Leaching behaviour of a sandy soil amended with natural and NH4+ and K+ saturated clinoptilolite and chabazite

ABSTRACT Using saturated or enriched zeolites as slow release fertilizers (SRFs) is considered as an environmental-friendly strategy to enhance use of macronutrients in sandy soils. In this paper, two natural zeolites, clinoptilolite (CLI) and chabazite (CHA) were used as mineral precursors to prepare NH4+/K+ saturated clinoptilolite (NH4+/K+-CLI) and chabazite (NH4+/K+-CHA) as zeolitic nutrient sources (ZNSs). Comparison between the nutrient retention capabilities of these ZNSs was one of the main objectives of this study. The NH4+/K+-CLI and NH4+/K+-CHA were prepared by soaking the zeolites in NH4Cl and KCl solutions, respectively. Leaching tests were performed on a sandy soil amended with chemical fertilizers (CFs), NH4+/K+-CLI and NH4+/K+-CHA to evaluate the leaching properties of them. The results indicated that approximately 84% and 88% of the NH4+ and K+ of soils fertilized with CFs were lost during the experiment, respectively. While, the NH4+ and K+ losses from soils amended with NH4+/K+-CLI and NH4+/K+-CHA were 29%, 23%, and 14%, 24%, respectively. Despite of drastic changes in leaching behavior of CFs, nutrient losses from ZNSs were more uniform during the experiments. No significant difference was found between the two studied zeolites on reduction of K+ loss. However, CHA was more effective in preventing NH4+ loss during leaching.

+ / K + saturated clinoptilolite (NH 4 + /K + -CLI) and chabazite (NH 4 + /K + -CHA) as zeolitic nutrient sources (ZNSs). Comparison between the nutrient retention capabilities of these ZNSs was one of the main objectives of this study. The NH 4 + /K + -CLI and NH 4 + /K + -CHA were prepared by soaking the zeolites in NH 4 Cl and KCl solutions, respectively. Leaching tests were performed on a sandy soil amended with chemical fertilizers (CFs), NH 4 + /K + -CLI and NH 4 + / K + -CHA to evaluate the leaching properties of them. The results indicated that approximately 84% and 88% of the NH 4 + and K + of soils fertilized with CFs were lost during the experiment, respectively. While, the NH 4 + and K + losses from soils amended with NH 4 + /K + -CLI and NH 4 + /K + -CHA were 29%, 23%, and 14%, 24%, respectively. Despite of drastic changes in leaching behavior of CFs, nutrient losses from ZNSs were more uniform during the experiments. No significant difference was found between the two studied zeolites on reduction of K + loss. However, CHA was more effective in preventing NH 4 + loss during leaching.

Introduction
Nitrogen (N) loss from agricultural soils is a worldwide issue causing environmental problems ranging from air and water pollution to climate change and stratospheric ozone depletion (Gholamhosseini et al. 2013;Kanter et al. 2015). Nutrients leaching can also cause environmental pollution and economic loss to farmers and also decrease the fertilizer use efficiency (FUE) (Xiang et al. 2008;Kanter et al. 2015). Among main pathways of N loss, leaching is one of the most important processes and is considered as a serious hazard (Widory et al. 2004;Ferretti et al. 2017). A great majority of the N spread in soils with chemical and organic fertilizers is lost by leaching and runoff (Dobermann & Cassman 2004;Malakouti et al. 2009;Duan et al. 2016;Vogeler et al. 2016). The same fate is reported for other nutrients such as K (Johnston and Goulding 1992;Johnston et al. 1993). For example, a precipitation of 20-45 mm/day within 10 days of fertilizer use can lead to loss of a large proportion (up to 50% or more) of the applied N and K through leaching (Ilias 2002). In particular, due to low cation exchange capacity (CEC), NH 4 + and K + following chemical fertilizers (CFs) application to sandy soils are more prone to leaching (Mackown and Tucker 1984;Kolahchi and Jalali 2007).
Enhancing nutrients efficiency by means of saturated/enriched or natural zeolites as nutrient and fertilizer carriers and sandy soil conditioners is one of the science-based solutions proposed by scientists (Mumpton 1984;Dwairi 1998;Coltorti et al. 2012;Li et al. 2013). Natural zeolites consist approximately 30-40% of channels of 0.1-0.4 nano meters diameter leading to unique pore selectivity for NH 4 + and K + (Manikandan and Subramanian 2014). High surface area and mesoporous structure of zeolites make them suitable as precursor for producing novel slow release fertilizers (SRFs) (Manikandan and Subramanian 2014).
Similar to other SRFs, the quality of saturated/enriched zeolites can be evaluated by means of leaching test, investigating species and quantity of the nutrient release. The way which nutrients are released from a soil amended with enriched zeolites largely depends on the species of the zeolite mineral present in the rock and thus their CEC (Mackown and Tucker 1984), the method of cations exchange and/or nutrients saturation (Hernandez et al. 1994;Notario Del Pino et al. 1995), and the geochemical behaviour of the ions (Wang and Peng 2010). Soil amended with NH 4 + and K + exchanged zeolites are subjected to minor leaching of these cations with respect to unamended soil independent from the type of soil and its texture (Park and Komarneni 1997;Perrin et al. 1998;Colombani et al. 2015;Campisi et al. 2016).
Identification and prevalence of zeolites use in fertilizer industry requires further and scientific practical researches especially on the line of addressing adsorption and release properties. Choosing a suitable zeolite as a precursor of SRFs is also a very important query to address because of its direct impacts on both environmental footprint and efficiency of the agricultural production processes. To the best of our knowledge, most of the previous researches were focused on evaluation of the quality and capability of an individual natural zeolite with focus on nutrient release properties from a special saturated/enriched zeolite. Very few studies have been reported on comparison of nutrient release pattern of different saturated/enriched zeolites. In this paper, the capability of two zeolites, i.e. clinoptilolite (CLI) and chabazite (CHA) as precursors of NH 4 + / K + fertilizers were evaluated using leaching experiments. The capability of reducing NH 4 + and K + loss from the soils amended with the two different employed zeolites was also investigated and compared with CFs.

