3.1. Influence of the Main Operating Conditions
As aforementioned, the struvite precipitation depends on several operating variables, but pH, initial concentration of P, Mg/P molar ratio, and temperature were considered the main ones.
According to the literature, the concentration (and speciation) of Mg
2+, NH
4+, and PO
43− is controlled by pH. Based on speciation curves in pure water, the optimum pH range for promoting struvite formation is between 8 and 11, as the concentrations of NH
4+ and PO
43− are the highest possible [
17]. In addition, the pH also affects the formation of other precipitates relevant to the system under analysis, such as MgHPO
4·3H
2O [
17]. According to the literature and the experiments conducted in the present study (
Figure S1, Supplementary Materials), the highest removal efficiency (near 80%) was observed at a pH of 9. Therefore, this pH value was set to carry out the following tests [
30,
31].
The P concentration in wastewater can vary within a wide range of values and, along with the Mg/P molar ratio, is a variable that influences the P removal efficiency by struvite precipitation. Thus, it is important to investigate the minimum initial P and Mg concentrations required to promote an efficient process (removal efficiency > 60% [
12]). These assays were carried out with synthetic solutions, and according to
Figure 2, negligible P recoveries were observed for concentrations below 30 mg P/L. Indeed, only 4% of P was recovered from a solution with an initial concentration of 20 mg P/L. Additionally, in the Visual MINTEQ prediction at the same pH and temperature conditions, the Saturation Index (SI) of struvite was negative (−0.190) for 20 mg P/L, showing no potential formation of this mineral. However, for 30 mg P/L the SI was already positive (0.249). Moreover, through Visual MINTEQ simulations, various minerals in the analyzed system may be formed, but only those with the highest potential for precipitation are summarized in
Table 2. In fact, some minerals are likely to precipitate in specific conditions since they are associated with SI > 0, whereas those that remain dissolved in solution have a negative SI. Thus, there was a decrease in SI for all species (Mg
3(PO
4)
2, MgHPO
4·3H
2O, and struvite) as the initial concentration of P decreased. Thus, this fact supports the results that RP decreased as the initial concentration of P in the solution also decreased.
Figure 3 shows the results for different Mg/P molar ratios, where the removal efficiency is superior for the highest value, as expected. For Mg/P 2, the amount of Mg is stoichiometrically higher than required to form struvite, promoting the P recovery by other complexes. For example, Mg
3(PO
4)
2 has an inferior solubility product constant (10
−25.2 [
32]) compared to struvite (with the most reported values between 10
−12.6 and 10
−13.26 [
17]), showing the potential to co-precipitate and form impurities in the final product (struvite). Indeed, other authors report the negative impact of higher Mg/P molar ratios in struvite purity [
30].
As shown in
Figure 3, the time for recovery of the same amount of P is reduced for higher Mg/P molar ratios. After 60 min of reaction, it was possible to remove about 70 and 80% of the P present in the solution for Mg/P 1 and 2, respectively. Even though a wide range of values was analyzed in this study, wastewater composition can be quite variable; thus, the Mg/P can also vary significantly. These results led to the conclusion that, with an Mg/P above 1, the process is potentially viable (removal efficiency > 60%). Considering the importance of obtaining the purest struvite product possible (better fertilizer behavior and market selling cost), a compromise between the removal efficiency and the purity of the material should be made. Thus, considering the predictions of Visual MINTEQ in
Table 3, Mg/P 1 provided a better response for both parameters.
When considering the real application scenario of struvite recovery after anaerobic digestion in the WWTP, which normally operates between 30 and 40 °C, it is important to verify whether temperature influences the formation of struvite.
Figure 4 and
Table 4 show that the removal efficiencies and final concentrations of P at 60 min of reaction were similar at all three temperatures. At 40 °C, the kinetic constant is superior (0.336 min
−1), indicating that the reaction occurs more rapidly, but the final R
Ef,P and C
eq,P are statistically equal to the other temperatures. Therefore, it can be concluded that implementing the phosphorus recovery method using struvite after anaerobic digestion is favorable, as the temperature does not inhibit its removal.
