Abstract

Phosphate (P) recovery from urban wastewaters is an effective strategy to address environmental protection and resource conservation, aiming at an effective circular economy. Off-grid wastewater treatment systems like urine-diverting toilets (UDT) can contribute to source separation towards nutrient recovery, namely phosphorus recovery. Effectiveness of P precipitation requires a process-based knowledge regarding pH, Mg:PO4, contact time and their interactions in P recovery and crystal morphology. Several studies failed to see the process as a whole and how factors influence both morphology and P recovery for UDT hydrolysed urine. This study addressed the above-mentioned factors and their interactions, and results showed that pH and Mg:PO4 ratio are the key factors for struvite precipitation, whereas contact time is relevant for crystal growth. The recommended set of factors proposed (pH 8.5, Mg:PO4 ratio of 1.2:1 and 30 minutes contact time) not only promotes a high precipitation yield – 99% of P with co-precipitation of at least 21% of ammonium (NH4+) – but also leads to larger crystals with lower water solubility (10% less crystals dissolved in water after 3 days). The obtained outcome facilitates the downstream process and leads to a more efficient slow-release fertiliser, as less P is wasted to receiving waters by leaching, minimising eutrophication processes.

INTRODUCTION

Approximately 60–65% of the phosphorus and 90% of nitrogen entering human systems is excreted in urine and discharged into sewage systems (Simha & Ganesapillai 2017). Due to the lack of efficiency of most nutrient removal processes applied in wastewater treatment plants, most of the nutrients end up accumulating in water bodies, triggering eutrophication phenomena in rivers, lakes and coastal zones (Martins et al. 2014). In this framework, conventional wastewater management concepts are being highly challenged because urine, despite representing only 1% of municipal wastewaters, contributes ∼75% and ∼45% of nitrogen and total phosphorus loads, respectively. Therefore, decentralised processes have been proposed as the new wastewater treatment paradigm and, amongst the available on-site technologies, urine-diverting toilets seem to be the most effective (Mendes et al. 2019). Indeed, in applying source separation of urine and faeces under proper process conditions, phosphorus compounds can be extracted from urine as struvite (NH4MgPO4·6H2O) (Simha & Ganesapillai 2017). Furthermore, struvite precipitation allows NH4+ recovery, adding up to an environmental advantage and an economic benefit regarding its use as fertiliser (Malila et al. 2019). The driver behind phosphorus recovery efforts is the asymmetrical geological distribution of phosphate rock, namely in Europe, where it is almost non-existent, as well as the negative environmental impacts from mining operations (Edahwati et al. 2016).

Controlled struvite precipitation can be achieved through the addition of Mg2+ salts (often in much lower concentration compared to NH4+ and PO43−) under tailored experimental conditions, i.e. pH, sedimentation time, contact time and ammonium content (Edahwati et al. 2016). Concerning the key factors affecting precipitation, two points gather consensus; the first is the pH range between 7 and 11.0 (Nelson et al. 2003), and the second is the ratio between magnesium to phosphate ions, hereby represented as Mg:PO4, which must be higher than 1:1 (Ezquerro 2010). MgCl2 or MgSO4 is usually used as magnesium source, but the optimal process conditions are highly dependent on the composition of the solution matrix being studied. Actually, while for wastewaters with high ammonium content the ideal values reported in literature are pH 9.2 and Mg:PO4 1.7:1 (Ezquerro 2010), for urine, the ideal values are pH 8.5, Mg:PO4 1.4:1 and 30 minutes of contact time (Xu et al. 2015), and for a synthetic hydrolysed urine solution, the ideal values are pH 9.5, Mg:PO4 1.6:1 and 20 minutes (Shalaby & El-rafie 2015). Furthermore, struvite recovery yield, crystal size, morphology and chemical composition must be optimised, in order to envisage downstream recovery at an industrial scale. For instance, Wei et al. (2018) reported a struvite precipitation process with 94% efficiency but with only 55% of crystals recovered due to the washout of small crystals. Barbosa et al. (2016) evaluated P recovery under constant pH and assessed the effect of the remaining experimental parameters on crystal size. The influence of pH and temperature on crystal size was also addressed by Zhang et al. (2016) but without much focus on P recovery.

