Phosphorus recovery by struvite precipitation: a review of the impact of calcium on struvite quality

Precipitation of phosphorus (P) as a struvite product from wastewater and local magnesium (Mg) sources is considered as a high economical prudent venture, as it has shown to bring down the cost of producing the product. However, there is a signi ﬁ cant concentration of calcium (Ca) in these Mg and most wastewater sources. The presence of Ca in wastewater and Mg sources during struvite precipitation has been shown to negatively affect the quality of the struvite produced. The low struvite quality greatly impedes its use as a fertiliser given the economic value of the product as an alternative P fertiliser. This review paper provides an understanding of how the presence of Ca affects the purity, crystal size, and morphology of the struvite. Different methods that have been used to mitigate the effect of Ca on struvite quality have been reviewed. The factors that affect the recovery of P and the percentage recoveries of P by different Mg sources have also been reviewed. This paper serves as a basis for further research where the possibility of removing Ca from wastewater and Mg sources is explored. When Ca is removed before struvite precipitation is carried out, the quality of the product will be greatly improved.

The precipitation of struvite requires the chemical addition of magnesium (Mg) in a non-spontaneous reaction to form magnesium ammonium phosphate (MgNH 4 PO 4 ·6H 2 O) (Barbosa et al. 2016).
The commonly used Mg sources used in the struvite reaction in their solution form are MgCO 3 , MgSO 4 , MgO, Mg(OH) 2 , and MgCl 2 . The alkaline nature of MgCO 3 , MgO, and Mg(OH) 2 helps by increasing the pH during the precipitation reaction. The struvite precipitation produces H þ , which decreases the pH, and so requires periodic dosing of NaOH to increase the pH. However, MgSO 4 and MgCl 2 are widely used for phosphorus recovery due to their greater solubility because of their reduced reaction time (Wu & Bishop 2004). Nevertheless, the use of these external Mg sources increases the cost of producing struvite and discourages the production of struvite as a fertiliser. Barbosa et al. (2016) reported that the cost of operation for the crystallisation of struvite mainly arises from the cost of reagent, power, and sometimes transportation. An increase in 75% cost has been reported due to the addition of magnesium (Peng et al. 2018), and this has hampered the application of struvite as a fertiliser due to the high operational cost (Wang et al. 2018). Research has shown that the addition of low-cost magnesium sources can significantly reduce the cost of producing struvite between 18 and 81% (Huang et al. 2011;Lahav et al. 2013 as cited by Wang et al. 2018).
Some natural Mg sources, which have shown potential to precipitate struvite, are wood ash (Sakthivel et al. 2012;Huang et al. 2017), brucite (Zhang et al. 2009;Huang et al. 2011;Hövelmann & Putnis 2016), dolomite (Roques et al. 1991;Holmberg et al. 2000;Chen et al. 2017), and bittern, brine, or seawater (Li & Zhao 2002;Ye et al. 2011;Lahav et al. 2013;Liu et al. 2013;Tran et al. 2016;Huang et al. 2017). These natural sources contain a significant amount of calcium. This review paper discusses the impact of Ca on struvite quality in P recovery, and the methods used for reducing the impact to date has been reviewed to enable the use of an effective method in future studies.

OVERVIEW OF RESEARCH
In the past 29 years, there has not been a clear pattern ( Figure 1) in terms of published work in studies on the effect of Ca on struvite quality until recently where publication in this area has increased. A total of 60 non-medical publications have been reviewed in the Scopus database. This paper provides a review of the influence of Ca on struvite quality specifically on purity, size, and morphology of the precipitate. In addition, factors that have been used to mitigate the effect of Ca have also been reviewed. In this review, the nutrient (P and N) removal efficiencies of different Mg sources have also been considered. Figure 1 shows a general view of non-medical research on how Ca influences struvite quality since 1991. The number of publications has been very low with an increase from 7 in 2019 to 11 as of October 2020. This can be corroborated by Li et al. (2019aLi et al. ( , 2019b in their review paper where they reported low publication on struvite quality research which is needed for industrial application. However, research activities in struvite precipitation have rather seen a gradual increase since 2010. The direction of research (Table 1) has been how Ca concentration in solution influences the struvite purity, crystal size, and morphology and the methods used to mitigate the negative impacts it has on struvite products. Published paper by wastewater type ( Figure 2) shows that majority of the wastewater source used is synthetic. The synthetic wastewater enables the researcher to manipulate the concentrations of Ca, Mg, NH 4 -N, and PO 3 -P in solution and observe how the different molar ratios affect struvite size, purity, and morphology. The different concentrations are also manipulated to enable a researcher to observe how the negative impact of Ca on struvite quality can be mitigated.
Currently, China is leading the research, in terms of published papers, on the influence of Ca on the quality of struvite as can be seen in Figure 3.