Based on the previous results, the ideal conditions to carry out the following tests were chosen: an initial concentration of 100 mg P/L, Mg/P 1, and a temperature of 20 °C. The XRD analysis showed that the solid precipitated in these conditions has a principal crystalline phase of struvite (92%wt), and a second phase of MgHPO
4·3H
2O (8%wt) (
Figure S3, Supplementary Materials).
3.2. Influence of Coexisting Ions
The effects of the two ions commonly present in wastewater (calcium and sodium) were studied, as shown in
Figure 5 and
Figure 6. Calcium has a significant influence on the initiation of the P removal reaction. Indeed, the P removal efficiencies, R
Ef,P, for the different Ca concentrations tested, in the first 5 min of the reaction, were statistically different from that of the blank solution (0 mg Ca/L) (
Figure 5b, bars marked with different letters). The kinetic constant increased from 0.0791 min
−1 to 0.1074 and 0.1699 min
−1 (
Figure S4, Supplementary Materials) for 0, 50, and 100 mg Ca/L, respectively, indicating that the removal of P was faster as the Ca concentration increased. However, after 60 min of reaction, the final removal efficiencies were similar for 0, 50, and 100 mg Ca/L, with values of 69, 73, and 77%, respectively. Nonetheless, in the case of adding 200 mg Ca/L, not only was the reaction the fastest (kinetic constant of 0.7355 min
−1;
Figure S4—Supplementary Materials), but the R
Ef,P was also greater than 80% in the first 5 min. In this case, the removal efficiency at 60 min increased by more than 20% compared to the blank (0 mg Ca/L).
According to the Visual MINTEQ predictions (
Table 5), as the concentration of Ca increased from 0 to 200 mg/L, the SI of struvite decreased, while the SI of hydroxyapatite and the other minerals increased. Other researchers, who reported that a higher concentration of Ca in the system enhances the formation of hydroxyapatite, corroborate the results of the present study. Struvite and hydroxyapatite are likely to form when the Mg/Ca molar ratio is 2. In contrast, when the Mg/Ca molar ratio is 0.5, only hydroxyapatite is expected to form [
18]. Indeed, XRD analysis of the precipitates formed at 100 revealed that the solid formed contained struvite as the only crystalline phase (
Figure S5, Supplementary Materials). On the other hand, the precipitate formed at 200 mg Ca/L was amorphous, suggesting the possibility of the formation of non-crystalline calcium phosphates. Similar results were found in a study developed by Campos et al. (2023), where the XRD patterns of solids obtained in the presence of calcium reveal low-intensity peaks of struvite and the possible formation of amorphous calcium phosphate as a coprecipitated material [
33]. Another study detected two phosphorus crystalline phases in synthetic solutions containing Mg, P, N, and Ca: struvite and hydroxyapatite [
34]. These authors concluded that the intensity of struvite peaks decreased as the Mg/Ca molar ratio decreases. The sample precipitated in a solution with a Ca content twice as high as Mg proved to be quite amorphous, with only hydroxyapatite being identified in the sample, indicating complete inhibition of struvite precipitation [
34]. New alternatives should be explored in further work to avoid the “contamination” of the sample with calcium, and the literature suggests an increase in the Mg/Ca molar ratio or selective calcium removal [
35,
36].
Regarding the presence of Na in the solutions,
Figure 6a,b demonstrates that, from 100 to 500 mg Na/L, the impact on the P removal efficiency was not statistically relevant (
p < 0.05). The same is not true for the concentration of 10,700 mg Na/L, where a significant inhibition of the P removal was observed (R
Ef,P ≤ 20%). Therefore, it is concluded that replacing the Mg source with seawater is not viable, since Na in high concentrations makes P removal difficult. In the literature, there are some studies where seawater is suggested as a source of Mg [
24,
37,
38]. However, other authors stated that seawater tends to form several complexes due to the presence of large quantities of Na
+ and Ca
2+ ions, thus reducing struvite purity compared to tests with MgCl
2 [
39]. At a concentration of 10,700 mg Na/L,
Table 5 shows that there was a significant reduction in the SI of struvite and Mg
3(PO
4)
2, which supports the lower P removal efficiencies.
