Optimization of Struvite Crystallization and Heavy Metal Recovery in Wastewater using Response Surface Methodology

Introduction

MAP (Magnesium Ammonium Phosphate) precipitation could occur spontaneously on the surface of process pipes, pumps and other industrial equipment leading to scale deposition of struvite (MgNH₄PO₄·6H₂O) minerals^1,2. The scale deposition of MAP bearing minerals may decrease the performance of equipment and increase the maintenance cost. In contrast, the induced MAP precipitation in the wastewater promotes the beneficial crystallization of struvite, which is potentially as fertilizer¹. Such precipitation technology has been widely implemented in wastewater treatment units for enhancing phosphate (PO₄-P) recovery³.

In addition to phosphate, the heavy metals and organic pollutants (such as PCB and polycyclic aromatic hydrocarbon) are frequently available in the wastewater, which must be subsequently treated prior to its discharge into the environment. Conversely, their accumulation in the soil becomes a serious problem for sustainable agriculture. Also, strict regulations have been imposed on heavy metal concentrations in fertilizer products. On the other hand, MAP precipitation is known as the best approach to heavy metals and phosphate recovery, which is related to the nucleation and growth of struvite crystals in the solution, whereas desirable amounts of struvite can be precipitated typically through the gravitational settling process¹.

Further, the conventional removal of toxic heavy metals from wastewaters may involve precipitation, cementation, sedimentation, filtration, coagulation, flotation, complexing, solvent extraction, membrane separation, electrochemical technique, biological process, reverse osmosis, ion exchange and, adsorption⁴. In the precipitation method, heavy metal and phosphate can be recovered by the crystallization of hydroxyl calcium phosphate and magnesium ammonium phosphate. However, the crystallization of hydroxyl calcium phosphate requires a long crystallization induction time and a large amount of seed crystal materials, hence leading to the difficulty for wide application of technology in the recovery of phosphate⁵. Recently, MAP crystallization for phosphate recovery offers its good separation performance, quick response rate, and high phosphorus recovery efficiency. This method is widely applied for phosphate removal from calf manure, leather tanning wastewater, swine wastewater, wasted sludge, industrial wastewater, municipal landfill leachate and poultry manure wastewater^{5, 6, 7}.

MAP crystallization in aqueous solution with a 1:1:1 molar ratio of ions of [Mg²⁺], [NH₄⁺] and [PO₄^2-], can be expressed as the following equation (1)¹:

The pH value, supersaturation, the temperature of the wastewater and the presence of metal ions could affect the extent of struvite production and its morphological characteristics^{3, 8, 9}. Specifically, the high heavy metal content makes slower kinetic of the struvite crystallization in term of reduction in value of constant rate^{6, 10}. While the value of constant rate decreases, the crystal growth also decreases due to the presence of a metal ion inhibitor. For instance, Zn was reported to the most effective in inhibiting crystal growth⁸. Correspondingly, a low or very low concentration of Cu, Pb, and Zn can be adsorbed on struvite surface. Moreover, the more concentrations of metal ions available in the solution make the decrease in the production of struvite^{8, 9, 11}. Accordingly, the possibility of increasing the struvite production eventually requires a compromising stage between optimizing metal removal and the need for increasing phosphorus recovery during crystallization. In such case, the improvement of struvite production could be achieved through better control of process parameters influencing on MAP precipitation for metal recovery in the solution.

Correspondingly, the efficiency of MAP precipitation for recovering heavy metals could be influenced by many process parameters including pH, temperature, time, starting molar ratio, and metal concentration. In this way, multiple variables controlling the removal efficiency could be optimized using an effective method, namely the response surface methodology (RSM). The availability of the experimental design software makes this method in the most efficient way of modifying the experimentation required to optimize many parameters. Here SRM has been successfully used to model and optimize crystallization processes related to wastewater systems^{2, 12} and it may be an effective design of the experimental method to optimize struvite production and heavy metal removal from wastewater.