Despite the above-mentioned studies, there is still a lack of guidelines on the optimisation of struvite precipitation load and appropriate crystal size and properties for downstream recovery. Many studies fail to address the process from a holistic perspective, mainly focusing on struvite formation or struvite precipitation, giving less importance to the morphology and properties of crystals and how they affect struvite quality. Furthermore, the applicability of struvite precipitates as fertilisers should also be taken into account. If struvite is easily soluble, plant uptake will be minimised, and washout to receiving water bodies will be more significant, increasing eutrophication problems. Therefore, it is also important to study the behaviour of the precipitated material in contact with water.

The goal of this research was to contribute to an enhanced P precipitation process and downstream recovery from hydrolysed urine (from urine-diverting toilets). Hence, firstly, the best combination of key operational factors, namely pH, contact time and Mg:PO4 ratio, was assessed using hydrolysed urine as a proxy of diverting toilets. Secondly, this knowledge was complemented with struvite precipitate and crystal formation characterisation in order to advance downstream processing, product recovery and effective fertiliser application.

MATERIALS AND METHODS

A design of experiments (DoE) approach using Minitab 18 software was used to optimise pH, Mg:PO4 ratio and contact time conditions. The DoE full factorial analysis was carried out with four pH levels (8.0; 8.5; 9.0 and 9.5), two Mg:PO4 levels (1:1 and 1.2:1) and two levels for contact time (20 and 30 minutes). The order of runs was randomised and repeated three times to get a total of three replicates.

Chemicals

Analytical grade chemicals (KH2PO4, NH4Cl, MgCl2.6H2O and NaCl from Panreac, CaCl2 and KCl from Chem-lab and Na2SO4 from Labkem) were used to prepare a solution with concentrations of PO43− and NH4+ close to the content of a urine solution after hydrolysation of urea to ammonium, following the protocol reported in Shalaby & El-rafie (2015). Table 1 presents the concentration of the various components. The pH corrections were made using 1 M NaOH solution (prepared using analytical grade NaOH pellets from Panreac).

Table 1

Synthetic hydrolysed urine solution composition, adapted from Shalaby & El-rafie (2015)

ReagentsMolar mass (g/mol)Concentration (mol/dm3)
KH2PO4 136.086 0.021 
NH4Cl 53.492 0.090 
MgCl2.6H2203.303 0.003 
KCl 74.551 0.054 
Na2SO4 142.042 0.016 
NaCl 58.443 0.079 
CaCl2 110.990 0.003 
ReagentsMolar mass (g/mol)Concentration (mol/dm3)
KH2PO4 136.086 0.021 
NH4Cl 53.492 0.090 
MgCl2.6H2203.303 0.003 
KCl 74.551 0.054 
Na2SO4 142.042 0.016 
NaCl 58.443 0.079 
CaCl2 110.990 0.003 

The experiments were conducted as follows: solid MgCl2.6H2O was added to 0.3 dm3 of synthetic urine solution in accordance with the pre-defined Mg:PO4 proportion, and the pH was adjusted to the pre-selected value (measured with an Orion 410A+). Afterwards, the solution was stirred (400 rpm) during the selected reaction time. After reaching the desired reaction time the solution was kept still for 30 minutes for the particles to settle; then the solids were recovered by filtration using quantitative paper, with pore size 7–9 μm, from Filter-Lab. The filters with precipitate were dried for 2 days in a Quincy-Lab oven. Phosphate analysis on the filtrate was made by the vanadomolybdate method according to standard methods (Rice et al. 2012), using a Hitachi U-2000 spectrophotometer.

