwi-Minsker, 2016}). This is superimposed by bands in the region between 3200 cm^-1 and 3400 cm^-1 and correspond to O-H vibrations of water. The presence of water in the zeolite is determined by a weak band at wavenumber 1650 cm^-1, due to the deformation of the O-H bond found in water molecules (^{Moulin et al., 2008}). The bands in the range 980 cm^-1 1200 cm^-1 are associated with asymmetric T-O-T stretching (T = Si or Al), and the bands found in 800 cm^-1 correspond to symmetrical stretching of the T-O-T bond (^{Fricke, Kosslick, Lischke, & Richter, 2000}).

The diffractogram obtained for the zeolite synthesized in this work (Figure 1B) exhibited diffraction angles 2θ, which was comparable to those previously reported (^{Treacy & Higgins, 2001}) and suggests that the zeolite used was a Na-X faujasite-type zeolite. Additionally, the sample synthesized by the aforementioned technique exhibited high crystallinity.

Figure 1 Infrared spectrum (A) and x-ray Powder diffraction (B) for the synthesized zeolite. 

The thermogram obtained for the zeolite prior to the metal adsorption process (Figure 2A) exhibited mass loss on two occasions. The first mass loss (1.43%) occurred at 110 ° C, which suggests a loss of surface water in the zeolitic material (^{Ríos, Williams, & Castellanos, 2006}). The latter (3.73%) occurred at 200 °C, which suggests the presence of water desorption in the internal zeolite structure (^{Khaleghian-Moghadam & Seyedeyn-Azad, 2009}). The thermogravimetric analysis of the zeolite after the removal process (Figure 2B) showed three mass loss events. An additional event occurring at 90 °C may be related to water molecules adsorbed during the adsorption process. The subsequent losses were the same as those observed in the zeolite before the adsorption process.

Figure 2 Thermogram for synthetic faujasite zeolite before (A) and after (B) the adsorption process in MWWs. Thermogravimetric analysis (solid line) and the derivative weight loss curve (dash line corresponds). 

SEM micrographs prior to adsorption (Figure 3A and 3B) indicated the presence of uniformly-distributed granules that correspond to the zeolite. There were also non-uniform nanocrystal agglomerates on nuclei of a higher constitution, which may be related to the growth of zeolites during synthesis (^{Dussan, Otálora, Chica, Bonilla & Otálora, 2012}). According to the results obtained from x-ray diffraction, these agglomerates were crystalline aggregates. The SEM results showed a sphere-shaped morphology that was characteristic of the zeolites, with sizes ranging from 500 nm to 2 600 nm (^{Gómez, 2001}). Previous studies with commercial zeolites have reported particle sizes between 590 nm to 1 020 nm (^{Zhang, Yang, Tang, & Yang, 2015}) and 2 300 nm crystals (^{Ansari, Aroujalian, Raisi, Dabir, & Fathizadeh, 2014}). Therefore, the morphological characteristics of the synthesized zeolite agreed with those reported by other authors. The ion removal process resulted in changes in the shape, size, and uniformity of the aggregates. In this case, the particles did not have a particular shape and were more agglomerated with each other. In addition, there were particles of much larger sizes, ranging from 700 nm to 6 000 nm, approximately. Usually, the morphology and particle size are not expected to change after ion exchange (^{Zhang et al., 2015}). However, other authors have reported a loss of crystallinity and morphology in crystals, which appear agglomerated without a particular shape after impregnation with metals (^{Gómez, 2001}).

Figure 3 SEM micrographs of synthetized faujasite zeolite before (A, B, C) and after (D, E, F) the adsorption process in MWWs. 

The analysis of EDX indicated the presence of oxygen (53.5%), silicon (19.2%), aluminum (15.3%) and sodium (12.0%) on the surface prior to the adsorption process. The chemical composition coincided with that previously reported, where the average formula is defined as Na₈₅Al₈₅Si₁₀₇O₃₈₄ (^{Moulin et al., 2008}). After the ion removal, there were traces of magnesium, potassium, and calcium, as well as an increase in oxygen levels. This may be due to the fact that the open structure of the zeolite provides a great ability to incorporate cations such as Mg²⁺, K^+, and Ca²⁺. The increase in the oxygen mass level suggests that the zeolite underwent hydration (^{Castaldi, Santona, Enzo, & Melis, 2008}).