In the SRM approach, the experimental domain (D) could be specified with a number of factors (k-factors), which are represented by independent variables (x₁, x₂,….x_k) and a polynomial model could be governed. Correspondingly, designs for an experimental run could be obtained, i.e. sets of experimental conditions, which present the estimates of less variance for coefficients and response. The polynomial model, with a p + 1 coefficient, is given to relating the experimental response to be optimized, y, with the k factors through the p variables (p ≥ k) as shown in Eq.(2).

Y_i = b₀ + b₁x₁ + b₂x₂ + … + b_kx_k ++ b₁₁x₁² + b₂₂x₂² + … + b_kkx_k² + b₁₂x₁x₂ + … + b_1kx₁x_k + … + b_k-1,_k x_k-1x_k x₁                                   (2)

Where x_k+1,x_k+2, …, x_p are the cross-products and powers of the k factors,

x₁, x₂…x_k, are the codified factors.

In such method, central composite, Dohelert, and Box Behnken designs are commonly employed in SRM, while the researcher can select the most suitable one for approaching the optimization problem.

The present research was undertaken to optimize MAP precipitation by SRM for struvite production and its effect on the metal ion uptake in wastewater. In this study, the central composite design (CCD) was employed to examine the effects of three metal concentrations (Cu, Pb, and Zn) on the optimization of mass yield response of struvite. The crystalline solid product was then characterized using the XRPD Rietveld method for mineral composition and SEM-EDX for morphology and chemical element analysis.

Materials and Methods

Preparation of Crystal-Forming Solution

The supersaturated solutions for the experiments were prepared by diluting the powder of MgCl₂·6H₂O and NH₄H₂PO₄ with analytical grade chemicals (Merck^TM), without further purification.  In this way, both powder crystals were separately dissolved in a double-distilled water to provide the equimolar of 0.25 M for each ion Mg²⁺, NH⁴⁺, and PO₄^-3. The pH of both solutions was adjusted to be 6.5 by adding 0.5 N KOH solution. Cu²⁺, Pb²⁺ and Zn²⁺ stock solution was prepared by dissolving a specified mass of the corresponding CuCl₂, PbCl₂, and ZnCl₂ in deionized water. Accordingly, the solution containing 0, 10 and 100 ppm of those metals was subsequently diluted into the stock standard of the MgCl₂·6H₂O solution. All solutions were then mixed in a 500 ml-glass beaker and stirred at 200 rpm and pH change was monitored by an Orion SA520 pH meter with a gel epoxy probe. The experiments for struvite crystallization was conducted by the gravitational settling process at ambient temperature. The precipitate was then obtained by filtering off in a paper (Schleicher and Schuell no. 604), washed with distilled water, and dried directly on the paper at room temperature. The time required the precipitation and filtration of struvite were about 90 minutes. The mass of dried precipitate was weighed using Sartorius weigh-scale.

Materials Characterization

Scanning electron microscopy (SEM) (JEOL JSM 5200) and energy dispersive X-ray spectroscopy (EDX) were used to examine the precipitate for recovering the ion-metal particles on the surface of struvite mineral. In this way, the dried precipitates with different particle sizes below 100 µm were previously embedded in epoxy on an Al-sample holder and sputtered with carbon for SEM/EDX analysis.

Phase identification of the precipitate was conducted by X-ray powder diffraction (XRPD) method through the data obtained from the Bragg-Brentano diffractometer (Philips 1830/40) measurement. The scan parameters (5-85^O 2q, 0.020 steps, 15 s/step) were set-up for recording XRPD data of the sample. A PC-based search- match program (MATCH software) was initially employed for identifying candidate crystalline phases, which was subsequently verified by the Rietveld method available in the program^{13, 14, 15}. The refined values of the cell parameters were used to calculate (wt. %) levels of mineralogical phases, which were performed by the program¹⁶.

Experimental Design and Optimization of Parameters

In this present study, the input variables (Cu²⁺, Pb²⁺, Zn²⁺ concentrations) and the yield response of the optimum mass (mg) was investigated by SRM and are given in Table 1. A multiple regression data analysis was conducted by the statistical v.6 software packages (StatSoft, Tulsa, OK, USA). The optimum values of the selected input variables were found by solving the regression equation and also by examining the response surface contour plots using equation [3]:

where Y is the estimated response; β₀, β_j, β_ij and β_jj constant coefficients; X_i and X_j are the coded independent variables or factors; ε is random error.