Struvite purity
The purity of the struvite product is determined by the percentage mass of the pure struvite (MgNH 4 PO 4 ·6H 2 O) in the total precipitate. The higher the struvite purity, the greater its effectiveness as a fertiliser.
Different methods have been used to determine struvite purity in the literature by analysing the concentration of NH 4 -N in the precipitate formed (Hao et al. 2013;Sun et al. 2020). NH 4 þ is removed from the solution only during the formation of struvite. It is not involved in any reaction, which competes for PO 3À 4 ; therefore, when a certain mass of the precipitate is first dissolved in HCl and then diluted with water, the purity of the struvite can be calculated by using the following equation: % purity of struvite ¼ n(NH 4 À N): (MW struvite ) mass of dissolved ppt Â 100 Journal of Water, Sanitation and Hygiene for Development Vol 11 No 5,708 where n(NH 4 -N) is the number of moles of dissolved struvite, which is determined by the titrimetric method, and MW struvite is the molecular weight of the struvite (Clesceri et al. 1999). Also, the quantitative X-ray diffraction (QXRD) technique has been used to determine the purity of the precipitated struvite. Lu et al. (2016) used QXRD to measure elemental concentrations of N, Mg, and P in the precipitate formed. This elemental information from QXRD is compared with elemental concentrations of N, Mg, and P using ICP. The concentration of N obtained from QXRD and ICP was very consistent and it was concluded that the QXRD technique is an effective method to determine MgNH 4 PO 4 ·6H 2 O content in struvite precipitate. The elemental analysis showed that N in the precipitates produced was primarily from the struvite phase, while Mg and P in the recovered precipitates were from both crystalline (i.e. struvite) and amorphous phases.
Furthermore, by comparing the % mass of Mg, NH 4 -N, and PO 4 -P in struvite precipitate with theoretical values of Mg, NH 4 -N, and PO 4 -P in pure struvite, the purity of the struvite can be estimated. The % mass of Mg, NH 4 -N, and PO 4 -P in pure struvite is 9.9, 5.7, and 12.6, respectively. The % composition of impurities (Ca, Fe, and CO 2À 3 ) in the supernatant can then be measured to ascertain how much of the impurities remain in the precipitate (Huang et al. 2006;Kemacheevakul et al. 2014).
In addition, energy dispersive spectrometry (EDS) has been used to identify the elemental composition of precipitates to measure the purity of struvite (Le et al. 2005). For samples with 0% Ca, the graph showed Mg, P, and O as the highest peaks. This is expected because these are the main elements in struvite in terms of percentage mass. At the Mg/Ca ratio of  1:1 and above, EDS spectrum reveals the formation of amorphous calcium phosphate compounds covering the struvite crystals.
The size attained by struvite crystal during product precipitation depends on a two-step process: nucleation and crystal growth. The growth of the crystal is affected by a combination of several factors such as the pH (Matynia et al. 2006), supersaturation, mixing turbulence (Matynia et al. 2006), and presence of foreign ions (Le et al. 2005;Liu & Wang 2019;Moragaspitiya et al. 2019;Hutnik et al. 2020). Le et al. (2005) found that at higher concentrations of Ca (with regard to Mg) in the solution, the mean size of the struvite is decreased. In their experiment to determine the impact of Ca on struvite size, they found that at the Mg/Ca ratio of 1:1 and above, the struvite formed was restricted and crystals were further inhibited by the production of amorphous calcium phosphate. Reaction media without the presence of Ca showed the highest percentage mass of struvite elements (Mg, P, and O) signifying the purity of the element (Le et al. 2005). This has been confirmed in a recent study by Moragaspitiya et al. (2019) that the absence of Ca in media generally resulted in the increase of crystal size with increasing NH þ 4 concentration. Jones (2002) attributed the slow growth rate of crystals by Ca in solution to the blocking off sites where crystals could form, resulting in the inhibition in the growth of struvite crystals. The size of the struvite crystal is significant to ease the recovery of the precipitate. This is because the finer struvite crystals, the more easily they can be discharged along with the effluent, resulting in a loss of crystals. Larger particle size has more benefits than smaller particle size; the larger the particle size, the easier it can be separated, the higher the P recovery efficiency, the lower their solubility, and they are less prone to be washed out (Ronteltap et al. 2010). This can be confirmed by Wei et al. (2018) in their study on phosphorus recovery from the urine of pregnant women. In that study, 94% of phosphorus was precipitated as struvite, but only 55% was removed from solution. Rahman et al. (2011) showed that the addition of seed materials did not significantly affect the crystal size of the struvite formed. Nevertheless, Liu et al. (2014) pointed out that adding seed materials significantly affect the crystal size of the struvite formed with the crystal size increasing from 33.7 to 57.0 μm with multi-stage addition of zeolite, whereas it was 44.7 μm with one-stage addition of zeolite. Therefore, the multi-stage addition of zeolite gave the highest effect on crystal growth. In other words, seed materials should be added periodically during a continuously running process.

