Mechanochemical synthesis of eco-friendly fertilizer from eggshell (calcite) and KH2PO4

The present article, based on reverse logistic, provides an environmentally beneficial and cost-effective method of producing calcium phosphate bioceramics (potassium calcium phosphates) aimed for soil conditioning, by a free-solvent mechanochemical process between eggshell waste and KH2PO4. The K3CaH(PO4)2 and possibly CaKPO4 phases formed provide a better nutrient management of P, K and Ca, when applied in soil system the material first release K and form hydroxyapatite and important P and Ca bioavailable source at pH ~ 5, which comprises a large part of many countries soils. Key-words: smart fertilizer, potassium calcium phosphates, eggshell, KH2PO4, nutrient management.


Introduction
Nowadays, the production of new materials is closely related to the reverse logistics, considered one of most important instruments for applying shared responsibility for the products lifecycle. Reverse logistics is an instrument of economic and social development characterized by a set of actions, procedures and means to enable the collection and restitution of solid waste to the business sector, for reuse, in its cycle or in other cycles productive, or other environmentally appropriate final destination (Tibben and Lembke, 2002), the reverse logistic is perfectly applied in the agriculture context. An application of this concept is to provide novel processes for the synthesis of different calcium and / or potassium phosphates from any source of calcium carbonate (CaCO3), making it a soil conditioner. Among sources, eggshells are important wastes of food industries and excellent CaCO3 sources (more specifically calcite) (Byeon, 2016), with monobasic potassium dihydrogen phosphate. Everyday millions of tons of eggshells are produced as biowaste around the world, disposed in landfills with no pretreatment and producing odors due to microbial growth.
Normally the reactions are processed by solid state mechanochemical activation, to produce different materials concerning physico-chemical features, especially the solubilization and diffusion of its constitutional elements and chemical species.
Basically, the solid-state mechanochemical activation consists in applying mechanical energy to promote a chemical reaction (Kouznetsov et al., 2016) from solid reagents using mainly friction and impact forces (Borges et al., 2015;Borges et al., 2016;Borges et al., 2017;Borges et al., 2018;. The process mechanism involved is still quite complex to evaluate, but the applied energy must be large enough to be able to break the reagent bonds and promote solid phase ion migration, in the case of this study, to form a new structure. The process mainly produces K3CaH(PO4)2 and possibly CaKPO4, among other non-crystalline compounds and also CO2 resulting from CaCO3 decomposition.
Compared to KH2PO4, all these phosphates may exhibit very different solubility according to the different bonds between different metal cations. It is influenced by the size of the metal cation and its charge, influencing the reticular energy of the structure (Jenkins et al., 1999). Besides that, the experimental release essays reveal that potassium dihydrogen phosphate exhibits much faster solubilization behavior than calcium or calcium and potassium phosphates. Composed solely of fertilizer components (HPO4 2-/PO4 3-, Ca 2+ and K + ) (Liu and Lal, 2015), the produced potassium calcium phosphates may be classified as smart release fertilizers, since they provide a delay of the bioavailability of P and K and also improve the bioavailability of Ca turning these materials an excellent alternative in these nutrient management. In addition, the solid-state mechanochemical activation allows the reactions to occur in a solvent-free environment (Zhao et al., 2020;Wang et al., 2018), producing no liquid or solid waste in the process, ensuring huge savings during the production process (Mujumdar and Huang, 2007).
The literature reports the mechanochemical synthesis of hydroxyapatite (Ca10(PO4)6(OH)2) from different precursors to those reported in this study (Chaikina et al., 2019), and even if from eggshell, the method uses phosphoric acid solution instead of powder phosphate salts (Francis and Rahman, 2016). Thus, as far as the authors know, no mechanochemical process has been reported for the production of calcium and potassium hydrogen phosphate (K3CaH(PO4)2) and potassium calcium phosphate (CaKPO4) from calcite provide by eggshell and KH2PO4, i.e. different structures from hydroxyapatite including potassium as one of the main components. It is in this sense that new technologies of ecological fertilizer production have been prioritizing environmental care using industrial and other potential wastes in the formulation of new products (Ivanov et al., 2019;Krishna et al., 2017;Santos et al., 2019;Cichy et al., 2019), avoiding unnecessary economic losses, are primarily relevant in the current socio-environmental scenario.