SEM and EDX

The particles size and morphology were analysed by scanning electron microscopy (SEM) using a Hitachi TM3030 tabletop microscope in a charge-up reduction mode which required no sample coating. Measurements on the SEM images were made with J-image software. For energy-dispersive X-ray spectroscopy (EDX) analysis, the samples were coated in a thin-film of gold/palladium to increase conductivity, and a JEOL JSM-70001F field emission SEM was used. Materials characterisation was complemented by X-ray powder diffraction (XRD). Diffractograms were obtained in a Pan'Analytical PW3050/60X'Pert PRO (θ/2θ) equipped with an X'Celerator detector and with automatic data acquisition (X'Pert Data Collector (v.2.0) software) using a CuKα radiation as incident beam, 40 kV-30 mA. Measurements were made by continuous scanning from 7° to 40°, with a step size of 0.017° 2θ and a time per step of 19.6 s.

Dissolution tests

For the assays that lead to a better yield and larger homogenous crystals, water solubility of the precipitate was assessed. For this, 100 mg of precipitate was dispersed in 100 cm3 of distilled water and kept under stirring at room temperature for 24, 48 and 72 h, and the solid was recovered by filtration and dried until constant weight.

Chemical equilibrium data and software platform

Modelling software, Visual MINTEQ 3.1, was used to calculate the solubility of solids, simulate equilibrium and speciation of inorganic solutes (Gadekar 2011). Visual MINTEQ works with ionic activities, so it is necessary to have the value of the ionic strength (I) which is calculated using Equation (1):
formula
(1)
where Ci is the concentration and Zi is the valence of the species present in solution.
For ionic solutions where 0.1 < I < 0.5, activity coefficient (γi) is usually calculated through the Davies equation (Equation (2)), in which A is the Debye–Hückel constant.
formula
(2)
Considering equilibrium values, the saturation index (SI) can be determined using Equation (3) (Tran et al. 2014), where IAP is the ionic activity product calculated by the product of ionic activities and ksp is the constant of the solubility product of the mineral. Solubility constants may differ between different matrixes, and in literature, ksp varies in the range from 4.4 × 10−14 to 3.9 × 10−10; this essay considered the ksp of struvite for urine at 25 °C as 10−13.26 (Ronteltap et al. 2007).
formula
(3)

The relevance of the SI parameter is a given and not only in the way it affects the crystals' properties and sizes, as it will also dictate if the solution is saturated or not towards a mineral, providing a prediction of how likely a crystal is to precipitate (Zhang et al. 2016).

RESULTS AND DISCUSSION

Effects and interactions of Mg:PO4, pH and contact time in P recovery maximisation

The effects and interaction of factors in P recovery were studied using a DoE to improve the understanding of formation and precipitation processes.

The data from DoE revealed that only two of the factors evaluated have significant main effects on P recovery, pH (F(3,32) = 50.697 p = 0.000) and Mg:PO4 (F(1,32) = 50.160, p = 0.000). Contact time was not significant (F(1,32) = 2.736, p = 0.108). In fact, while an increase in pH and Mg:PO4 leads to a higher recovery of P, increasing the contact time from 20 to 30 minutes leads to almost the same P recovery (difference is lower than 0.2%), not justifying higher contact time if more than 98% is already being recovered, as seen in Table A1 (in Supplementary Data), which summarises the data obtained in this study.

In the range of Mg:PO4 values considered (1:1 and 1.2:1), there is no influence of contact time on P recovery, which is always around 99%. However, as Mg:PO4 increases, there is a noticeable decrease in the amount of Mg2+ predicted to precipitate, i.e. around 88% for Mg:PO4 1:1 and 72–74% for Mg:PO4 1.2:1 (Table A1, Supplementary Data). The data reveals that not all P is reacting with Mg present, even when Mg:PO4 is 1:1, suggesting a probable reaction with calcium and/or potassium ions. This fact can be viewed as an advantage if struvite is to be used as fertiliser, because the presence of micro-nutrients such as K+ or Ca2+ enriches its value (Taddeo et al. 2018).

