Removal of Phosphorus from an Aqueous Solution by Nanocalcium Hydroxide Derived from Waste Bivalve Seashells: Mechanism and Kinetics.

Excessive supply of phosphorus, a vital macronutrient for all organisms, can cause unwanted environmental consequences such as eutrophication. An increase in agricultural and industrial activities has created a considerable imbalance in the phosphorus cycle with continuing adverse effects on sustainability and ecosystem health, thereby stipulating/postulating the significance of phosphorus removal. A unique and sustainable concept for the removal of phosphorus through the utilization of waste bivalve seashells was proposed in the present study. Flat-surfaced and hexagonally shaped nanocalcium hydroxide particles (∼96% purity) with size ranging from 100 to 400 nm have been synthesized, and phosphorus from its aqueous solution is treated via precipitation. An optimization study has been conducted using the Box-Behnken design of response surface methodology, which highlights that with a calcium/phosphorus mass ratio, pH, and temperature of 2.16, 10.20, and 25.48 °C, a phosphorus removal efficiency of 99.33% can be achieved in a residence time of 10 min. Also, under the same conditions, diluted human urine was analyzed and phosphorus removal efficiency of ∼95% was observed. Through experimental results, semiquantitative phase analysis, and transmission electron microscopy, it has been found that the reaction was diffusion-controlled, which was further confirmed through shrinking core diffusion modeling. The present study manifests the promising potential of waste seashell-derived nanocalcium hydroxide for phosphorus treatment and its precipitation in the form of value-added hydroxyapatite.

Introduction

n class="Chemical">Phosphorus (P) is reclass="Chemical">n class="Chemical">ferred to as a limiting nutrient in the context of water degradation through eutrophication. Excess P concentration can trigger abnormal growth of aquatic plants, particularly, algae.[1] Many researchers have provided evidence highlighting the acceleration of eutrophication when P concentration exceeds 0.02 mg L–1.[2] The United States Environmental Protection Agency endorses a total P concentration of <0.05 mg L–1 for each stream entering any natural reservoir.[3] The European Union has more stringent legislation and recommends threshold limits for P concentration as <0.01 mg L–1 (nonrisk) and >0.1 mg L–1 (risk condition) in lakes and other natural reservoirs.[3] According to the Water Framework Directive, the permissible limit for P concentration has been reduced to 0.1 mg L–1, which was earlier set to 1–2 mg L–1.[4] Various chemical, biological, and physical approaches have been developed for effective P removal, such as chemical precipitation, ion exchange, electrochemical adsorption, membrane filtration, and crystallization.[5−8] Most of these methods possess some limitations, such as high cost, complex operation, low purity, and secondary wastes, while crystallization is mainly referred to as an economically viable method exhibiting a considerable recovery rate, helping in production of valuable products and minimization of environmental risks.[8,9]

Recently, the use of wastes, such as steel slag,[9] calcite,[10] eggshell,[11] red mud,[12] seashells,[13,14] and lime mud,[15] for P removal has been actively attempted, since these materials have low cost and wide availability. n class="Species">Bivalves are oclass="Chemical">ne such group of waste available iclass="Chemical">n abuclass="Chemical">ndaclass="Chemical">nce aclass="Chemical">nd are geclass="Chemical">nerally dumped either iclass="Chemical">n opeclass="Chemical">n fields or iclass="Chemical">n laclass="Chemical">ndfills.[16] The Food aclass="Chemical">nd Agriculture Orgaclass="Chemical">nizatioclass="Chemical">n (2016) reported that Europe, Japaclass="Chemical">n, South Korea, aclass="Chemical">nd Thailaclass="Chemical">nd are amoclass="Chemical">ng the leadiclass="Chemical">ng producers of class="Chemical">n class="Species">bivalves with the estimated quantity of 632 000, 377 000, 347 000, and 210 000 tonnes, respectively. Above all, China itself dumped annually more than 10 million tonnes of bivalve waste in landfills.[16,17] It has been reported in previous studies that the adverse effects of bivalves, such as evolution of hazardous gases, including ammonia, amines, and hydrogen sulfide, and health issues, like malaria, diarrhea, and cholera, could occur when dumped in public waters and open fields.[14] Being rich in calcium, these bivalve wastes can be valorized into desired calcium products, which can be utilized in the removal of P. This can simultaneously eliminate three problems: (i) waste management of bivalve shells; (ii) treatment of P-contaminated wastewater; (iii) reutilization of P in the form of value-added products.

Till date, most of the studies on P removal utilizing n class="Species">bivalves aclass="Chemical">nd other class="Chemical">n class="Chemical">calcium-rich waste materials have been providing contradicting results. For example, some studies emphasized the adsorption of P as the major removal mechanism,[9,18,19] while some other studies stressed on a precipitation mechanism.[5,20,21] Their treatment time in the previous studies ranged from several minutes to days. In addition, none of those studies focused on any sustainable technique to reduce the resultant high pH of water after P removal. There is ample literature available on the removal of P by the golden standard of activated carbons through adsorption, but those studies either focused on low-P-concentrated aqueous solutions or provided considerably low performance with high-P-concentrated waters.[22,23] Therefore, for the removal of high-P-concentrated wastewater, development of new sustainable P removal agents is required. Only few studies on the removal of P from wastewater with seashells were found but, to the best of our knowledge, none with the seashell-derived nanocalcium hydroxide (N-CH).

Thus, there is an urgent need to address these voids and to provide a more sustainable solution that can efn class="Chemical">fectively treat P from wasteclass="Chemical">n class="Chemical">water and form a valuable product without causing any secondary effects to the environment. Therefore, the objectives of the present work are as follows: (i) valorization of bivalves, such as abalone, mussel, scallop, oyster, and Manila clam shells, into N-CH with similar physicochemical properties; (ii) performance of P removal by synthesized N-CH; (iii) the influence of various parameters on P removal efficiency; and (iv) elucidation of a unique and distinctive formation mechanism of the product formed with the interaction of P and N-CH (i.e., hydroxyapatite) through batch experiments, spectroscopic investigations (Fourier transform infrared (FTIR) spectroscopy), semiquantitative phase analysis (X-ray diffraction (XRD)), and shrinking core modeling (SCM).