Materials and Methods
The eggshells were collected by the authors in their homes, washed with water and dried at room temperature, presenting a diameter between 1 and 10 mm before the grinding process; The KH2PO4 PA was supplied from Vetec.
The milling process was performed with high-energy ball mill, Fritsch Pulverisette 7, with a 250 ml zirconia bowl, 5 zirconia balls of 15 mm diameter. The essays were performed using the weight ratio of 1:1 (w/w) of eggshell and KH2PO4 were performed firstly for 15 min, 30 min, 1 h, 2 h, 3 h, 6 h, 12 h and 24 h at 600 rpm, and secondly at 100, 200, 400 and 600 rpm for 2 h, in order to evaluate acid-basic reactions between these precursors to form potassium calcium phosphates. Considering the purity of KH2PO4 and assuming that the eggshells are mainly composed by CaCO3, the mixture corresponds to 0.75 mmol of KH2PO4 and 1 mmol of CaCO3. After milled, the samples were characterized by X-ray diffraction (XRD), nuclear magnetic resonance, magic-angle spinning (NMR) MAS, thermogravimetric analysis (TGA), and differential scanning calorimetry (DSC). Details of each characterization are described in Supplementary Material.
The nutrient content of the samples and precursors were obtained dissolving the samples by using acid digestion in H2SO4/H2O2 PA mixture (8:1 v/v, by 24 h at 250 °C), and determined by atomic absorption (K and Ca) and colorimetric (P) methods (Table S1, Supplementary Materials). Before release essays at citric acid solution (2 wt%), the material has been pelletized to simulated commercial products. The pellets were obtained in steel pelletizer applying the pressure of 5 bar, and the weight was standardized to 100 mg/L of P. One single pellet were used in each essay in triplicate, that was placed in 10 mL beaker, this small beaker was immersed into 300 mL of 2% citric acid solution in a 600 mL beaker, the system was kept closed with polyvinyl chloride (PVC) film, in a constant temperature room at 25 °C and under a magnetic stirring. The stirring ensures homogeneous distribution of diffused elements, and aliquots of 1.2 mL at different times (from 15 min to 30 days) have been collected to evaluate the release behavior and allow the calculation of a release kinetic model of Ca, K and P.
The phosphorus diffusion rates in soil were performed in Petri dishes containing moist soil, according to the method described by Degryse and Mclaughlin (2014). This method consists to use a pre-treated paper to mark the P diffusion in soil. In each Petri dish with diameter of 8.6 cm, 78 g of dry and sieved 2.0 mm mesh soil were added to achieve the desired density of 1.2 g.cm -3 . The chemical and physical soil properties are shown in Table S2. In order to achieve 70% of the maximum soil water retention capacity, 15 mL deionized water were added. The papers are scanned and analyzed using image software (GNU Image Manipulation Program, v. 2.6.11, Free Software).
The scanned images are converted to black-with binary images, using a threshold color value (e.g. on a 0-255 scale), and the area of the high P-zone is quantified using the histogram command.
In order to investigate the release behavior closer to real conditions, soil release essays were carried out under aerobic conditions in plastic container containing 50 g of sandy soil (Table S2) with 75% water saturation. The experiments were performed at 3, 7, 15 and 30 days, using a single pellet for each container, the pellet weight was standardized to 300 mg of P per kg of soil. After each release period the solid residues samples were removed from the soil and characterized by XRD; potassium and calcium were extracted from the soil by HCl (0.05 Mol/L) and KCl (1 Mol/L) methods, respectively; and phosphorus was extracted from soil using Mehlich 1 method (Embrapa, 1979).
A colorimetric analyses of molybdenum blue was used to quantify phosphorus released content into the solution (Murphy and Riley, 1962) and potassium and calcium released content were quantify by flame atomic absorption spectroscopy (FAAS), with a PerkinElmer PinAAcle 900T instrument, in flame atomization mode.