Morphology
A scanning electron microscope (SEM) has been used to examine the images of the morphology of struvite precipitates. Tansel & Monje (2018) has reported the presence of impurities, other ions, and changes in pH as factors that can change crystal growth patterns which will result in irregularities in crystal morphology. The shape of struvite products is rod-like in the absence (Figures 5(a), 6(a) and 7(a)) of Ca (Le et al. 2005;Moragaspitiya et al. 2019). It has been described as needle-like (Figure 4(c)) also in the absence of Ca (Le et al. 2007;Doino et al. 2011;Muster et al. 2013;Muryanto et al. 2017). In experiments where Ca ion is present at different Mg/Ca ratios, the morphology has been described as arrow-head (Figure 4(d)) (Zhao et al. 2019), prismatic (Bayuseno & Schmahl 2020), pyramidal, star, rectangular platelet, elongated platelet, X-shape (Figure 4(b)), or coffin-lid shape (Figure 4(a)) (Prywer & Torzewska 2012).
As the Mg/Ca decreases, the needle-like shape of the struvite associated at the Mg/Ca ratio of 1:0 begins to become shorter ( Figure 5(b)), irregular (Figures 5(b) and 7(b)), and thicker ( Figures 6 and 7(c)) with increasing Ca concentration. This has been attributed to the interference of Ca in the struvite formation, where Ca reacts with phosphates to form undesirable calcium phosphates (Le et al. 2009).
This was supported by Moragaspitiya et al. (2019) in a recent study on the impact of Ca on struvite crystallisation. Liu & Wang (2019) found that at the Mg/Ca ratio of 1:1 and above, amorphous calcium phosphate compounds precipitate and cover the struvite crystals. Abbona & Franchini-Angela (1990) have explained that due to the lower solubility of calcium phosphates, it nucleates more easily and as a result precipitates faster and easier than magnesium phosphates. At the Mg/Ca ratio of 1:1 and above, energy dispersive spectroscopy (EDS) mapping reveals the formation of an amorphous Ca layer on the struvite crystal (Le et al. 2005). The presence of Ca during struvite precipitation does not affect phosphorus recovery but greatly affect the purity of the struvite.

METHODS OF MITIGATING THE EFFECTS OF CALCIUM
As a means of mitigating the negative impact of Ca, the following measures have been identified in the literature.
1. Increasing the concentration of ammoniacal nitrogen  in the solution (Moragaspitiya et al. 2019;Shaddel et al. 2020). Struvite purity has shown a significant increase when the concentration of NH 4 -N is increased. Moragaspitiya et al. (2019) showed that struvite purity can increase by 5.2% when the concentration of NH 4 -N is doubled. In their study, a synthetic solution was prepared to mimic wastewater effluent. In their results, when Mg/Ca is 1:1, the % weight of the struvite produced was 68.11, 73.31, and 76.75 at NH 4 -N concentrations of 328, 700, and 1,000 mg/L, respectively. Crutchik & Garrido (2011) established that when NH 4 -N is high, factors that favour struvite precipitation over Ca  Journal of Water, Sanitation and Hygiene for Development Vol 11 No 5,711 phosphate precipitation like buffer capacity of the solution and nucleation rate of the solution are favoured. Moragaspitiya et al. (2019) also confirmed that NH 4 -N acts as a pH buffer by preventing the reduction of pH during struvite reactions. 2. Shaddel et al. (2020) studied the effect of a high Mg/P ratio to mitigate the impact of Ca on struvite quality. When Mg/P is kept high (1.67:1, 2.3:1, and 3.1:1) at high pH, the concentration of Ca is kept constantly under 1% in the precipitate produced. 3. Ye et al. (2011) have investigated pH control as a mitigating factor and plotted the percentage amorphous calcium phosphate (% ACP) and the ratio of saturation index of ACP to saturation index of struvite (SI ACP /SI MAP ) against pH, as shown in Figure 8. They established that the presence of Ca increased the SI ACP and as a result the % ACP. When the pH is kept at 9.5 and below, the % ACP was kept at an extremely low concentration. The pH adjustment was done along with keeping the Mg concentration in seawater (the Mg source used for the study) at 1% and above. 4. The mitigating factor of a high Mg/Ca ratio has been suggested by many investigators (Liu et al. 2013;Li et al. 2019aLi et al. , 2019bMoragaspitiya et al. 2019;Shaddel et al. 2020). Adjusting the Mg/Ca ratio during struvite precipitation has been widely used to mitigate the negative effect of Ca.
Struvite purity falls below 60% (Figure 9) when the ratio of Ca/Mg is 1 but can be as high as 90% if the Ca/Mg ratio is 0.5 or less. This was confirmed by Kemacheevakul et al. (2014) in their study on the improvement of purity in struvite recovered from hydrolysed urine. This is very important if struvite precipitation is to be an economically viable venture. To increase the purity means adding more Mg to the process to compensate for the presence of Ca but this may not be an economically prudent venture. This means that a very high concentration of calcium in wastewater makes struvite precipitation unfeasible, as it lowers the purity significantly. Struvite purification with regard to calcium removal has not received the attention given to the economic value of the product.