Results and Discussion

Physicochemical Characterization of Synthesized N-CH

The elemental compositions of raw bivalve seashells were determined by X-ray fluorescence (XRF) and were compared with that of natural limestone, as depicted in Table S1. The obtained elemental compositions of raw n class="Species">bivalves were fouclass="Chemical">nd aclass="Chemical">nalogous to that of class="Chemical">natural limestoclass="Chemical">ne. The semiquaclass="Chemical">ntitative phase aclass="Chemical">nalyses of raw class="Chemical">n class="Species">bivalves and synthesized N-CH were carried out by XRD analysis. All of the peaks of each raw sample were identified to be of calcium carbonate. The two major phases of raw bivalves in the present study were the rhombohedral calcite phase having a space group R3̅c (space group no. 167; PDF card no. 86-0174) and the orthorhombic aragonite phase having a space group Pm̅cn (space group no. 62; PDF card no. 05-0453), which are calcium carbonates presented in Figure a. However, the proportion of the calcite and the aragonite phases varies depending on the type of bivalve. For abalone, the amounts of aragonite and calcite phases were approximately 76 and 22%, respectively. Similar trends in the case of Manila clam were observed. That is, the amounts of aragonite and calcite phases were around 92 and 6%, respectively. For mussel, scallop, and oyster, the calcite phase dominates with around 72, 94, and 92%, respectively, whereas the amounts of the aragonite phase were around 25, 3, and 3%, respectively. Similar XRD patterns for bivalve seashells, particularly oyster and mussel shells, were reported in recent studies, confirming the dominant phase of calcite in oyster vs a mixture of aragonite (62%) and calcite (38%) in mussel shells.[24,25]

Figure 1

XRD patterns and a semiquantitative phase amount of (a) all of the raw bivalve seashells, (b) synthesized N-CH from respective seashells, and of post P-treated N-CH after (c) 1 min, (d) 2 min, and (e) 5 min.

XRD patterns and a semiquantitative phase amount of (a) all of the raw bivalve seashells, (b) synthesized n class="Chemical">N-CH from respective seashells, aclass="Chemical">nd of post P-treated class="Chemical">n class="Chemical">N-CH after (c) 1 min, (d) 2 min, and (e) 5 min.

The XRD patterns of the synthesized n class="Chemical">N-CH with difclass="Chemical">n class="Chemical">ferent bivalves are shown in Figure b. The patterns of the synthesized N-CH with different bivalves showed peaks due to the hexagonal portlandite phase having a space group P3̅m1 (space group no. 164; PDF card no. 87-0674) and the vestigial calcite phase (space group no. 167; PDF card no. 86-0174) that pre-existed as residual (unreacted calcite remnants during N-CH synthesis) along with calcite formation due to carbonation during the synthesizing processes. It was also reported that the small amount of calcite can also be formed due to instantaneous carbonation during the drying process.[25] This signifies that the crystalline characteristics of synthesized N-CH particles with different bivalves were mostly identical. To determine the amount (%) of the phase of calcium hydroxide in the synthesized N-CH, N-CH particles from respective shells were mixed in equal proportion by weight and a semiquantitative phase analysis was performed. The results showed a high proportion of calcium hydroxide in the resultant N-CH (∼96%) along with calcite (∼3%) and halite (∼1%) as minor impurities, depicted in Figure S1. The added value of this mixing procedure includes the determination of any possible deviation in the proportion of calcium hydroxide and P removal performances by N-CH derived from each shell.

The functional groups of bivalve seashells and synthesized n class="Chemical">N-CH were characterized by FTIR, as showclass="Chemical">n iclass="Chemical">n Figure . For bivalve shells, as showclass="Chemical">n iclass="Chemical">n Figure a, spectra were corroborated with XRD aclass="Chemical">nalysis, coclass="Chemical">nfirmiclass="Chemical">ng the vibratioclass="Chemical">n modes of C–O boclass="Chemical">nds iclass="Chemical">n class="Chemical">n class="Chemical">CO32– groups and highlighting the presence of both calcite and aragonite phases. Aragonite exhibits characteristic bands at 1083 cm–1 (ν1 symmetric stretching) and at ∼856 cm–1 (out-of-plane bending-ν2) in its spectrum. Interestingly, for aragonite, ν1 symmetric stretching at 1083 cm–1 is both infrared- and Raman-active, whereas in the case of calcite, it is only Raman-active. Thus, the peak at ∼1080 cm–1 is generally used to determine aragonite from a blend of aragonite and calcite.[26] In addition, ν2 vibrations (out-of-plane bending) are infrared-active for both phases but possess characteristic shifts. For instance, for aragonite, ν2 vibrations emerge at ∼855 cm–1, whereas for calcite, this vibration turns up at ∼875 cm–1. Other bands that were attributed to aragonite are a doublet at 712 (in-plane bending, ν4a) and 701 cm–1 (ν4b) and another doublet at 1464 (asymmetric stretching, ν3a) and 1448 (ν3b). For calcite characteristic shifts for ν3 and ν4, vibrations appear at ∼1453 and ∼713 cm–1, respectively. Overall, the presented data are also in agreement with those from Ferraz et al.[26,27] The minor absorption bands in a range of 2350–2400 cm–1 denote the adsorbed carbon dioxide from the surrounding environment on the surface of the particles.[28] For synthesized N-CH as shown in Figure b, the sharp peak of the absorption band at 3643 cm–1 corresponds to the OH– stretching, confirming the presence of calcium hydroxide.[29] Other minor absorption peaks defining the CO32– groups formed due to carbonation during the synthesis of N-CH.[27]

Figure 2

FTIR patterns for (a) the raw bivalve seashells, (b) synthesized N-CH, and (c) N-CH collected post P removal experiment.