The pH effect is related to the conjugation of Reactions (1)–(3), which occur when pH is increased (Dhakal 2008).
formula
formula
formula
Increasing pH will lead to higher amount of HPO42− (Reaction (1)) and PO43− (Reaction (2)) in solution, with their relative quantities being dependent on the exact pH value. From the simulation data, it was predicted that over 58% of P in solution is as HPO42− at the pH levels 8.0 and 8.5. Only at pH 9.5 does the amount of PO43− approach 1% of total P, being approximately non-existent at lower pH values. However, increased pH will also lead to NH4+ deprotonation (Reaction (3)), promoting the volatilisation of NH3 out of solution and decreasing the concentration of NH4+ remaining to interact with phosphorus to form struvite. The literature shows a decrease in concentration from 99% to 64% while pH increases from 7.0 to 9.0 (Ezquerro 2010). The results, whereby HPO42− is more available than PO43−, also point towards the statement that mentions Reaction (4) as the more likely reaction leading to struvite formation over the direct reaction (Reaction (5)) (Liu et al. 2013).
formula
formula
Figure 1 presents the response surface of main effects and interaction of pH and Mg:PO4 for 20 minutes (maximum P recovery). Equation (4) describes the role of each variable, showing the second order tendency of pH factor and the small contributions of time as an interaction factor when compared to the pH/Mg:PO4 interaction. In the equation, the variable for Mg:PO4 needed to be recoded (1.2:1 = 2 and 1:1 = 1) to transform the variables from a categorical to numerical form so Minitab could consider them to appear in the equation:
formula
(4)
Figure 1

Response surface for Mg:PO4 ≥ 1:1 at 20 minutes.

Figure 1

Response surface for Mg:PO4 ≥ 1:1 at 20 minutes.

The interaction of factors pH and Mg:PO4 were revealed to be significant F(3,32) = 11.354, p = 0.000. On the other hand, the interactions between Mg:PO4 and contact time, F(1,32) = 0.146, p = 0.705, pH and contact time F(3,32) = 1.330, p = 0.282 or the three-way factor interaction, F(3,32) = 0.582, p = 0.631, were not significant. The Mg:PO4 ratio affects the pH value where P recovered gets its maximum – 99.0% at pH 9.0 for Mg:PO4 1.2:1 and 98.7% at pH 9.5 for Mg:PO4 of 1:1. It is also important to notice how small the difference of pH 9.0–8.5 is: generally <0.5%, which is a similar difference in P recovery within pH levels (∼1%) already present in the study done by Ronteltap et al. (2010). Figure 1 presents the interaction between pH and Mg:PO4 factors, as the best values for P recovery cannot be found following only Mg:PO4 or pH values. In fact, if contact time is fixed at 20 minutes, to find the best set of values to maximise P recovery, a combination of both pH and Mg:PO4 is needed. The increase of P recovery with increased Mg:PO4 is justified, because when the stoichiometry of struvite compounds is above 1:1, the loss of Mg due to side reactions forming small quantities of other magnesium salts is mitigated. To find the pH value to use along with the Mg:PO4, the second order of the curve must be followed to find the best pH value. The optimal set of values, leading to 99% of P recovery, were Mg:PO4 1.2:1, pH 8.5–9 and mixing time of 20 minutes. These values render the higher Mg:PO4 (1.6:1) ratios pointed to by other authors (Shalaby & El-rafie 2015) unnecessary and costly.