NUTRIENT REMOVAL EFFICIENCIES USING DIFFERENT MAGNESIUM SOURCES
Generally, the recovery of PO 4 -P is calculated by using the following equation: The presence of Ca 2þ does not affect the percentage recovery of nutrients from wastewater. Zhang et al. (2015) observed that high phosphorus recovery reached a maximum of 95% (Figure 11) at the Mg/Ca molar ratio of 7:1. It was further Journal of Water, Sanitation and Hygiene for Development Vol 11 No 5,712 observed that the struvite content of the precipitate formed was only 17%. However, pH, Mg/P molar ratio, and reaction time are the significant factors that affect PO 4 -P recovery.
PO 4 -P recovery can occur between the pH of 7-and 11 but lower % PO 4 -P are removed at pH lower than 8 as can be seen in Figure 10. Mijangos et al. (2004) has been corroborated by Pastor et al. (2008) in an experiment conducted to determine the conditions influencing struvite precipitate. They observed that when pH was gradually increased from 8.2 to 9.5, at almost the same reaction time, % PO 4 -P recovery changed from 37.3 to 76.1%, respectively. Huang et al. (2017) attributed this reason to Journal of Water, Sanitation and Hygiene for Development Vol 11 No 5,713 orthophosphate speciation at different pH. At low pH of 8, there is a lower quantity (87%) of HPO 2À 4 in the total orthophosphate, while the proportion of HPO 2À 4 increases significantly (99%) at pH 9. This is likely to be the reason why PO 4 -P recovery as struvite is high at high pH values. Interestingly, Zhang et al. (2015) in their study on phosphorus recovery from bioeffluents of wastewater treatment plant (WWTP) observed that a rather huge drop in phosphorus recovery occurred at pH beyond 11.   Journal of Water, Sanitation and Hygiene for Development Vol 11 No 5,714 The molar ratio (specifically Mg/P molar ratio) of participating ions during struvite precipitation is very important because it affects the level of the solution. Recovery of total phosphorus (TP) is very high at a high Mg/P molar ratio (Table 2). Ye et al. (2016) observed that TP recovery increases significantly when the Mg/P ratio is 1.8 and above. Korchef et al. (2011) explained that an increase in Mg concentration in solution decreases the adhesion of crystals on the walls of the reactor vessel which result in high PO 4 -P recovery.
From Table 2, it can be seen that high-grade Mg sources (e.g. MgCl 2 and MgSO 4 ) can be replaced by natural Mg-rich sources in the production of struvite. However, strategies must be adopted to improve reactivity and removal of impurity  ions (Ca 2þ ) from both wastewater and Mg sources that have the potential to hinder struvite quality. It can be observed that this reduces the crystal size of the struvite product.

CONCLUSION
The literature has shown that the presence of Ca ions in solution during struvite precipitation negatively affects struvite quality, in terms of purity, size, and morphology. All papers reviewed revealed that at higher Ca concentrations, the crystals formed had reduced size and contained amorphous Ca phosphate substances rather than crystalline struvite. Even though Ca negatively affected struvite quality, its presence in solution did not negatively affect PO 4 -P removal from wastewater irrespective of the type of Mg source used. PO 4 -P recovery was rather affected by pH, Mg/P molar ratio, and reaction time. The high percentage recovery of P is always achieved at a higher Mg/P molar ratio. Even though the effect of Ca in solution during struvite precipitation has been studied, gaps still exist in the area. Research must be carried out on the removal of only Ca from Mg-rich sources to improve struvite purity, size, and thus its use as a fertiliser. The selective removal of Ca before struvite precipitation is a research gap that must be explored. Also, struvite precipitation from low phosphorus-rich sources should be looked at, since most research is focused on phosphorus-rich wastewater.

DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.