FTIR patterns for (a) the raw bivalve seashells, (b) synthesized n class="Chemical">N-CH, aclass="Chemical">nd (c) class="Chemical">n class="Chemical">N-CH collected post P removal experiment.

The morphological analysis of synthesized n class="Chemical">N-CH was carried out through field emissioclass="Chemical">n scaclass="Chemical">nclass="Chemical">niclass="Chemical">ng electroclass="Chemical">n micclass="Chemical">n class="Chemical">roscopy (FE-SEM). A hexagonal nanoplate (∼100–400 nm) derived from respective shells forming a compact layered structure through multiple self-attachments was observed in Figure . The morphologies of synthesized N-CH samples were analogous, thereby confirming the concordance of similar morphology. Similar results have also been reported in the recent literature, confirming the production of calcium hydroxide with a similar shape and crystal size range.[30,31]

Figure 3

FE-SEM images of synthesized nanocalcium hydroxide derived from (a) abalone shell, (b) Manila clam shell, (c) mussel shell, (d) scallop shell, and (e) oyster shell.

n class="Chemical">FE-SEM images of syclass="Chemical">nthesized class="Chemical">n class="Chemical">nanocalcium hydroxide derived from (a) abalone shell, (b) Manila clam shell, (c) mussel shell, (d) scallop shell, and (e) oyster shell.

Brunauer–Emmett–Teller (BET) analysis was carried out to identify the specific surface area of synthesized n class="Chemical">N-CH. The specific surface areas were very low for all class="Chemical">n class="Chemical">N-CH and fell in the range of 4.5–5.0 m2 g–1. Similar results were obtained in previous studies where the specific surface area of the calcium hydroxide nanoparticles synthesized with calcium nitrate decahydrate in an aqueous medium was approximately 8 m2 g–1.[32,33] Also, from FE-SEM microimages (Figure ), it can be noticed that the nanoparticle surfaces were considerably flat with compact layers that justify the occurrence of low surface area.[32]

Precipitation Performance of Phosphorus

The detailed experimental results for n class="Chemical">PPhosphorus with seashell-derived class="Chemical">n class="Chemical">nanocalcium hydroxide (N-CH) are displayed in Figure a. It was observed in the phosphorus removal tests that P removal rates were spontaneous in the initial 60 s and then gradually reached equilibrium conditions in approximately 10 min. The initial spontaneity can be due to the wide accessibility of N-CH’s surface area for phosphate ions’ contact to persuade a chemical reaction.[15] This probability might gradually reduce due to saturation (discussed in a later section) and thereby considerably reduce the reaction rate prior to achieving an equilibrium state. The precise differences in the PPhosphorus at 10 min by the N-CH derived from different shells are illustrated in Figure b. The minor differences of PPhosphorus observed with each shell-derived N-CH can be a resultant effect of diminutive differences in surface areas and variabilities in contact between N-CH and phosphate ions. The optimum Ca–P mass ratio for maximum PPhosphorus in the present study was approximately 2.16, as shown in Figure . The obtained experimental data showed a close similarity in the patterns of PPhosphorus representing a comparable performance of synthesized N-CH irrespective of the source shell. To confirm this behavior, equal portions by weight of N-CH derived from each shell were mixed and identical experiments were conducted, and the results were found in correspondence with those conducted with individual shell-derived N-CH. This result indicates that the PPhosphorus of synthesized N-CH is not dependent on the source shell.

Figure 4

Experimental results of (a) P treatment through N-CH from the respective sea shell; an equal amount of mixed N-CH was also tested, (b) comparison bar graph for experimental results of P treatment through N-CH from the respective seashell after 10 min. (Experimental conditions: P concentration = 20 mg L–1; volume = 150 mL; pH (prior to the addition of N-CH) = ∼7.5; temperature = 21 °C; and stirring rate = 300 rpm.).

Figure 5

P removal efficiency with a changing Ca–P mass ratio. The Ca–P mass ratio onward, P removal efficiency was found to be maximum (∼97%).

Experimental results of (a) P treatment through n class="Chemical">N-CH from the respective sea shell; aclass="Chemical">n equal amouclass="Chemical">nt of mixed class="Chemical">n class="Chemical">N-CH was also tested, (b) comparison bar graph for experimental results of P treatment through N-CH from the respective seashell after 10 min. (Experimental conditions: P concentration = 20 mg L–1; volume = 150 mL; pH (prior to the addition of N-CH) = ∼7.5; temperature = 21 °C; and stirring rate = 300 rpm.).

Since other anions and cations are also present in wasten class="Chemical">water, it is importaclass="Chemical">nt to aclass="Chemical">nalyze the competiclass="Chemical">ng precipitatioclass="Chemical">n betweeclass="Chemical">n P aclass="Chemical">nd other ioclass="Chemical">ns to class="Chemical">n class="Chemical">N-CH; therefore, a human (male) urine sample was chosen to test for P removal. Table S2 illustrates the average composition of ureolyzed human (male) urine. Figure demonstrates the experimental results with the real urine sample compared with only P in an aqueous medium. A slight dip (∼3.5%) in PPhosphorus was observed with the urine sample when compared with an aqueous solution at the equilibrium condition. This drop can be explained with the presence of SO42– ions, as these ions possess a tendency to form calcium sulfate.[34] It has also been reported that with high sulfate concentration (SO42–/PO43– > 10) sulfate considerably inhibits P removal.[34] Co-precipitation of SO42– and PO43– on calcium nanoparticle surface with the solution containing both SO42– and PO43– was also evident in the recent study, thereby inhibiting P removal.[35] This overall performance highlights the potential of N-CH with high selectivity toward P over other anions.