Precipitate characteristics

The process scale-up will require more than the maximisation of P precipitation, because composition and precipitate characteristics must be considered for downstream recovery and fertiliser use. This is especially important when, as was obtained in this study, P recovery differs by only around ≈1% between pH 8.5 and 9.5. As such, the resulting product of the overall process must yield the best combination of high P recovery and crystal morphology and properties to facilitate struvite recovery from the reactor and application on fields. SEM images presented in Figure 2 show that crystal morphology varies between elongated for higher pH and a mix of branched and elongated for lower pH values. Microphotographs in Figure 2(a)2(c) show how crystals decreased in size and became thinner as pH evolved from 8.5 to 9.5, keeping Mg:PO4 at 1.2:1 and 30 minutes of contact time.

Figure 2

SEM images of precipitates obtained at different values of pH.

Figure 2

SEM images of precipitates obtained at different values of pH.

Longer contact time leads to an increase in crystal mean length, leading to overall larger crystals. In general, the increase in SI, up to 8.0–8.5, leads to larger crystals; however, between 8.5 and 9.0 the tendency inverts, and an increase in SI leads to a decrease in the crystals' mean area. At Mg:PO4 1:1 in 30 minutes, pH 9.5 crystals showed slightly higher length than those of pH 9.0, but these were 4 μm wider than the 9.5 pH ones.

The distribution of crystal area with pH, for precipitates obtained with Mg:PO4 ratios of 1:1, is presented in the Supplementary Data (Figure A1 and A2), and for Mg:PO4 ratio of 1.2:1, it is presented in Figure 3 (20 minutes) and Figure 4 (30 minutes). SI values are displayed also in the Supplementary Data (Table A2).

Figure 3

Distribution of crystal area with pH for Mg:PO4 ratio of 1.2:1 and 20 minutes.

Figure 3

Distribution of crystal area with pH for Mg:PO4 ratio of 1.2:1 and 20 minutes.

Figure 4

Distribution of crystal area with pH for Mg:PO4 ratio of 1.2:1 and 30 minutes.

Figure 4

Distribution of crystal area with pH for Mg:PO4 ratio of 1.2:1 and 30 minutes.

The figures representing the distribution of the areas of the crystals with pH show some interesting features. At pH 8.0, both contact times show the widest range of crystal areas when compared to other pH values, which indicates the presence of large crystals alongside tiny ones (known as fines). This wide range of crystal areas is also indicated by the high value of the standard deviation of the mean crystal area. The results are in line with those stated by Su et al. (2014), who pointed to pH 9.0 as the pH presenting a smaller number of fines. In this study, the pH range between 8.5 and 9.0 is revealed to have lower fines overall than other pH levels. For pH 8.5, the distribution of crystals shifts to higher values of the crystal area for both contact times, confirming the mean value results as being the pH that originates larger crystals in the conditions studied. However, pH 9.0 shows the narrowest range for crystal area for both contact times, as seen by the smallest value of standard deviation. The decrease of the standard deviation value between pH 8.0 and 9.0 has already been reported by Ronteltap et al. (2010), who showed also that pH 8.5 has larger sized crystals on average when compared to pH 9.0. pH 9.5, in turn, always has the smallest crystals in general, with the curve shifting to smaller crystal areas. All of these match the data of SI. In this study, all SI values are positive and above 2.5, reflecting a supersaturated solution in respect to struvite. It can be seen that there is an SI value between 2.7 and 3.5 (between pH 8.0 and 9.0), above which crystal size drops with increasing SI, resulting in smaller and thinner crystals. This highly contradicts the conclusions of Shalaby & El-rafie (2015), who considered an SI value of 14.98 and pH of 10 as ideal for P recovery, with the increase in SI being obtained through the increase of ion concentration while keeping the Mg:PO4 ratio constant. As a matter of fact, they did not investigate the possible increase in P recovery with crystal properties. In this study, crystals formed at an SI closer to 4.0 are smaller and thinner than crystals forming at lower SI values. Also the crystals formed at pH 8.5 and lower SI values are larger than those obtained at higher values of both pH and SI. Crossing this information with the data obtained from the response surface, it is possible to conclude that, for the pH where P recovery has its absolute maximum, appropriate crystal size was being compromised.