Table 1: Range and level of independent variables.

Independent variables	Low Level (-1)	Range and Level Center Level (0)	High Level (+1)
Concentration Cu2+(ppm)	0	10	100
Concentration Pb2+(ppm)	0	10	100
Concentration  Zn2+(ppm)	0	10	100

 

Results and Discussion

Properties of the Solid Precipitate

The corresponding solids were then subjected to the XRPD Rietveld method as given in Figure 1a. The crystals are primarily composed of struvite and sylvite (KCl). Upon the quantitative Rietveld analyses, struvite was precipitated in the wastewater streams as the major mineral (about 99 wt.%) with a small amount of sylvite (< 0.5 wt.%). Although all data are not shown, obviously, the presence of three metal ions made unchanged the mineral composition of the precipitates (only 50 ppm of Cu, Pb, and Zn is shown in Figure 1b). Figure 2 shows the SEM image of the crystals obtained exhibiting aggregate formations of irregular prismatic-like crystalline of struvite. This formation was also confirmed again by EDX analysis in Figure 2, showing the chemical elements of Mg, O, and P corresponds to struvite composition. Moreover, the uptake of Cu²⁺ on the struvite surface was indicated by the element of Cu in the EDX spectrum. The similar results of metals (Pb and Zn) captured on the struvite surface were also observed by SEM/EDX, although the results were not shown in this paper. Accordingly, MAP precipitation demonstrated the recovery of metals and phosphate from the solution. An economic return of this precipitation process has been previously reported for the struvite production in the wastewater¹⁷. Therefore, this precipitation technology is potentially adapted to yield optimum struvite, which has its capability of heavy metal recovery, and the precipitating product can be used for fertilizer.

Figure 1a: Plot of XRPD Rietveld analysis of the crystals precipitated in the absence of metal ions. b) XRPD patterns of the precipitates with 0 ppm, 50 ppm Cu, 50 ppm Pb, and 50 ppm Zn. Click here to View figure

 

The peaks are S (struvite) and Sv (sylvite) respectively.

Figure 2: SEM image of a prism-like shape crystal morphology and EDX spectra obtained from the precipitates at the 30°C in the presence of 50 ppm Cu. Click here to View figure

 

Predicted Model and Statistical Analysis

Model of response variables was optimized using SRM with the input data (Table 1), which consisted of 3 factorial design 2⁽³⁾ providing nc = 8; ns = 6; no = 2 and run = 16. Correspondingly, the concentration range of Cu²⁺ (X_1, 0-100ppm), Pb²⁺ (X_2, 0-100ppm) and Zn²⁺ (X_3, 0-100ppm) was selected for the calculation. Factors and their levels for SRM correspond to the low level (-1) = 0, high level (+1) = 100, and the center point (0) = 10. The whole design of yield response of mass precipitate (mg) is presented in Table 2.

Table 2: Design of experiments with independent variables.

Run		Independent variables	Zn2+(ppm)	Responses
	Cu2+(ppm)	Pb2+  (ppm)		The mass of precipitate (mg)
1	50	50	50	8.875
2	50	50	100	9.36
3	50	100	50	8.21
4	50	100	100	8.105
5	100	50	50	9.105
6	100	50	100	9.665
7	100	100	50	8.225
8	100	100	100	8.825
9	32.9552	75	75	8.605
10	117.0448	75	75	9.525
11	75	32.9552	75	9.445
12	75	117.0448	75	8.01
13	75	75	32.9552	9.075
14	75	75	117.0448	9.515
15	75	75	75	8.26
16	75	75	75	8.26

 

Regression analysis of the mathematical model for the optimization of independent variables was fitted to the polynomial Eq. [3], and the adjusted model obtained for struvite production (Y) can be expressed as Eq.(4).