Figure 6

Experimental results of P treatment in a diluted urine sample and an aqueous solution through mixed N-CH. (Experimental conditions: P concentration in urine ∼22 mg L–1; P concentration in an aqueous solution ∼21 mg L–1; volume = 150 mL; pH (prior to the addition of N-CH) = ∼7.5; temperature = 21 °C; and stirring rate = 300 rpm.).

Experimental results of P treatment in a diluted urine sample and an aqueous solution through mixed n class="Chemical">N-CH. (Experimeclass="Chemical">ntal coclass="Chemical">nditioclass="Chemical">ns: P coclass="Chemical">nceclass="Chemical">ntratioclass="Chemical">n iclass="Chemical">n uriclass="Chemical">ne ∼22 mg L–1; P coclass="Chemical">nceclass="Chemical">ntratioclass="Chemical">n iclass="Chemical">n aclass="Chemical">n aqueous solutioclass="Chemical">n ∼21 mg L–1; volume = 150 mL; pH (prior to the additioclass="Chemical">n of class="Chemical">n class="Chemical">N-CH) = ∼7.5; temperature = 21 °C; and stirring rate = 300 rpm.).

Optimization through the Box–Behnken Methodology

Table illustrates the range of independent variables for investigating the efn class="Chemical">fect oclass="Chemical">n a depeclass="Chemical">ndeclass="Chemical">nt variable, i.e., class="Chemical">n class="Chemical">PPhosphorus. The ranges for X1, X2, and X3 were decided on the basis of preliminary experimental results and relevant literature.[21,36] It is critical to consider the change in the pH value of a phosphorus solution due to addition of N-CH as it modifies the solution pH due to release of -OH ions. Therefore, “E” refers to pH at an equilibrium state after adding N-CH, whereas “I” refers to the initial pH before adding N-CH. The variations in the pH value of the solutions on addition of N-CH particles is shown in Figure S2. The experiment with conditions at the center point (1.35, 8.47E, 25) was repeated five times and similar results were obtained, indicating the repeatability and reproducibility of the data.

Table 1

Independent Factors and the Level for PPhosphorus Optimization Study

	range and levels
factors	lower	center	upper
Ca–P ratio (X1)	0.54	1.35	2.16
pH (X2)	2I	5.5I	9I
	5.76E	8.47E	11.17E
temperature (°C) (X3)	10	25	40

Initial pH value.

pH value at the equilibrium state.

The implementation of response surface methodology (RSM) provides an empirical relationship between independent variables and the response function of a system. A mathematical relationship in terms of a quadratic polynomial equation can be developed, which represents the complex correlations between independent and dependent variables. MINITAB 16, a statistical and regression software, was used to determine the response function coefficients for a dependent variable through the correlation of experimental outcomes. The selected dependent and independent values along with obtained experimental data for n class="Chemical">PPhosphorus are illustrated iclass="Chemical">n Table S3. The obtaiclass="Chemical">ned quadratic polyclass="Chemical">nomial equatioclass="Chemical">n represeclass="Chemical">nticlass="Chemical">ng the respoclass="Chemical">nse fuclass="Chemical">nctioclass="Chemical">n, i.e., class="Chemical">n class="Chemical">PPhosphorus, is presented as followsThe R2 value was determined as 0.9897. The predicted results obtained through eq are contiguous to the experimental outcome, which represents the prediction capability (R2 = 0.9015) of the software used. The coefficients in eq demonstrate that PPhosphorus increases with the Ca–P mass ratio (X1), pH (X2), and temperature (X3). The Ca–P mass ratio and pH values have a dominant effect on PPhosphorus compared with temperature. From Table S3, it can be observed that all of the experiments at a low pH (pH = 5.76E) provided negligible efficiencies since solubilities of calcium products are considerably high in low-pH solutions and therefore unable to precipitate.[37] An efficiency of 97.1% was achieved at values for Ca–P mass ratio, pH, and temperature as 2.16, 11.17E, and 25 °C, respectively. The extent of the influence of temperature is considerably low and can be seen when the Ca–P mass ratio and pH are kept constant at 2.16 and 8.47E, respectively, and with a temperature rise from 10 to 40 °C, with PPhosphorus incremented by only ∼4%.

For extended influences of these independent variables on n class="Chemical">PPhosphorus, aclass="Chemical">nalysis of variaclass="Chemical">nce (ANOVA) was carried out aclass="Chemical">nd the results are depicted iclass="Chemical">n Table . Respoclass="Chemical">nse fuclass="Chemical">nctioclass="Chemical">n predictioclass="Chemical">ns were iclass="Chemical">n correspoclass="Chemical">ndeclass="Chemical">nce with the obtaiclass="Chemical">ned experimeclass="Chemical">ntal results (R2 > 0.90). A comparisoclass="Chemical">n betweeclass="Chemical">n model variaclass="Chemical">nce aclass="Chemical">nd residual variaclass="Chemical">nce caclass="Chemical">n be made through ANOVA by F-test. Similar variaclass="Chemical">nces resemble a ratio class="Chemical">near uclass="Chemical">nity, which iclass="Chemical">ndicates a class="Chemical">narrow possibility of variables to have aclass="Chemical">ny sigclass="Chemical">nificaclass="Chemical">nt iclass="Chemical">nflueclass="Chemical">nce oclass="Chemical">n the respoclass="Chemical">nse fuclass="Chemical">nctioclass="Chemical">n. The iclass="Chemical">ncoclass="Chemical">nsisteclass="Chemical">ncy aclass="Chemical">nd lack of fit are evideclass="Chemical">nt if the probability of the F value represeclass="Chemical">nted by the p value is very small (<0.05). The p value iclass="Chemical">n the preseclass="Chemical">nt modeliclass="Chemical">ng reclass="Chemical">n class="Chemical">fers to the probability of obtaining the observed F value in the case of a true null hypothesis. Smaller p values (<0.05) therefore illustrate the need for repudiation of the null hypothesis. The resultant data from eq are significant, as confirmed by the F-value test and ANOVA by fitting the independent variable’s data with the response function quadratic polynomial model. It was found that all of the coefficients in eq have a p value less than 0.05 (significant), except those for X1X3 and X2X3 (insignificant). The linear effect of coefficients for the Ca–P mass ratio X1 (p < 0.039), pH X2 (p < 0.000), and temperature X3 (p < 0.010) is significant, indicating direct proportionality to PPhosphorus. In addition, the interactive effect of the Ca–P ratio and pH (p < 0.000) is significant, while those of the Ca–P mass ratio and temperature (p < 0.687) and pH and temperature (p < 0.790) are insignificant. These insignificant terms do not have an influential effect on PPhosphorus, therefore being ignored in eq . The p values of the coefficients of quadratic terms, i.e., the Ca–P mass ratio X12 (p < 0.000), pH X22 (p < 0.000), and temperature X32 (p < 0.010), are also significant. For the quadratic model expressed in eq , ANOVA provided an R2 value of 0.9897; F value, 123.55; a probability of ∼0.90; and a coefficient of variation (CV) of 4.3, indicating a high potency of the model and highlighting the accuracy and reliability of experiments (Table ).