Results and Discussion
The XRD patterns of the powders obtained after milling are shown in Fig. 1. The milling process between KH2PO4 and eggshell (98% of CaCO3) performed from 2 h and 400 rpm produce a mixture of phosphates, mainly K3CaH(PO4)2 (PDF database n° 22-1218 indicated by black columns, and black dash lines guiding the comparison with the samples) and possibly CaKPO4 (PDF database n° 33-1002 indicated by red columns, and red dash lines guiding the comparison with the samples). Bellow these milling parameters (time and milling speed) only the reactants diffraction patterns are found, i.e., no evidence of chemical reaction is observed. There is a direct relationship between the time and milling speed and the increasing of total energy involved. In general, it was noticed that only a minimal energy is required to initiate the mechanochemical reaction and, as a consequence, the diffraction patterns have huge changes attributed to the formation of different polymorphic potassium calcium phosphate phases. When milling time increases, secondary reactions begin, for example, mechanochemical dehydroxylation of the new formed phosphates. One can propose the non-stoichiometric reaction of the main products, with CaKPO4 more evident after 24 h of milling: The TGA curve of KH2PO4 show two main loss events: since it is an anhydrous material, the moisture loss is not seen and the first step at 200°C should be associated to crystallization water loss. The second step at 300°C corresponds to polycondensation, concomitant with structural water loss. The residual mass (89%) regards to KPO3 formation is consistent to the theoretical (87%) (Fig. 2): The thermal degradation of eggshell has two main steps: a first mass loss step below 300 °C attributed to moisture elimination concomitant with organic membrane degradation; and a second step related to calcite decomposition below 700 °C (Cree and Rutter, 2015;Omari et al., 2016). The eggshells are mainly composed by CaCO3, as attested by XRD pattern (Fig. 1), which is also confirmed by the small difference from experimental residue (56%) and the theoretical (59%): In the case of milled samples (Fig. 2), three mass loss events are observed: the first step, centered at 280°C, is related to crystallization water loss concomitant with physically adsorbed moisture elimination. The second step, centered at 420°C, is associated to partial structural water loss and structure dimerization, and the final step, centered at 710°C corresponds to final structural water loss reaction and structure polycondensation (Vlase et al., 2005). In addition, all the samples milled at 600 rpm do not present of calcite decomposition below 700 °C but in the samples milled for 2 h at 400 rpm this event is still observed. It suggests that in low rotation speeds do not provide enough energy to degrade the total amount of calcite to form calcium potassium phosphates. It is observed that all samples processed at 600 rpm present a similar thermal decomposition, but with different residuals masses according to the milling time. Water loss at 340°C reduces with milling time probably due to water formation and adsorption during the process and subsequently desorption increasing time milling.

Figure 2
Thus, to better understand these differences and the effect of milling process in material structure, the milled samples were calcined to characterize the solid fractions after each main loss mechanism (280 °C, 420 °C and 710 °C) (Fig. 3). In all cases, a similar thermal behavior related to crystalline phases formation was identified -except for milling during 2 h at 600 rpm, this sample has an unclear phosphate phase, but after   (Fig. 4). The shift at δ = 3.28 ppm (2 and 6 h at 600 rpm) and at δ = 2.82 ppm (24 h at 600 rpm) are reported to calcium phosphates compounds, for example, hydroxyapatite (Ca10(PO4)6(OH)2) and octocalcium phosphate (Ca8H2(PO4)6.5H2O) (Frossard et al., 1994). In addition, the signal at δ = 0.74 ppm is related to disubstituted phosphate, suggesting H(PO4)2 5species agreeing with the KCaPO4 formation.

Figure 4
A shift was also observed at around δ = -5 ppm (marked at δ = -4,78 ppm) in sample milled by 2 h at 600 rpm, attributed to pyrophosphate formed by phosphate dimerization (Godinot et al., 2017). It suggests that this phosphorous chemical environment is an intermediate compound during K3CaH(PO4)2 and KCaPO4 formation.