Y = 14.56481 – 0.05912 X₁+ 0.00035 X₁²– 0.03652 X₂+ 0.00016 X₂² – 0.06896 X₃+ 0.00048 X₃²  +  0.00004 X₁X₂  +  0.00016 X₁X₃  –  0.00011 X₂X₃            (4)

Analysis of variance (ANOVA) was performed for determining the significance of the model given in Table 3. The function of more significant variables obviously fits with the quadratic regression model, because the calculated coefficient of determination (R²) was quite high (0.982). Moreover, the proposed model is considered statistically significant, according to the F-test with 95 % of confidence. Here F-value is defined as the ratio between MSF (mean squares of the factor) of the MSE (mean squares of error). Correspondingly the F-value of 86.362041 is much greater than the value of F-table (4.1) confirming that the model is significant at 95 % of confidence. Only 1.8 % of the total variation did not fit the model.

Table 3: ANOVA analyses of the yield response of crystalline mass.

Source	Sum of squares (SS)	Degree of freedom (DF)	Mean Square (MS)	F-value	F-table	R2
S.S. RegressionS.S. ErrorS.S. Total	5.141284 0.357191 5.498475	9 6 15	5.141284 0.059532	86.362041	4.1	0.982

 

The experimental results were also analysed in chart Pareto (Figure 3). Upon the analysis, p-value less than 0.05 is considered to have an insignificant effect contributed to the response. On the basis of the results given in Figure 4, it can be seen that concentrationof Pb²⁺ (X₂) has major linear (L) effect for controlling struvite production, followed by concentration (L) of Cu²⁺ (X₁) and Zn²⁺ (X₃); and their quadratic effects (Q) for concentrations of Zn (X₃)²; Cu²⁺ (X₁)² and Pb²⁺ (X₂)².  However, the quadratic effect of concentration of Pb²⁺ (X₂)² can be ignored because they provide insignificant factor on the yield response of struvite precipitated.

Figure 3: Pareto chart of optimization for independent variables on the yield response of struvite production. (L) is the linear and (Q) is the quadratic interaction of variables. Click here to View figure

 

Optimization of Independent Variables for the Optimum Struvite Production

The graphical results of interactive independent and dependent variables are depicted by 3D and 2D contour plots, which enable to determine the optimum recovery of struvite (mass in mg) (Figs. 4-6). Different forms of the contour plots indicate the different interactive effects, where the significant interactions between the variables are shown in an elliptical contour plot. In contrast, a circular contour plot shows the insignificant interactive effects. Figure 4 presents the interaction between concentrations of Cu²⁺ and Pb²⁺ for the yield of struvite. As expected, the increased yield of struvite occurred at the highest concentration of Cu²⁺. However, the increased concentration of Pb²⁺in the solution made the reduced production of struvite.  Additionally, Figure 5 presents the interactive effects of Cu²⁺ and Zn²⁺ concentrations on the yield of struvite. When increasing the concentrations of Cu²⁺ and Zn²⁺ were reached at 70 ppm, the high amount of struvite could be produced. In contrast, the concentrations of Cu²⁺and Zn²⁺ increased to above 70 ppm leading to the reduced production of struvite. Therefore, both Cu²⁺and Zn²⁺ have significant effects on retarding the struvite production. Conversely, the effect of Pb²⁺ on the struvite production was more apparent than Cu²⁺ and Zn²⁺ removal.

Figure 4: Response surface contours for mass scales production showing effects of interactive concentrations of Cu2+and  Pb2+. Click here to View figure

 

Figure 5: Response surface contours for mass scales production showing effects of interactive concentrations of Cu2+and  Zn2+. Click here to View figure

 

Furthermore, the interaction between the concentration of Pb²⁺ and concentration of Zn²⁺ on the yield of the mass of precipitate are shown in Figure 6.  Apparently, the increase of concentrations of Pb²⁺ from 50 to 100 ppm made a decrease in the struvite production. Conversely, the more concentrations of Zn²⁺ available in the solution could be recovered at the increase of struvite production. Of the three metal ions, Zn²⁺could be the most insignificant effect in promoting crystal growth of struvite. This result was in close agreement with the previous reports in the literature^{8, 17}.

Figure 6: Response surface contours for mass scales production showing effects of interactive concentrations of Pb2+ and  Zn2+ Click here to View figure

 

Further, the use of the statistical v.6 software enabled to obtain the response surfaces which in turn allow the determination of optimum concentration from 2D and 3D contours. It can be concluded from Figures 4–6 that the optimum mass of struvite was quite effective to trap metals and the computed results are shown in Table 4.