Removal of ions in synthetic water (SW)

The results obtained from preliminary adsorption tests in SW showed that the optimum zeolite dose for the removal of ions ranged between 0.01 g.L^-1 and 1.0 g.L^-1. Previous experiments with SW have shown that the ability to remove ammoniacal nitrogen can be improved by increasing the dose of zeolite. The use of a 0.6 g.L^-1 dose in the concentrations evaluated (12 mg.L^-1 and 25 mg.L^-1) can remove ions up to 98%. The initial concentration of ammoniacal nitrogen did not significantly affect the level of adsorption obtained in each experiment. As for water hardness, the removal percentage remained constant (100%) in experiments with higher doses of zeolite, which suggests that the dosage of zeolite and the initial concentration of calcium and magnesium ions do not affect the removal process. This behavior partially coincides with that reported by ^{Manrique, Bonilla, Chica, Otálora and Salamanca (2015}), where the dosage of zeolite was shown to affect hardness removal. Also, the use of doses greater than 0.3 g.L^-1 resulted in levels of removal greater than 90%.

Figure 4 shows Na-X zeolite's ability to remove heavy metals from synthetic water. For low concentrations of Cu²⁺ (8 mg.L^-1), Fe ³⁺ (15 mg.L^-1) and Zn²⁺ (10 mg.L^-1), the zeolite dose was associated with ion adsorption rates ranging from 90% to 100%. As for high ion concentrations, the use of doses greater than 0.02 mg.L^-1 resulted in removal levels above 80%, since higher doses of zeolite leads to an increase in both the surface area and adsorption points, and therefore it promotes ion exchange (^{Burakov et al., 2018}). The Pb²⁺ ions behaved differently since a higher ion concentration (10 mg.L^-1) leads to an increase in removal levels, regardless of the dose of zeolite. For low Pb²⁺ ion concentrations (3 mg.L^-1) the level of removal was drastically reduced by using low doses of zeolite (less than 0.04 mg.L^-1), while the use of higher doses of zeolite resulted in removal levels ranging from 50% to 80%. This behavior may be associated with a decrease in the hydrated radius of the ion (0.261 nm) and the ability to diffuse in the zeolite structure. However, a weak interaction between the Pb²⁺ ion and the zeolite was prevalent, and this was favored by the low ion concentration present in the solution. Faujasite-type zeolite was shown to have greater selectivity for Zn²⁺ in the absence of other ions. This may be due to the fact that unlike Pb²⁺ion, Zn²⁺ has a greater ability to interact with the active zeolite sites (^{Mekatel et al., 2012}). Similar results were obtained by other studies where zeolite exhibited greater selectivity for the Zn²⁺ ion than for Pb²⁺ in monocationic solutions Zn²⁺ > Pb²⁺> Cd²⁺ (^{Castaldi et al., 2008}).

Figure 4 Removal percentage for Pb²⁺ (A), Zn²⁺ (B), Cu²⁺ (C), and Fe³⁺ (D) in synthetic water at different concentrations, according to the zeolite dose. 

Removal levels of around 90% can be obtained for initial concentrations of Cu²⁺ (8 g.L^-1) by using low doses of zeolite (<0.0625 g.L^-1). Removal levels of Cu²⁺ up to 55% by using zeolite doses of 4 g.L^-1 and Cu concentrations up to 250 mg.L^-1 have been previously reported (^{Visa, 2016}). This shows that the zeolite evaluated in this study was effective for the removal of this type of ion when used in low doses.

Removal of ions in municipal wastewater (MWW)

Figure 5 and Figure 6 show the removal percentages obtained by using different doses of zeolite for each of the ions and the organic matter (quantified as COD) present in MWW, with different initial concentrations of organic matter (high, medium, and low). The competitive effect in such a complex water matrix resulted in an overall decrease in the effectiveness of the adsorption process. Factors such as the concentration, the size and charge of the hydrated cation, the anionic species associated with the cation, and the characteristics of the zeolite, play an important role in the ion exchange results (^{Ríos, Williams, & Roberts, 2010}).

Figure 5 Removal percentage of each of the metal ions in MWWs with high (A), medium (B), and low (C) initial concentrations of organic matter. Pb²⁺ (--), Zn²⁺(--), Cu²⁺(--), Fe³⁺ (--). 

Figure 6 Percentage of adsorption of organic matter as COD (A) and ammoniacal nitrogen (B) in MWWs with high (--), medium (-0-), and low (--) initial concentration of organic matter. 