Zeolites
The CLI and CHA zeolites are from a mine located in Semnan (Iran) and Grosseto (central Italy), respectively. Zeolite samples were crushed and sieved into particle sizes ranging from 0.5 to 1 mm. The separated fractions were washed with deionized water to remove residual small particles adhered to the mineral surface and then dried at room temperature for 24 hr. A detailed chemical and mineralogical characterization of sample CHA can be found in Malferrari et al. (2013).
An X'Pert PRO-PANalytical diffractometer was used to perform semi-quantitative mineralogical phase analyses on CLI zeolite. Results indicates that CLI consists mainly of clinoptilolite (about 44 weight %) with minor amounts of heulandite and tridymite. Total zeolitic content of chabazite zeolite was 70.9% which consists of 68.5% of chabazite, 1.8% of phillipsite, and 0.3% of analcime.
Chemical analysis of the samples CLI (Table 1) was carried out on pressed pellets of powdered rock via a wavelength dispersive Philips PW 1480 X-ray fluorescence (XRF) spectrometer. Loss on ignition (LOI) was determined by heating the sample under investigation in an oven at 1100°C.
Electrical Conductivity (EC) and pH of both CLI and CHA (Table 1) were measured in Milly-Q water (Millipore, USA) in a 1:2 (w/v) ratio, with a HANNA RS 180-7127 conductivity meter and an Orion 9102BNWP pH-meter connected to an Orion 4-star pH-ISE benchtop (Thermo Fisher), respectively. CEC (Table 1) was measured for sample CLI following the method adopted from Chapman (1965).

Soil
The sandy soil samples were taken from an agricultural field located in the Po river delta coastal zone Lido delle Nazioni, Ferrara, Italy (44º 45ˈ 15.44˝N, 12º14ˈ 22.84˝E). This agricultural area was derived from artificially levelled coastal dunes. The soil belongs to the cartographic unit 'CER1' according to the Emilia Romagna soil map 1:50.000 (http://geo.regione.emilia-romagna.it/cartpedo/, in Italian) and it is classified as Endogleyic Calcaric Arenosol (WRB 2007). The field is dedicated to cultivation of summer cereals (e.g. maize, wheat, barley, and sorghum) and occasionally to open field horticulture. To improve fertility of the soil, both chemical and especially organic fertilizers (Manure, slurry) application are common in this region. The soil was subjected to grain size analysis according to the method described by Di Giuseppe et al. (2015). Soil organic matter (OM) was measured by dry combustion (Tiessen and Moir 1993); water content was measured gravimetrically after heating for 24 h at 105°C, while soil pH and EC were measured alike for zeolite samples. As expected, the soil was characterized by very low OM, sub-alkaline pH and low salinity ( Table 2). The amount of soil to be loaded in the column was air dried at room temperature for 24hr and sieved to 2 mm.