From Figures 3 and 4, it is also important to highlight how the increase of contact time from 20 to 30 minutes, at all pH levels, leads to higher values of crystal areas. The increase in contact time also reduces the difference between the distribution curves of the crystal areas obtained at pH 8.5 and 9.0.

The present results are in agreement with the statement that larger crystals can be obtained with low pH values (Kozik et al. 2014), optimum SI value and a residence time long enough to promote agglomeration (Villa Gomez et al. 2015). Agglomeration is promoted when particles in suspension in the supersaturated solution, small enough (<1 μm) for van der Waals' forces to exceed the gravitational forces, collide with each other (Adnan 2003). However, in the literature there is no information about the maximum SI value. From the results of this study, for the hydrolysed urine matrix, the ideal value is in the range of pH 8.5–9.0, corresponding to SI values between 2.7 and 3.5. In literature it is denoted that a combination of high SI values and higher pH leads to smaller and thinner crystals, as high SI values promote nucleation over crystal growth (Kozik et al. 2014), leading to increased number of smaller crystals or undetectable nuclei in solution at the cost of larger crystals. However, the increase in contact time of particles with the supersaturated solution leads to the decrease of mean SI in solution, resulting in bigger crystals at 30 minutes than at 20 minutes, as was also visible in this study. The identification of pH and SI values, above which thinner and smaller crystals are generated, is of high importance. Small crystals (fines) make the retrieving process more difficult at a larger scale, and as the fines will become suspended in solution resulting in non-uniform operation (Adnan 2003), the process will be more expensive.

Figure 5 shows the stacked XRD diffractograms for Mg:PO4 1.2:1 with pH increase for 30 minutes contact time. The diffractogram identified the crystals as struvite, which was expected according to the studied conditions.

Figure 5

Stacked diffractograms at Mg:PO4 1.2:1, showing the comparison between the four pH levels fixing contact time at 30 minutes. S – struvite (reference taken from Hao et al. (2008)).

Figure 5

Stacked diffractograms at Mg:PO4 1.2:1, showing the comparison between the four pH levels fixing contact time at 30 minutes. S – struvite (reference taken from Hao et al. (2008)).

Figure 6 shows the EDX information for pH levels 8.5 and 9.0 at 20 minutes and the obtained 30 minutes mixing time for pH 8.5. This result further confirms the presence of struvite by showing both phosphate elements and Mg2+ as being the most abundant. However, EDX also points out the presence of K+ ions, which seem to be from K-struvite (which has a diffractogram similar to regular struvite (Chauhan et al. 2011)) and/or its precursor KH2PO4. K-struvite is known to be favoured by higher pH values than is regular struvite, which is in accordance with the obtained results, as more K+ is detected at pH 9.5 than at pH 8.5. As was expected, despite being in much lower concentration than magnesium (Mg2+/Ca2 ratio ≈ 9/1), Ca2+ ions are present as calcium phosphates, even if only as tiny crystals in amounts that make them almost unnoticeable, as confirmed by EDX (Figure A3 in Supplementary Data).

Figure 6

EDX results for precipitates obtained for Mg:PO4 1.2:1.

Figure 6

EDX results for precipitates obtained for Mg:PO4 1.2:1.

Crystal properties give useful information for process design. Combining the P recovery, composition and size of crystals, the ideal values were found to be situated at Mg:PO4 1.2:1, with 30 minutes contact time. From EDX it is also possible to estimate the minimum amount of NH4+ expected to be recovered together with struvite. For that, we use the recovered amount of P (99%) and deduce from the value the amount expected to precipitate as K-struvite. To calculate the amount of K-struvite by percentage, it is assumed that only two Mg2+ phosphates are present – struvite (MgNH4PO4.6H2O) and K-struvite (MgKPO4.6H2O), and then the ratio of Mg to K is obtained using the EDX data, and from that a maximum value of 11% of K-struvite expected to be formed was obtained. From this matrix and the conditions used, a minimum of ∼21% of NH4+ is expected to precipitate together with 99% of P.