Table 2

Analysis of Variance (ANOVA) Test for the Response Function PPhosphorus

source	sum of squares	Df	mean square	F value	p value
regression	16 605.8	7	2372.26	123.55	0.000
X1—Ca–P	15.1	1	15.11	7.9	0.039
X2—pH	1801.4	1	1801.35	93.81	0.000
X3—temperature	202.7	1	202.65	10.55	0.010
X12	601.3	1	601.27	31.31	0.000
X22	1901.3	1	1901.32	99.02	0.000
X32	199.0	1	199.01	10.36	0.010
X1X2	1232.0	1	1232.01	64.16	0.000
X1X3	2.2	1	2.25	0.08	0.687a
X2X3	1.8	1	1.82	0.06	0.790a
residual error	172.8	9	19.20
lack of fit	168.5	5	33.70	31.35	0.003
pure error	4.3	4	1.08
total	16 778.6	16

Insignificant value.

Figure illustrates the response surface plots depicting the relationship between the Ca–P mass ratio, pH, and temperature over n class="Chemical">PPhosphorus. The iclass="Chemical">nflueclass="Chemical">nce of the Ca–P mass ratio with respect to pH aclass="Chemical">nd temperature is showclass="Chemical">n iclass="Chemical">n Figure a–d. It was observed that class="Chemical">n class="Chemical">PPhosphorus increases with the Ca–P ratio till it reaches a maximum value of 2.16 (Ca–P molar ratio = 1.69), beyond which PPhosphorus nearly remained constant, representing a linear relationship, also shown in Figure . Figure b depicts the maximum PPhosphorus, which falls in the Ca–P mass ratio range of ∼1.8–2.16. Figure d also confirms a similar range of the Ca–P mass ratio for the maximum attainable PPhosphorus. A similar study was conducted where the maximum removal of phosphorus (98.85%) was achieved with the Ca–P molar ratio of 2.07, which is much higher compared with the present case, i.e., 1.69.[38] Another recent study also reported that the Ca–P molar ratio of 1.667 was suitable for the formation of hydroxyapatite that is a highly stable and practically insoluble form of Ca–P product in an aqueous medium (solubility product (Ksp) = 3.7 × 10–58).[36,39] In the present case also, the hydroxyapatite phase was observed as a precipitated product (discussed later in detail). However, if the Ca–P mass ratio is greater than the stoichiometric ratio of hydroxyapatite, calcium hydroxide can be formed in addition to hydroxyapatite.[40]

Figure 7

Three-dimensional (3-D) response surface plot for the effect of (a, b) Ca–P and pH, (c, d) temperature and pH, and (e, f) Ca–P and temperature on PPhosphorus.

Three-dimensional (3-D) response surface plot for the efn class="Chemical">fect of (a, b) Ca–P aclass="Chemical">nd pH, (c, d) temperature aclass="Chemical">nd pH, aclass="Chemical">nd (e, f) Ca–P aclass="Chemical">nd temperature oclass="Chemical">n class="Chemical">n class="Chemical">PPhosphorus.

The efn class="Chemical">fect of temperature oclass="Chemical">n class="Chemical">n class="Chemical">PPhosphorus is presented in Figure c–f. Unorthodox results were observed as the PPhosphorus value at around 25 °C was comparatively higher than that at 10 and 40 °C, respectively. A reasonable explanation behind this unorthodoxy can be made using two phenomena that can occur simultaneously: (a) solubility of N-CH and (b) nucleation of hydroxyapatite. Both of these phenomena can be significantly influenced by small variations in temperature.[21,41] At low temperatures, the solubility of N-CH (Ksp = 4.7 × 10–6 at 25 °C) is comparatively high and decreases gradually as the temperature increases (Ksp = 7.83 × 10–7 at 100 °C).[41] Here, the Ksp at high temperatures is lower because of Le Chatelier’s principle, i.e., heat is one of the products of the N-CH solubility reaction affecting the reactivity; therefore, the system tends to shift in the reverse direction, making N-CH less soluble. The case with nucleation of hydroxyapatite is exactly reverse. That is, a relatively higher temperature is favorable for the formation of hydroxyapatite as its formation in the present case is endothermic (ΔH = 195.4, 283, and 305.8 kJ mol–1 with PO43–, HPO42–, and H2PO4–, respectively), depending on the form of phosphate present in the solution, which, in turn, depends on the solution pH.[42] For example, it was reported in previous studies that there is considerable acceleration in the rate of formation of hydroxyapatite with temperature. For instance, the formation of pure phase hydroxyapatite takes place in 24 h at 25 °C, while it takes only 5 min at 60 °C.[21] Therefore, the coupled effect of these two phenomena holds the tendency to compensate each other to an extent, thereby providing slightly higher PPhosphorus at a temperature of ∼25 °C.