NH 4 + and K + saturation
Zeolites were saturated by preparing batch suspensions. NH 4 + -saturated samples (N-CLI and N-CHA, respectively) were prepared using 5 g of each zeolite soaked individually in 25 mL of 1M NH 4 Cl solution in static condition at 20 ± 2°C. According to Hedström (2001) and Lin et al. (2013), the pH of the two solutions was adjusted to 6.5 to favour NH 4 + adsorption. All the batch experiments were carried out in three replicates analysed daily to check the amount of the adsorbed NH 4 + . The soaking solutions were refreshd every 5 days. At each sampling time, the supernatant was discarded and the solid was washed with deionized water until there was no sign of Cl − in the supernatant. Cl − was determined volumetrically using 0.01 N AgNO 3 as a titrant. To measure the NH 4 + content, after drying the saturated zeolite particles, the adsorbed NH 4 was desorbed 3 times (5 minutes each) with 1M KC1 (0.2 w/v ratio) by agitating on a reciprocating shaker (100 rpm). The supernatants were then collected and NH 4 + measured with an ion selective electrode (ISE) Orion 95-12 connected to an Orion 4star pH -ISE benchtop (Thermo Fisher).
K + -saturated CLI and CHA (hereafter samples K-CLI and K-CHA, respectively) were prepared in the same way but with 1M KCl solution. The amount of K adsorbed was extracted 3 times with 1M NH 4 Cl solution by agitating on a reciprocating shaker. K content of the supernatant was

Nh 4 + and K + leaching
Among the methods employed to evaluate the quality of SRFs, soil leaching method can provide a more reliable evaluation of the nutrient release feature (Xiang et al. 2008). Hence, this method was used to evaluate the nutrient release feature of different zeolitic NH 4 + and K + sources produced in this research. NH 4 + and K + leaching was measured through column leaching experiments (Faccini et al. 2014). Columns were used for leaching experiment with the following content: filled by unamended soil (C1), soil amended with CLI (C2), soil amended with CHA (C3), soil amended with N-CLI (C4), soil amended with N-CHA (C5), soil amended with K-CLI (C6), soil amended with K-CHA (C7), soil amended with (NH 4 ) 2 SO 4 (C8) and, soil amended with K 2 SO 4 (C9). Three replicates were made to compare the leaching behaviour of NH 4 + and K + , as well.
Columns of 15 cm in height and 2 cm in internal diameter were filled with the above mentioned prepared soil samples. Assuming constant soil bulk density (BD) and the column volume, 65 g of dried soil was required to completely fill the columns. To prepare C4, C5, C6, C7 and C8, proper amounts of NH 4 + or K + source (contains 17.36 mg of ammonium or 12.36 mg of potassium) were homogenously mixed with the entire dry soil volume, then the columns were filled. NH 4 + content of each nutrient source was different. Hence, to have equal amount of NH 4 + (17.36 mg) in C4, C5, and C8 treatments for leaching tests, depending the NH 4 + content of each nutrient source different amounts of N-CLI, N-CHA, and (NH 4 ) 2 SO 4 were added to the columns. In this research work, 0.77, 0.77, and 0.07 g of N-CLI, N-CHA, and (NH 4 ) 2 SO 4 were added to C4, C5 and C8 columns, respectively. In regards to K + , 0.25, 0.27, and 0.027 g of K-CLI, K-CHA, and K 2 SO 4 were added to the columns C6, C7, and C9, respectively. To prepare C2 and C3, a definite weight of CLI and CHA, equal to the maximum amount of applied zeolite in C4 and C5 (0.77 g) was also applied to the soil.
The columns filled with the above mentioned treated soil samples were then saturated with deionized water from the bottom before the experiments. The experimental flushing device consisted of a peristaltic pump with flow rate of 100 ml h −1 accomplished with polyethylene tube (Faccini et al. 2014;Colombani et al. 2015). The leaching experiments have been carried out according to the methodology described in Colombani et al. (2015). The flux was more than one order of magnitude higher than the heaviest flash flood rainfall occurred in the period of 2011-2015 (33 mm h −1 ). The data was recorded by a DAVIS Vantage Pro2 Plus installed in meteorological station of Codigoro about 10 km from the soil sampling area.
A pre-experiment was setup to measure the pore volume of the soil sample. First, the column was filled with air-dried soil and weighted. Then, the soil was saturated with deionized water from the bottom using a pump. The flow was immediately stopped after exiting the first drop and the saturated column was weighted. The pore volume then was calculated by using simple weightvolume relationships. The pore volume was determined as 21 ml. The schematic diagram of the column experiments is presented in Figure 1.
A total of 20 leachate samples (20 PVs) were periodically collected and analysed for NH 4 + and K + measurement. The number of PVs to be collected (20) was determined based the PVs required for the complete depletion of the NH 4 + from column C8.