According to the statistical analysis, the zeolite dose significantly impacts (p < 0.05) ion removal (Pb²⁺, Fe³⁺, Cu²⁺, Zn²⁺, Mg²⁺, Ca²⁺, and ammoniacal nitrogen) as well as organic matter content (COD) in MWWs (Figure 5 and Figure 6). The results show that the higher the zeolite dose, the greater the removal of all the components found in MWWs. COD removal levels between 75% and 95% can be obtained by using zeolite doses greater than 0.4 g.L^-1 for the three types of MWWs (high, medium, and low organic matter concentration). This may occur since the higher the initial concentration of the adsorbate, the lower the adsorption equilibrium capacity, and the greater availability of the substrate to adhere to the adsorbent (^{Koshy & Singh, 2016}). Regarding interaction between factors (zeolite dose - initial organic matter concentration), a statistically significant effect (p < 0.05) was found with respect to the removal of ammoniacal nitrogen (positive effect) and Zn²⁺ (negative effect), which impacts the removal process. An additional contribution to the removal process are second-order interactions between two factors (zeolite dose² and initial organic matter concentration²) that individually influence the response. This shows that the zeolite dose² has a negative second-order effect on the removal of organic matter, Fe³⁺, Zn²⁺, and water hardness, while the concentration of MWWs² has a positive second-order effect on Cu²⁺ and Zn²⁺ removal.

Some authors have reported similar results in studies where Pb²⁺ removal was greater than the removal of Cu²⁺ in a multicomponent water matrix. However, when higher doses of zeolite are used the effectiveness of metal ion removal is close to 100% (^{Zheng & Zaoui, 2011}; ^{Visa, 2016}).

There is evidence of a matrix effect on Pb²⁺ and hardness removal, which shows that an increase in the organic matter concentration in MWWs leads to a decrease in the zeolite's removal ability. The opposite effect occurs regarding removal of Fe³⁺ and Zn²⁺, since an increase in initial concentrations of organic matter in MWW_S results in an increase in ion removal. Metal cations (Zn²⁺, Fe³⁺, Pb²⁺, Cu²⁺) behave like Lewis acids, since their acid strength depends on their ionic radius, charge, and electronegativity. In general, these cations have such a high charge/radius ratio that they quickly interact with other ions, atoms and molecules to produce a more thermodynamically stable structure. In this case, stabilization was achieved by the interaction with oxygen and/or organic molecules found in the water, and the interaction with the negatively-charged zeolite, which acts as a Lewis base, providing the cation with additional stability.

Organic molecules found in MWWs can interact with transition-metal ions (Fe³⁺, Cu²⁺, and Zn²⁺) in order to form organometallic complexes. These compounds can affect metal adsorption since the metal complex exhibits different characteristics (solubility, charge, size) (^{Mekatel et al., 2012}). The formation of complexes was not assessed by this work. However, the results obtained for Zn²⁺ (Figure 5) suggest the formation of complexes with organic matter, which promotes adsorption in MWW containing high organic matter concentrations. Meanwhile, Pb²⁺ is not able to form metal complexes, and its adsorption is better in MWW systems with low concentrations of organic matter.

Among metal ions, Fe³⁺ has the lowest removal rate (between 50% to 65%) with a 0.5 g.L^-1 zeolite dose. This is explained by the precipitation of Fe³⁺, since it tends to form ferric hydroxide at pH values higher than 2.2. In addition to being a flocculent precipitate, this element has a strong basic character (^{Zhao et al., 2017}). A level of hardness removal of up to 100% was obtained in this study. Therefore, the lower the initial concentration of MWW the higher the removal percentage.

The removal of organic matter (Figure 6A) is improved by the low initial concentration of organic matter in MWWs, where removal levels of up to 94% have been achieved by using a 0.5 g.L^-1 zeolite dose. In MWWs with high initial organic matter concentrations, the percentage of removal does not exceed 85%. Competition for the zeolite exchange sites occurs between colloidal organic matter having a cationic surface charge and other cations present in MMWs. Therefore, an increase in the organic matter and ion concentrations leads to an increase in cations that have a greater affinity towards exchange and adsorption sites, which also causes a decrease in the levels of organic matter removal (^{Manrique, Bonilla, Chica, Otálora, & Salamanca, 2015}).

Figure 6B shows the percentages of ammoniacal nitrogen removal achieved by using different zeolite doses and initial concentrations. A level of removal of this ion of up to 94% was reached in MWW with low initial concentration. As for MWWs with medium initial concentrations, the level of removal was 91%, while in MWWs with high ammoniacal nitrogen concentration the removal percentage was 90% using a 0.5 g.L^-1 zeolite dose. The lower the initial concentration of the NH₄ ⁺ ion, the higher the percentage of removal, which can be attributed to the availability of active sites in the zeolite. Some authors have reported that cation exchange with NH₄ ⁺ Na⁺, K⁺, Ca²⁺, and Mg²⁺ ions accounts for much of NH₄ ⁺ removal (^{Rožić, Cerjan-Stefanović, Kurajica, Vančina, & Hodžić, 2000}). A previous study with naturally-occurring zeolite (clinoptilolite) found that greater ammoniacal nitrogen removal occurs when Na⁺ is present in the exchange sites. In addition, a competitive effect occurrs between the ammonium ions and other cations such as Ca²⁺, Mg²⁺, and K⁺. Nevertheless, selectivity remains predominant for the ammonium ion (^{Cooney, Booker, Shallcross, Stevens, & Geoffrey, 1999}). The zeolite used in this study exhibited greater Mg²⁺ and Ca²⁺removal levels.