The pH of the solution was one of the most influential parameters in practical P treatment operations from wasten class="Chemical">water iclass="Chemical">n the preseclass="Chemical">nt study, as it caclass="Chemical">n sigclass="Chemical">nificaclass="Chemical">ntly afclass="Chemical">n class="Chemical">fect the physicochemical properties of the precipitant (N-CH), nature of contaminants (such as H2PO4–, PO43–, and HPO42–), and even the precipitated product (hydroxyapatite). Specifically, the pH value can significantly influence the solubility of N-CH, as at low pH, the dissolution of Ca2+ ions is comparatively higher, which gradually decreases with increasing pH.[18] The dissolved Ca2+ ions then initiate to precipitate at pH > 12.4 due to the saturation of water with calcium hydroxide.[43] Additionally, the pH value significantly alters the existing form of phosphate in solutions because of its polyprotic nature. The phosphate dissociation equilibrium in an aqueous solution with respective pK values is depicted in the following equations[44]At pH < 2.2, phosphate was present in the form of H3PO4 due to very high acidity, and thereby the precipitation in the form of hydroxyapatite was not possible. When 2.2 < pH < 7.3, the dominant form was mostly H2PO4– and could form a precipitated product. With pH > 7.3, the dominant form of phosphate was HPO42–, and as the result suggests in Figure , it can be the form of phosphate most suited for the nucleation with Ca2+ ions to form the precipitated product. Finally, when the pH value was approximately 12.3, PO43– was a dominant form.[34] Finally, the solubility of hydroxyapatite also followed a behavior similar to that of calcium hydroxide. Its solubility is much higher at low pH, which gradually decreases and almost becomes insoluble at pH > 7.4 with Ksp ∼ 10–59 at 25 °C.[45] This is the reason that even with high solubility of N-CH at low pH, hydroxyapatite cannot be precipitated, providing negligible PPhosphorus. Thus, as per the explanation, the reaction for the formation of hydroxyapatite can be expressed as follows[36,42]The resultant PPhosphorus is the outcome of the synergistic effect of the Ca–P mass ratio, pH, and temperature. Conditions favoring one variable resulted in thwarting another. This causes the origination of an elliptical bubble representing a region with a parallel effect of the most favorable conditions shown in Figure d,f. A continuous drop of 0.033 pH unit °C–1 with increasing temperature was also reported, highlighting the interdependency of these variables.[46] The accurate values of optimized conditions achieving the maximum PPhosphorus (99.33%) for the Ca–P mass ratio, pH, and temperature were 2.16, 10.20E, and 25.48 °C, respectively (illustrated in Figure S5). Similar pH was also observed by a study suggesting a pH range of 9–11 at room temperature for considerably better crystallization of the thermodynamically most stable form of calcium phosphates, i.e., hydroxyapatite.[19,36]

Finally, with stringent environmental regulations, treated n class="Chemical">water with such a high pH caclass="Chemical">nclass="Chemical">not be discharged iclass="Chemical">nto surface class="Chemical">n class="Chemical">waters. Thus, the carbonation method has been applied to control the pH of the treated water. One hundred and fifty milliliters of treated water was further processed with accelerated carbonation having a CO2 injection rate of 100 mL min–1 with a continuous stirring rate of 300 rpm for 2 min. The resulting pH abruptly reduced from 11.17 to 6.71 in only 2 min due to the formation of carbonic acid. A pH ∼6.5–7.5 is considered safe for discharging treated water into surface waters.[47]

Mechanism and Kinetic Studies

The calcareous precipitation containing treated P was found to be n class="Chemical">hydroxyapatite, as coclass="Chemical">nfirmed by XRD aclass="Chemical">nd FTIR preseclass="Chemical">nted iclass="Chemical">n Figures c–e aclass="Chemical">nd 2c, respectively. Miclass="Chemical">nor impurities iclass="Chemical">n the form of calcite (2–3%) were also observed due to atmospheric class="Chemical">n class="Chemical">carbonation during the treatment process. It seems to be a simple chemical reaction between phosphate ions and calcium ions, but this overall reaction mechanism is rather more sophisticated. Depending on the Ca–P mass ratio and reaction time, other reactions are accompanied during the formation of hydroxyapatite. Previous studies provided evidence for the formation of octacalcium phosphate first (Ca–P molar ratio = 1.33), which is highly unstable and transforms into amorphous calcium phosphate (Ca–P molar ratio = 1.5).[21] In later stages, this amorphous calcium phosphate first mutates to calcium-deficient hydroxyapatite [Ca10–(HPO4)(PO4)6–(OH)2–·nH2O], which almost instantaneously transforms into most stable hydroxyapatite [Ca10(PO4)6(OH)2].[21] This overall transformation is very rapid and can be understood by “successive transformation” or Ostwald’s rule of stages for achieving the most thermodynamically stable form, as shown in eq .[48] The transmission electron microscopy (TEM) nanoimages provided a visual and compositional testimony of the newly developed crystalline phase alteration during the P removal experiment, comparable to the previous study.[21] The texture of the mellowly fine nanostructure is shown in Figure S4, indicating hydroxyapatite as homogeneous aggregates.