The literature corroborates these results, indicating that Na
+ concentrations above 1150 mg/L increase the induction time for crystal formation. Despite the increase in supersaturation of the solution, the induction time may rise due to the accumulation of positive charges of Na
+ around molecules with negative charges, such as struvite, hindering the nucleation process [
40]. In addition, Kabdasli et al. (2018) explained that this increase in Na
+ charges on the surface of the nuclei forms a barrier that slows down the transport of Mg
2+ and NH
4+ to them [
41].
3.3. Influence of Seeding
Two types of seeds were tested to evaluate the influence of this phenomenon, struvite (inducing secondary nucleation) and biomass ash (primary nucleation).
Figure 7 and
Table 6 highlight the results obtained. The present results refer to the first 10 min because it is in the initial reaction period that the seeds can act to accelerate the process.
As can be seen in
Figure 7 and
Table 6, the kinetic constants are similar for all the experiments without and with seeds of different types and sizes. The P removal efficiencies after 10 min were statistically equal between the experiments without and with seeds of struvite (
p < 0.05). Although seeding is expected to boost nucleation efficacy and reduce induction time by increasing reaction surface area [
25], other studies also found no evident differences when seeding with struvite compared to the identical reaction without seeding [
27]. The reason for this could be that the reaction reaches equilibrium quickly, and the newly created crystal nuclei have a larger surface area than the seeds introduced into the system [
27]. Regardless, the use of struvite seeds presents an advantage in that the surface for nucleation is the same as that which will be formed, resulting in a uniform product. In addition, the use of struvite minimizes the contamination of the system with other impurities.
Using biomass ash, seeds smaller than 63 µm had the maximum removal effectiveness (almost 70%). This enhancement could be attributed to the fact that the surface area of the seeds is similar to the surface of the new struvite nuclei generated, promoting the development of this mineral on the surface of the seeds and allowing very small nuclei to form on these surfaces without being lost [
1]. Another crucial factor verified with the biomass ash seeds was an increase in the initial pH of the solution from 4.6 to 7.2. Despite these tests being developed in synthetic solutions, this increase in pH allows for a reduction in the use of a pH control solution (e.g., NaOH), reducing the costs in real-scale processes. Other studies in the literature use alternative sources for seeding such as biochar. A study from the literature carried out experiments with two types of biochar, obtaining a 43% increase in crystal size when compared to the process without seeding. These authors also reported that the crystal formed with biochar seeding had an internal crystalline structure similar to the struvite produced without seeding. However, there was also an increase in heavy metals concentrations, which is a disadvantage to this product for applications in the soil. In addition, the study showed that there is no beneficial effect in increasing the amount of seed from 0.75 to 1 g/L [
26]. A study conducted by other authors also disclosed an increase in P removal efficiency from 88 to 97% (biochar seed) and 95% (struvite seed) compared to the process without seeding [
42]. Thus, these alternative seeding sources could be an asset to enhance the process.
The precipitate obtained in the presence of biomass ash seeds (size < 63 µm) was investigated by XRD to determine the level of impurities in the final product. The principal crystalline phase determined was struvite (about 68%wt), while P was also removed as (Mg
3Ca
3PO
4)
4. The XRD also found quartz (11%wt) and calcium carbonate (18%wt) phases, which was expected given that biomass ash contains silica, carbonates, and calcium (
Figure S7, Supplementary Materials).
Figure 8 presents SEM images with the precipitate obtained without, and with, biomass seeds (size < 63 µm). There are some differences between the samples due to the presence of compounds with silica, calcium, and carbonates, as is shown in
Figure 8b,d. However, in both images, the presence of struvite (crystals with more elongated form) can be observed.