Calculations and statistical analysis
Shapiro-Wilk and Levene's Test were performed for testing the data normality and homogeneity of the variance. One-way ANOVA and Fisher (LSD) tests were then employed for evaluation of significant differences between the treatments at p = 0.05 level. SigmaPlot 12.3 was also used for statistical analysis.

NH 4 + and K + saturation
The results of saturation experiments showed that 10 and 3 days were required to produce N-CLI and N-CHA, respectively. In the case of K-CLI and K-CHA, the saturation occurred after 7 and 10 days. The shorter time required for saturation of CHA with NH 4 + and K + with respect to CLI can be attributed to its crystal chemistry and framework structure. There are two types of exchangeable sites in chabazite framework located respectively in the small pseudo-hexagonal prism and large ellipsoeidal cages that favourably fits the ionic radius of both NH 4 + and K + (0.13-0.14 nm) (Torraca et al. 1998). According to Wang et al. (2013), the larger pore entrance in chabazite with respect to clinoptilolite results in a better accessibility of cations to exchangeable sites and leads to faster kinetic of adsorption. Additionally, the higher Al content of chabazite than clinoptilolite can increase the presence of exchangeable sites .
The faster rate of chabazite saturation with NH 4 + and K + can also be explained in a different way which seems to be more close to reality. Theoretically, higher Al/Si ratio in zeolites leads to higher adsorption capacity (Li et al. 2014). Here, despite the Al/Si ratio of CHA is approximately 3 times higher than CLI (Table 1), the adsorption capacities of both zeolites were found to be very similar. The faster rate to reach equilibrium state in CHA may be due to the presences of Ca 2+ and K + in its crystal structure leading to the interruption in NH 4 + and K + adsorption on CHA and preventing full saturation.
The results of saturation studies showed that the maximum adsorption for NH 4 + and K + in samples N-CLI and N-CHA was 1.25 mol kg −1 and 1.24 mol kg −1 , respectively. The maximum adsorption for NH 4 + and K + in samples K-CLI and K-CHA was also 1.24 mol kg −1 and 1.14 mol kg −1 , respectively. The saturation values for N-CLI and K-CLI samples nearly overlap the CEC measured for the natural sample CLI (127 cmol c kg −1 ). However, the CEC of CHA (39 cmol c kg −1 ) is underestimated when compared to the maximum adsorption capacity obtained from saturation experiments. The low CEC determined for chabazite can be attributed to the issues caused by cation selectivity of the zeolite during the measurement using conventional method: Chapman's Method (1M NH 4 OAc, pH 7) (Ming and Dixon 1987). In the Chapman's method, the CEC is calculated based on the amount of sodium replaced with exchangeable cations. Since K + is replaced with difficulty by Na + (Ames 1960) the degree of Na + saturation depends on the original amount of K + on the zeolitic exchange sites. As K + and Ca 2+ are the most abundant cations in chabazite (Table 1), replacing these ions by Na + , may has occurred incompletely. These findings are in agreement with the findings of Jama and Yucel (2006) which reported that high amount of K + and divalent cations on zeolite may results in impossibility of Na + replacement in the exchangeable sites. The observed results indicate that the method of Chapman (1965) underestimates the CEC for Chabazite used in this study.
The N-CLI and N-CHA produced in this research contained 22.5 and 22.32 mg NH 4 + , respectively. The K content of K-CLI and K-CHA was 48.75 and 44.46 mg g −1 , respectively.