The kinetic studies of P removal or n class="Chemical">hydroxyapatite formatioclass="Chemical">n through class="Chemical">n class="Chemical">calcium hydroxide were rather more conflicting in previous studies. For instance, a study found the hydroxyapatite formation reaction as of the first order while some studies suggested the need for further investigation.[21] To converge this void, a series of experiments were conducted along with semiquantitative phase analysis by XRD to identify the kinetics behind the genesis of synthesized hydroxyapatite. A 500 mL P solution having a concentration of 209 mg L–1 was treated with N-CH having a Ca–P mass ratio of 2.16. This implies the presence of 104.5 mg of P in 500 mL of the solution. The precipitated particles at 1, 2, and 5 min were separated and semiquantitative XRD analysis was carried out, as depicted in Figure c–e. It was observed that 96.9% of N-CH was converted into hydroxyapatite after a residence time of 1 min, whereas only 44.12 mg (42.22%) of phosphorus was utilized. Theoretically, for this amount of hydroxyapatite, ∼110 mg of P is required. This unorthodox result can be understood with the fact that XRD can only provide the crystal’s surface analysis. After 2 and 5 min of residence time, the N-CH conversions were 96.8 and 98.4%, respectively, which are almost similar to N-CH conversions in 1 min. However, a significant increment in the amount of P removal (50.08 and 64.19 mg) was observed and continued with residence time, as shown in Table S4. A possible theoretical explanation of the observed results can be phosphate ions’ diffusion into N-CH, resulting in a stretching front of the reacted material surrounding an inner core of nonreacted N-CH; brief steps regarding the description of the diffusion mechanism are provided in Figure S5. According to previous studies, the shrinking core model (SCM) for the present reaction behavior can be the best suited kinetic model.[49,50] They also suggested that film diffusion control, surface chemical reaction, and internal diffusion process can be responsible for the slowest step, controlling the system. To confirm that the reaction rate is controlled by a chemical reaction or diffusion, the respective kinetic modeling was performed. Equation expressed the integral rate equation for the chemical process and as the rate-controlling step. If diffusion is considered as the rate-controlling step where diffusion is taking place through an insoluble core (a solid layer) surrounding the unreacted core, eq can be rewritten as shown in eq (49)where KCP and KD refer to the chemical rate constant and diffusion rate constant, respectively; x is the fraction reacted; and t is the reaction time. As per eqs and 8, the plot of [1 – (1 – x)1/3] vs t is a straight line if the rate-controlling step is a chemical process, whereas if the plot of [1 – 2/3x – (1 – x)2/3] vs t is a straight line, this confirms the diffusion process as the rate-controlling step of the system. Figure a presents the linear plot of [1 – 2/3x – (1 – x)2/3] vs t with KD = 0.010 and an R2 value of 0.983. Figure b illustrates a linear plot of [1 – (1 – x)1/3] vs t with KCP = 0.032 and an R2 value of 0.843. Here, the rate constant (KD) was found to be the best SCM fit, based on the correlation coefficient (R2 = 0.983 > 0.843) of kinetic data. Therefore, it can be conferred that diffusion is the dominant rate-controlling mechanism for P removal or formation of hydroxyapatite.

Figure 8

Kinetic plots representing (a) the diffusion rate constant, KD and (b) the chemical rate constant, KCP, at 25 °C.

Kinetic plots representing (a) the diffusion rate constant, KD and (b) the chemical rate constant, n class="Chemical">KCP, at 25 °C.

Conclusions

In the present work, n class="Chemical">N-CH was syclass="Chemical">nthesized from waste bivalve seashells through a chemical precipitatioclass="Chemical">n method followed by efficieclass="Chemical">nt P removal from aclass="Chemical">n aqueous solutioclass="Chemical">n. The syclass="Chemical">nthesized class="Chemical">n class="Chemical">N-CH from each type of seashell possessed corresponding physicochemical properties (hexagonal shape with a size range of 350–400 nm), which highlights wide availability and reutilization potential of the wastes as a valuable raw material. The study conducted by response surface methodology explained the complex dependency of independent variables on PPhosphorus and within each other. The optimized parameters for a maximum PPhosphorus value (99.33%) were identified as the Ca–P mass ratio, pH, temperature, and time values of 2.16, 10.20E, 25.48 °C, and 10 min, respectively. Ureolyzed diluted urine was also treated at the same experimental conditions, and the resulting PPhosphorus value of ∼95% highlights the extent of selectivity for P removal by N-CH. A significant extension in the mechanism and kinetic studies has been made by providing experimental and SCM modeling (R2 = 0.983 and KD = 0.010) evidence exhibiting diffusion as the rate-controlling step. This unique mechanism was rarely discussed before in the context of removal of P from the aqueous solution. However, additional research is required to estimate the feasibility of the N-CH treatment process in pilot or full-scale water treatment plants.

Experimental Section

Materials

n class="Species">Abalone, class="Chemical">n class="Species">Manila clam, mussel, scallop, and oyster shells were collected from a local supplier in Daejeon, South Korea. Hydrochloric acid (35–37% concentration) and sodium hydroxide (97% purity) were purchased from Junsei Chemical Ltd., South Korea. A stock solution with P concentration of 500 mg L–1 was prepared by dissolving potassium phosphate dibasic (K2HPO4) in distilled water, while other lower concentration solutions were prepared by the dilution method. All mentioned chemicals were of analytical grade, and raw materials were used without any further purification.

n class="Species">Human uriclass="Chemical">ne was collected from the male toilet of Miclass="Chemical">neral Resources Research Divisioclass="Chemical">n (Korea Iclass="Chemical">nstitute of Geoscieclass="Chemical">nce aclass="Chemical">nd Miclass="Chemical">neral Resources, Daejeoclass="Chemical">n, South Korea). The uclass="Chemical">nequally diluted aclass="Chemical">nd ureolyzed (4 days) uriclass="Chemical">ne (with class="Chemical">n class="Disease">flushing in urinals) was diluted further to meet the desired initial P concentration before the experiment. The average composition of ureolyzed urine is illustrated in Table S2.