Adsorption isotherms

Unlike the result obtained (R ² ) from the Freundlich adsorption model, the Langmuir model best fits the adsorption results, with coefficients of determination greater than 0.97. It is possible to assume that monolayers form where ion adsorption occurs in specific sites that are energetically equivalent and far enough apart, so that interaction between adsorbed ions in adjacent sites will not occur.

Table 1 shows the parameters corresponding to this type of isotherm in certain ions. The maximum adsorption rate (q _m ) is given in descending order as follows: Zn²⁺> Fe³⁺ - Cu²⁺> NH₄ ⁺> Pb²⁺. The zeolite's maximum Pb²⁺ adsorption capacity is induced by weak attractive forces caused by the size of the hydrated ion, as previously mentioned (^{Mekatel et al., 2012}). The q _m capacity of Pb²⁺ on the synthesized zeolite is lower when compared to other types of adsorbents, such as sawdust (^{Leyva, Berber, Mendoza, & Aragón, 2004}).

The maximum adsorption capacity of ammoniacal nitrogen (28.49 mg g^-1) is comparable with values obtained for two synthetic zeolites with an adsorption capacity ranging from 10.2 mg.g^-1 to 24.8 mg.g^-1. Other adsorbents, such as sugarcane bagasse ash, have a maximum adsorption capacity of 19.9 mg.g^-1 (^{Prieto-García, Quintana-Puchol, Rodríguez-Díaz, Arteaga-Pérez, & Mollineda-Trujillo, 2012}). These results show that synthesized faujasite has a higher adsorption capacity for ammonium ion compared to other materials.

A similar behavior occurs for Zn²⁺ and Cu²⁺, with q _m values ranging from 93.46 mg.g^-1 to 55.26 mg.g^-1, which are higher than those reported for clinoptilolite-type zeolites (^{Pavón & Briones, 2009}). The same happens with other adsorbents, such as tuff and bentonite (^{Rueda, Volzone, Lago, & Ortiga, 2007}).

The synthesized faujasite has a maximum adsorption capacity q _m of 68.55 mg.g^-1 for Fe³⁺, while the maximum adsorption capacity of iron ion reported for a natural zeolite is 24.29 mg g^-1, which shows that the zeolite synthesized in this study had better performance (^{Pavón & Briones, 2009}).

All in all, the maximum adsorption values for zeolite in solutions where ions are found independently are either comparable or exceed the results reported for other adsorbents with respect to each ion.

As for the Langmuir adsorption constant (K) for each ion, the results suggest that the values were higher than those reported by other authors, who reported constants of K = 0.251 Lg^-1 for Fe³⁺ adsorption , K = 0.150 Lg^-1 for Zn²⁺ , K = 3.4x10^-6 for Cu²⁺ and K = 2x10^-6 for Pb²⁺ ((^{Shavandi et al., 2012}; ^{Visa, 2016}).

Table 2 shows the fraction of sites occupied on the surface, also known as coverage fraction (θ = X/q _m ), where X accounts for the number of ions (mg) adsorbed per gram of zeolite, and q _m is the maximum adsorption capacity. In all cases, the formation of a monolayer on the surface of the zeolite (θ > 0.9) occurred for Cu²⁺ and Pb²⁺ ions, as ion concentration increased.

With respect to zeolite’s ion adsorption capacity, when ions are present in MWW (real matrix), a competition for adsorption sites occurs, as well as a competition for sites that are conducive to ion exchange. Ions with a better affinity towards zeolite will be able to occupy larger coverage fractions.

To determine the coverage fraction by using this matrix, it was necessary to independently take into account the fraction corresponding to each ion, and then group them in order to calculate ion behavior. These values were obtained by the Langmuir equation. Table 3 shows the aforementioned results. According to these values, the coverage fractions were higher for cations at low concentrations, since a lower number of ions are competing for the active sites on the zeolite surface. The cations with the highest coverage fraction were Cu²⁺ at 8 g^.L^-1, followed by Fe³⁺ and Pb²⁺, which had similar coverage fractions at both high and low concentrations.