Out of the 22 crystals that were examined, and are shown
Figure 8a, the length found was between 9.5 and 48.4 µm, while four had larger diameters, from 39 to 48.4 µm. Regarding the crystals obtained with biomass ash seeds, the size range was between 6.6 and 40 µm.
A complementary mapping analysis was conducted.
Figure S8 (Supplementary Materials) shows that, across the elongated shape of the struvite crystal in both precipitate samples, other elements were identified. Mg, N, and P are primarily grouped in the crystal area (more intense color), confirming that it corresponds to struvite. The Ca appeared when biomass ash was used for seeding, although it was dispersed over the whole sampling area. This suggests that the Ca mineral surrounding the struvite crystal may be related to calcium carbonate and/or (Mg
3Ca
3PO
4)
4.
3.4. Precipitation with Wastewater
As highlighted in the results shown in
Table 7, the P removal efficiencies surpassed the expected and predicted values in the analyses of the synthetic solutions. Furthermore, the recoveries obtained were not significantly influenced by the point in WWTP at which the samples were collected. Removal efficiencies of 95% [
22] and 93.2% [
15] were also found in the literature for Mg/P molar ratios of 1 and 2.5 in wastewater matrices. According to the XRD analyses of sample S2.2, struvite represented only 15%wt of the solid, while hazenite (KNaMg
2(PO
4)
2·14H
2O) was about 23%wt. In this case, the main crystalline phase detected was halite (NaCl), representing about 62% of the solid (
Figure S9, Supplementary Materials). This result may be due to the high concentration of sodium and chloride present in the sample, as the WWTP is located near the sea. In these cases, the precipitate of struvite should be previously washed to avoid contamination before its application in soil.
The results in the present study showed that P removal was mainly due to the formation of two minerals, hazenite, and struvite (in lower mass percentage). In the literature, several studies report two compounds derived from struvite, struvite-Na (NaMgPO
4·7H
2O) and struvite-K (KMgPO
4·6H
2O). The formation of these compounds tends to occur when sodium and potassium are in excess in the matrix; in both cases, the ions Na
+ or K
+ substitute the NH
4+ [
43,
44]. Furthermore, the solubility product of struvite is between 10
−12.6 and 10
−13.26 [
9], that of struvite-K is 10
−10.62 [
45] and that of struvite-Na is 10
−11.6 [
46]. Another study addressing hazenite formation stated that this mineral is similar to struvite. The study demonstrated that P composition in hazenite is about 11.2%, while that in struvite is 12.6% [
47]. Therefore, hazenite may have fertilization interest, and like struvite, the compound may have low solubility in water, acting as a slow-release fertilizer.
According to the Visual MINTEQ predictions, the mineral with the highest probability of precipitating in wastewater conditions was hydroxyapatite, with an SI of 16.970, compared to that of 1.408 for struvite, mainly due to the calcium concentration in the matrix. However, no calcium mineral was detected in the XRD analyses, which indicates that calcium probably precipitated in the amorphous phases. Regarding struvite-K, and struvite-Na, or the combination of both (hazenite), there were also no predictions for their formation. These differences between the real and synthetic solutions result from the fact that the former effluent contains many elements and organic matter that determine the complex equilibrium between the phases that precipitate and the substances that remain in the solution.
SEM analysis was carried out for the precipitate from the WW2.2 sample.
Figure 9a,c shows the images obtained for the synthetic solution, while
Figure 9b,d shows the images for the real sample. Differences are easily detected, as the images in
Figure 9a,c depict well-defined struvite crystals, whereas
Figure 9b,c shows not only struvite crystals (corresponding to more elongated structures) but also other clusters of particles nearby the crystals, making their identification difficult. Some of the clusters next to the struvite are probably halite, as it is the compound that represents most of the solid precipitate sample.