Synthesis of Nanocalcium Hydroxide (N-CH)

The shells were separately washed and dried at 105 °C for 4 h in an oven. The shells were then pulverized with a vibrating cup mill (TMC, Fritsch, Germany), and sieved particles (<75 μm) were used for the experiments. For the elimination of n class="Chemical">carbon dioxide aclass="Chemical">nd volatile impurities, sieved shells were calciclass="Chemical">nated at 900 °C for 2 h.[14] Tweclass="Chemical">nty grams of the calciclass="Chemical">nated shell was theclass="Chemical">n carefully mixed with 250 mL of class="Chemical">n class="Chemical">hydrochloric acid (1 M) at a constant stirring rate of 300 rpm. The solution was then filtered (MF-Millipore membrane filter, mixed cellulose ester, 0.2 μm) and heated to 90 °C since the solubility of atmospheric CO2 in water is minimum at this temperature.[14] This temperature also greatly favors the origination of a perfect hexagonal calcium hydroxide nanoparticle.[30] Finally, 250 mL of the sodium hydroxide solution (1 M) was added dropwise to the filtered solution at the same stirring rate. The mixture slowly turned milky in color, which confirms the synthesis of N-CH. These particles were then separated through filtration and carefully washed to remove any trace amount of HCl and NaOH. The synthesized N-CH were stored in a desiccator after drying at 105 °C for 4 h. This process was repeated to synthesize N-CH particles from all five bivalve shells, i.e., abalone, Manila clam, mussel, scallop, and oyster shells. A schematic diagram for the synthesis of N-CH from bivalve shells is illustrated in Figure S6.

Test Methods

Batch Experiments

The experiments to investigate the efn class="Chemical">fect of Ca–P mass ratios, pH, coclass="Chemical">ntact time, aclass="Chemical">nd temperature oclass="Chemical">n the P removal from P stock solutioclass="Chemical">n were performed iclass="Chemical">n a series of batch experimeclass="Chemical">nts. The pH adjustmeclass="Chemical">nts were made through class="Chemical">n class="Chemical">HCl (1 M) and NaOH (1 M). The experiments were performed with a fixed P concentration of 20 mg L–1 at 21 °C with constant stirring at 300 rpm. P stock solutions of 150 mL were treated with 3, 7, 10, 12, and 14 mg of N-CH, respectively, while phosphorus removal efficiency (PPhosphorus) was recorded at designated time intervals until the equilibrium was obtained. For minimizing the sample turbidity, each sample was filtered (MF-Millipore membrane filter, mixed cellulose ester, 0.2 μm). Figure S7 represents a schematic diagram of P removal from stock solution with bivalve-derived N-CH.

Chemical Analysis

Total n class="Chemical">phosphorus, total chemical oxygeclass="Chemical">n demaclass="Chemical">nd (COD), aclass="Chemical">nd class="Chemical">n class="Chemical">ammonia analyzing kits “HS-TP-L” (range: 0.01–3 mg L–1), “HS-COD-Mn-L” (range: 0.6–20 mg L–1), and “HS-NH3(N)-L” (range: 0.2–6 mg L–1) in accordance with the standard ascorbic acid, Nessler, and reactor digestion spectrophotometric method, respectively, have been used to evaluate the P, total COD, and ammonia concentrations in the required experiments through a UV–vis spectrophotometer (Humas Co., Ltd., South Korea).[51] Sulfate, chloride, and nitrate were determined by ion chromatography (881 compact IC pro, Switzerland). The measurements for P, total COD, and ammonia were conducted at 880, 540, and 415 nm wavelengths, respectively, which correspond to the maximum absorbance. The P efficiency was calculated as follows9where C0 and C denote initial and terminal concentrations of P in the solution with N-CH.

Optimization through Response Surface Methodology

The efn class="Chemical">fects of the three iclass="Chemical">ndepeclass="Chemical">ndeclass="Chemical">nt variables, i.e., Ca–P mass ratio (X1), pH (X2), aclass="Chemical">nd temperature (X3), oclass="Chemical">n class="Chemical">n class="Chemical">phosphorus removal efficiency (PPhosphorus) were investigated through the Box–Behnken statistical experiment design of response surface methodology (RSM). The optimization method includes response analyzation of statistically designed combinations, identification of coefficients best fitted for the experimental data of that response function, prediction of response for the fitted model, and analyzing the model adequacy. A mathematical relationship in terms of a quadratic polynomial equation was developed, which represents the complex correlations between independent and dependent variables. This eq can be expressed aswhere “l” is the coefficient of independent variables and Z denotes the response function.

Crystalline Characteristics

The phase identification and structural analysis of raw shells and produced n class="Chemical">N-CH were iclass="Chemical">nvestigated by powder X-ray diffractioclass="Chemical">n (XRD) operaticlass="Chemical">ng at 40 kV aclass="Chemical">nd 30 mA employiclass="Chemical">ng Cu Kα radiatioclass="Chemical">n (λ = 1.54 Å) iclass="Chemical">n a raclass="Chemical">nge of 2θ from 10 to 90° (BD2745N, Rigaku, Tokyo, Japaclass="Chemical">n). Semiquaclass="Chemical">ntitative phase aclass="Chemical">nalysis (X’Pert MPD, Philips, Netherlaclass="Chemical">nds) was used for iclass="Chemical">nvestigatioclass="Chemical">n of sample purity aclass="Chemical">nd kiclass="Chemical">netic study. A Fourier traclass="Chemical">nsform iclass="Chemical">nfrared (FTIR) spectclass="Chemical">n class="Chemical">roscopy analysis (6700 FTIR, Thermo Scientific Nicolet, Massachusetts) was conducted in the range of 400–4000 cm–1 for the identification of characteristic functional groups. The specific surface area was obtained using a Brunauer–Emmett–Teller (BET) analysis through Quadrasorb SI, Quantachrome Instruments, Florida. For microstructural analysis, samples were distributed on aluminum stubs followed by platinum coating through a sputter coater. A field emission scanning electron microscopy (FE-SEM) analysis (D1627, Sirion, Eindhoven, Netherlands) was then conducted to investigate the microstructural characteristics of synthesized N-CH. Transmission electron microscopy (TEM) analysis (JEM 2100, JEOL, Japan) was conducted to determine the nanostructural characteristics of the product formed after P removal (i.e., hydroxyapatite).