NH 4 + and K + leaching
The leaching behavior of different N and K sources are illustrated in Figure 2-4. The values represent 3 replicates of each treatment. As shown in Figure 2, NH 4 + leaching in soils amended with NH 4 + -saturated zeolites is drastically lower than soil amended with CF. A sharp decline was observed in pattern of NH 4 + release from C8 during the first three PVs (Figure 2). Following this initial leaching, approximately 59% of total NH 4 + was leached out (Figure 4). The cumulative leaching results shows that NH 4 + leached from C8 is about 84% of the total NH 4 + . While, the NH 4 + leached from the C4 and the C5 was notably low (Figure 4). The slow rates of NH 4 + release from both of the NH 4 + -saturated zeolites can keep its concentration available in soil solution at a lower level reducing the leaching losses. These results are in agreement with findings of Faccini et al. (2014) which reported that only 8.7% of the NH 4 + content of ammonium enriched chabazite was subjected to leaching. The comparison of C1 with C2 and C3 showed that adding natural zeolite to soil have no significant effect on leaching of NH 4 + or K + . This might be attributed to low amount of used zeolite in soil (0.77 g). The importance of zeolite quantity on NH 4 + leaching is also emphasized by Weber et al. (1983).
In order to compare the leaching behaviour of K + with NH 4 + release (Figures 3 and 4), the same number of PVs was considered. Potassium was also released faster and in greater amounts from soil samples amended with CF similar to what already has been observed for ammonium (Figure 3). The cumulative percentage of K + leached showed that more than 61% of the K + added with K 2 SO 4 fertilizer in the C9 was leached within the first three PVs.; while, these amounts were approximately 7% for both C6 and C7 (Figure 4). The total K + leached from the C9, C6, and C7 was about 87%, 23%, and 24%, respectively.
Although the type of zeolite plays a significant role in leaching of NH 4 + , the ANOVA test showed no significant differences between leaching behaviour of K + from the K-CLI and K-CHA (Table 3). It seems that the CHA is more capable of maintaining NH 4 + on its exchange sites with respect to the CLI. The lasting effect of zeolite on nutrient retention in soil is also reported by some previous researchers (Ferguson and Pepper 1987;Li et al. 2013). Table 3. Mean values of leaching from different N and K fertilizers.
The slow and steady releases of both NH 4 + and K + from the saturated zeolites indicated that they are suitably capable to perform a controlled release of NH 4 + and K + and decrease nutrients loss by leaching in sandy soils with low organic matter contents. However, the capability of these zeolitic NH 4 + and K + sources as SRFs should be tested in presence of plant to ensure a rate of nutrient release adequate for plant nutrition.
Due to relative abundance of zeolite mines in both Iran and Italy, the raw material (Zeolite) cost is low. The majority of the cost of producing SRFs by using zeolites in these two countries are transportation and fertilizer production process costs. Hence, the economic comparison of the produced ZNSs with some slow release fertilizers can give valuable information on the economic Figure 3. a) Leaching behaviour of K from different treatments: unamended soil (C1), soil amended with clinoptilolite (C2), soil amended with chabazite (C3), soil amended with K + -saturated clinoptilolite (C6) and soil amended with K + -saturated chabazite (C7), soil amended with K 2 SO 4 (C9). b) Magnified graph representing the data and the leaching behaviour of C1, C2 and C3. feasibility of these fertilizers. Addressing the above mentioned issue is the next goal of this research.

Conclusions
NH 4 + and K + saturation behavior of clinoptilolite and chabazite rich zeolites was investigated using column leaching tests performed on a sandy soil amended with both natural and NH 4 + or K + saturated zeolites to assess their contribution in controlling nutrient loss compared to an unamended soil. Based on the results of this study, the following 6 major conclusions can be drawn: (1) The Chapman (1965) method underestimates the CEC of Chabazite.
(2) In spite of higher Al/Si ratio in CHA, the adsorption capacities of both zeolites were nearly similar. This can be related to high Ca 2+ and K + contents of CHA leading to an incomplete saturation.
(3) The saturation of CHA with both NH 4 + and K + occurred in a shorter period of time than the CLI and this can be attributed to the different crystal chemistry and framework structure of these two main zeolitic species.
(4) Soils amended with the two NH 4 + saturated zeolites, especially the N-CHA, undergo a reliable lowering of NH 4 + and K + leaching with respect to soil amended with CFs. . Cumulative (%) of NH 4 + and K + leached from different treatments: soil amended with NH 4 + -saturated clinoptilolite (C4), soil amended with NH 4 + -saturated chabazite (C5), soil amended with K + -saturated clinoptilolite (C6), soil amended with K + -saturated chabazite (C7), soil amended with (NH 4 ) 2 SO 4 (C8) and soil amended with K 2 SO 4 (C9). (5) Although the zeolite type does not affect the leaching behaviour of K + but the release of NH 4 + from C5 was significantly lower than C4. Hence, the chabazite retains ammonium better than clinoptilolite. (6) The cation exchange and slow release characteristics of both CLI and CHA make them suitable as a precursor for production of SRFs.