Solubility kinetics
The kinetic study of the nutrient release was firstly evaluated in citric acid solution (wt 2%) according to the recommended procedure from MAPA to analyze fertilizer solubility (MAPA, 2014) (Fig. 5), and according to the total content of each nutrient in the samples and precursors (Table S1).

Figure 5
Phosphorus and potassium release behavior were similar for the precursor (KH2PO4) and milled samples, but slower for milled samples, this fact shows that the phosphates obtained by milling improve the nutrients management maintaining their bioavailability. On the other hand, calcium release is rather different for eggshell to the milled samples. Even that CaCO3 in the eggshell may be soluble in acids, the kinetic was extremely slow in comparison, evidencing the solubility of phosphates formed.
The nutrients release curves were fitted using the linear form of pseudo-first order, pseudo-second order and intraparticle diffusion models -used in the literature.
But, phosphorus and potassium presented better correlation with pseudo-second order model, considering 'R 2 ' and the values related to concentration at equilibrium time "qe" and "experimental qe" ( and increasing the calcium bioavailability. Table 1 3

.2 Soil release essays
The kinetic study of the nutrient release was also evaluated in soil (Fig. 6). The temporal profile for the precursor KH2PO4 indicates a P increase from 0 mg/kg to 107 mg/kg after 3 days of release keeping constant after 30 days, representing 40% of total P from precursor (KH2PO4). Differently, for sample 2 h/600 rpm, the phosphorus release is more progressive (from 0 mg/kg to 25 mg/kg at 3 days and 45 mg/kg at 30 days) corresponding to 15% of total P content recovered from the milled sample. These results indicate that sample 2 h/600 rpm presented a longer delay in phosphorus release compared to the precursor (KH2PO4), corroborating with the results observed in citric acid solution.
In addition, an extra experiment was done to visualization of the P diffusion zone and comparison with KH2PO4 and Struvite. Then, Fig. S3  slower than precursor at the end of 30 days. The kinetic study of K release in soil shows a slight decrease at 3 days when 2h/600rpm sample is compared to KH2PO4, after that, both materials reach around 100% release (Fig. 6).

Figure 6
There is no increase in the bioavailable calcium when eggshell precursor (CaCO3) is applied in soil, but this concentration increases to 30.97 mg/kg after 30 days for 2 h/600 rpm sample, corresponding to around 8% of calcium recovery from sample ( Fig. 6). These results corroborate with citric acid solution release essays and attest efficient use of 2 h/600 rpm sample as a source of bioavailable calcium compared to the precursor eggshell -CaCO3. The slower release for calcium and phosphorus for 2 h/600 rpm sample and the P radial diffusion are probably related to hydroxyapatite formation -(Ca10(PO4)6(OH)2) (PDF database n° 9-432 indicated by green columns) -identified from 3 until 30 days in soil conditions (Fig. 7). In an additional experiment, samples ground at 600 rpm were suspended in water, dried and then characterized by XRD, in which the phase transition to Hydroxyapatite was observed for all of them (Fig. S5).
Studies of the crystallization behavior of the precursor and samples were carried out using a TA Differential Scanning Calorimeter (DSC) (TA Q100 Controller) with refrigerated cooling system (RCS). Thermal degradation of samples was evaluated in the 25°C -600°C range using a Q500 analyzer (TA Instruments, New Castle, DE, USA) under synthetic air flow of 60 mL.min -1 (80% N2 and 20% O2) with and heating rate of 10°C.min −1 .
The thermogravimetric analyses were performed in a Thermogravimetric Analyzer TGA-Q500 (TA Instruments, USA) under synthetic air atmosphere (N2/O2) at a flow rate of 60/40 mL/min and using platinum crucibles. The temperature range used was ~20 to 900 °C at 10 °C min -1 .

31
P NMR MAS analysis were carried out using a Avance III HD / Bruker equipment. The sample was rotated at 10KHz, using the sequence HPDEC (Direct Bias with decoupling) and the chemical shift reference was ADP (Adenosine diphosphate) with phosphate line at 0.8 ppm.  Figure S1 -XRD data for 2 h/600 rpm before and after thermal treatment at 280 °C.

Kinetic study in citric acid solution 2%
Soil release essays