Table 4: Optimum the mass of struvite (mg).

Factor	Optimum concentration (ppm)	Optimum mass of precipitate (mg)
Concentration Cu2+(ppm)	59.4564
Concentration Pb2+(ppm)	70.7930	8.342
Concentration  Zn2+(ppm)	77.2114

 

Validation of the Predicted Value for the Optimal Variables

To validate the predicted value was performed by conducting experimental run with the optimum values of the process variables. Table 4 presents the optimum yield of struvite calculated using the input data of each concentration of Cu²⁺ (59.4564 ppm), Pb²⁺ (70.7930 ppm) and Zn²⁺  (77.2114 ppm), and hence providing 8.342 mg of the mass of precipitate. Referring to the experimental results is shown in Table 5, the mass precipitated at these concentrations was reached at 8.562 mg. Here the calculated % error in the mass of precipitate response was 2.56 %. This result implies that the calculated response of struvite production has an accuracy of 97.44 %.

Table 5: The predicted and experimental value of the response at the optimum level of metal concentration

Optimum variables		Optimum  Result (mg)	Experimental Result (mg)	% Relative error
Concentration Cu2+(ppm)	59.4564
Concentration Pb2+(ppm)	70.7930	8.342	8.562	2.56%
Concentration  Zn2+(ppm)	77.2114

 

Relative error (%) = [(Experimental result– Optimum result by SRM)/Experimental result] × 100%.

It is noteworthy that the induced MAP precipitation through the formation of struvite from the synthetic wastewater offers an opportunity to recover nutrients and heavy metals. This method produced struvite which can be potential as a slow-release fertilizer. It is not only employed for sustainable recovery of wastewater phosphorus, but also for absorbing metals and organics which can enhance the property of fertilizer product with environmentally friendly. Moreover, a series of batch experiments had been conducted to assay the effect of heavy metals on the optimal struvite product. It demonstrates that the high concentration of metal ions in the range of 50 -80 ppm could be recovered with the optimum struvite crystals produced. Nevertheless, remaining works related to the agronomic bioavailability of struvite should be performed in a greenhouse experiment in order to be admitted as a synthetic plant fertilizer. Agronomic assays showed that struvite found in this study is similar to that of synthesis ammonium phosphate fertilizers.

In terms of the proposed study, pH of 6.5 was found to be a desirable value for removing heavy metal from wastewater by MAP precipitation^{18, 19}, while changes in temperature may have no influence on the struvite production. In wastewater systems containing heavy metals with other inorganic substance, the presence of K⁺ in the solution may have no significant effect on the efficiency. However, the presence of Ca²⁺ may decrease the percentage of Mg²⁺ and NH₄⁺ removal. Additionally, the presence of metal ions in the solution resulted in a reduction of mass and the crystal size of struvite^20,21.

In general, the induced MAP precipitation was applied in this study offering various benefits such as their low and rapid processes, simple operation and control, flexibility to change of temperature. Unlike in biological systems, the precipitation can facilitate variable input loads and flow such as seasonal flows and complex discharge. Whenever it is needed, a chemical plant for wastewater treatment could be easily developed. Besides, the treatment system developed involves only a lower space and installation cost. Their advantages, however, are outweighed by a number of disadvantages such as their high operational costs due to the chemicals required, high-energy consumption and handling costs for sludge disposal. Nevertheless, in reducing additional cost of chemicals for the induced MAP precipitation, an industrial-grade (IG) MgO and Mg (OH)₂ has been proposed as the low-cost magnesium sources, which can result in struvite to be prevalent in the precipitate²¹.

Conclusions

SRM optimization of variable concentrations of heavy metals in the wastewater provided that the availability of Pb²⁺ has the significant effect on the mass yield of struvite during precipitation. The optimum result of the mass response scale (8.342 mg) was reached at the concentration of Cu²⁺  of 59.4564 ppm,  Pb²⁺of 70.7930 ppm, Zn²⁺ of 77.2114 ppm, respectively. Accordingly, a good achievement of modelling and optimization of struvite precipitation for heavy metal recovery was presented in this paper. The influence of selected parameters and thus model validation with insignificance parameters could be confirmed. Finally, an easy, simple, and cost-effective precipitation method could be implemented for wastewater treatment containing heavy metals, whereas the induced MAP precipitation provided the valuable fertilizer product rich in phosphorous.

Acknowledgments

The authors express their gratitude to the University of Pembangunan National “Veteran” Surabaya, East Java, Indonesia for providing this PhD research project.

References

1. Doyle, J.D.; Parsons S.A. Water Res. 2002, 36, 3925-3940.
   CrossRef
2. Capdeviellea, A.; Sykorová, E.; Biscans, B.;  Bélinea, F.; Daumer, M-L.  J. Hazard. Mater. 2013, 244-245, 357-369.
   CrossRef
3. Doyle, J.D.; Oldring, K.; Churchley, J.; Price C.; Parsons, S.A. J. Environ. Eng.-ASCE, 2003, 129, 419-426.
   CrossRef
4. Barakat, M.A. Arabian J. Chem. 2011, 4, 361–377.
   CrossRef
5. Ge, F.; Li, M-M.; Ye, H.; Zhao, B-X. J. Hazard. Mater. 2012, 211– 212, 366– 372.
6. Ronteltap. M.; Maurer, M.; Gujer, W.  Water Res. 2007, 41, 977–984.
   CrossRef
7. Wilsenach, J.; Schuurbiers C.; Van Loosdrecht, M.C.M.  Water Res. 2007, 41, 458–466.
   CrossRef
8. Muryanto, S.;  Bayuseno, A.P.  Powder Technol. 2014, 253, 602-607.
   CrossRef
9. Stratful, I.; Scrimshaw, M.D.; Lester, J.N.  Water Environ. Res. 2004, 76, 437-4199.
   CrossRef
10. Saidou, H.;  Atef, K.; Sami, B.M.; Mohamed, B.A.  Open J. Inorg. Chem. 2015, 5, 41-51.
    CrossRef
11. Suzuki, K.; Tanaka, Y.; Kuroda, K.; Hanajima, D.; Fukumoto, Y. Bioresour. Technol. 2005, 96,  1544-1550.
    CrossRef
12. Demeestere, K.; Smet, E.; Van Langenhove, H.; Galbacs, Z. Environ. Technol. 2001, 22, 1414-1428.
13. Rietveld, H.M. J. Appl. Crystallogr. 1969, 2, 65-71.
    CrossRef
14. Rodriguez-Carvajal, J.  Program Fullprof 2k, version 3.30, Laboratoire Leon Brillouin, France, June 2005.
15. Caglioti, G.;  Paoletti, A.; Ricci, F.P.  Nucl. Instrum. 1958, 35, 223-228.
    CrossRef
16. Mahieux, P.-Y.; Aubert, J.-E.; Cyr, M.; Coutand, M.; Husson, B.  Waste Manage. 2010, 30, 378-388.
    CrossRef
17. Shalaby, M.S.; El-Rafie, Sh.; Hamzaoui, A.H.; M’nif, A.  Chem. Biochem. Eng. Q. 2015, 29, 35-46.
    CrossRef
18. El-Gawad, H.A.;  Kattab, I.A.;  Mahdy, A.N.;  Moselhy, H.; Ibrahim, O.A.  RJPBCS, 2017, 8,1366-1380.
19. Wang, H.; Wang, X-J.; Wang, W-S.; Yan, X-B.; Xia, P.; Chen, J.; Zhao. J-F.  J. Chem. Technol. Biotechnol. 2016, 91, 3045-3052
    CrossRef
20. Ivanetsa, A.I.; Kitikova, N.V.; Shashkova, I. L; Oleksiienko, O.V.; Levchuk. I.; Sillanpaa, M.; JECE. 2014, 2, 981-987.
21. Peng, C.; Chai, L-Y.; Tang, C-J.; Min Ali, X-B. M.; Song, Y-X.;Qi, W-M. J. Chem. Technol. Biotechnol. 2017, 92, 325-333.
    CrossRef

---

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

---

This work is licensed under a Creative Commons Attribution 4.0 International License.