When applying the precipitates to soil, the ones more suitable to be used as fertilisers are the less soluble ones, i.e. those able to remain in the soil longer before being leached and carried away by rain. In Figure 7, a chart of dissolved material with time is presented, showing clear difference between pH 9.5 and the other two pH levels. For pH levels 8.5 and 9.0, the values remain approximately the same throughout the experiment.

Figure 7

Evolution of dissolution of crystals, formed at different pH values, with time.

Figure 7

Evolution of dissolution of crystals, formed at different pH values, with time.

The difference in dissolved material between pH 8.5 and 9.0 is basically non-existent, but for pH 9.5, it is quite considerable, with 10% more remaining undissolved. This is relevant when considering applying, for example, 1,000 kg of crystals to the fields as P fertiliser, as 100 kg would not be leached, resulting in considerable savings in fertiliser in the long term. This difference in dissolution of the crystals may be explained by considering the sum of peak intensity and FWHM (full width at half maximum) for the most relevant peaks in the diffractograms – 2θ values of 15.9°, 21.0°, 21.6°, 27.2° and 32.1° (Figure 8).

Figure 8

Chart with peak intensity and FWHM sum for the most relevant struvite peaks.

Figure 8

Chart with peak intensity and FWHM sum for the most relevant struvite peaks.

The sum of peak intensities is higher for pH 9.0 than for pH 8.5. This occurrence reflects what was found with SEM microphotographs where crystals were revealed to be smaller and with a thinner needle-like shape morphology at pH 9.0 than at pH 8.5, having more crystals with smaller sizes reducing the count sum. The FWHM is an important parameter as it reflects the structure organisation and crystallinity; higher FWHM means less crystallinity, which fits the dissolution results. The difference in FWHM sum is similar to the difference in dissolved material between pH 9.0 and 8.5, as after the third day, the difference between dissolved material in these pH levels is almost non-existent. The diffractograms (Figure 5) also explain the different behaviour of the crystals formed at pH 8.5 and 9.0 compared to those formed at pH 9.5. While the crystals in the first two pH levels have larger and narrower width at half height, pH 9.5 has smaller peaks with broader width at half height. This reveals the latter one as less crystalline with a higher number of smaller crystals, and pH 8.5 and 9.0 as more crystalline with a higher number of larger crystals, which confirms the information taken from the distribution of crystal area. Lower crystallinity and smaller crystals with higher FWHM explain why pH 9.5 is the one showing higher dissolution material, making the precipitate less suitable to be used as fertiliser.

CONCLUSIONS

Aiming at the maximisation of P recovery, the interaction of Mg:PO4 and pH is highly significant, with crystal size and properties also being relevant for full scale application and P recovery and reuse. This study shows that focusing only on P recovery yield is not appropriate. Recovery efficiency showed a difference of 1% between pH 8.5 and 9.5; however, adding the data obtained from crystal properties, the best process conditions for larger crystal production are for pH 8.5–9.0, along with SI values of 3.0–3.5 for Mg:PO4 1.2:1, resulting in almost 200 μm2 of difference in crystal size from pH 8.5 to pH 8.0 or 9.5. Water dissolution tests have also shown that precipitates obtained at pH 8.5 and 9.0 are less soluble, ∼10% less, than precipitates obtained at pH 9.5, providing better use as slow-release fertilisers and minimising leaching, which is important for reducing the accumulation of P in receiving waters through water runoff. In summary, this study reveals that the best process conditions towards P recovery with better performance as agriculture fertiliser are Mg:PO4 1.2:1, pH 8.5 and 30 minutes of contact time in hydrolysed urine. These conditions, obtained at laboratory scale, reveal the possibility of recovering 99% of P, with a simultaneous recovery of at least 20.6% of NH4+, with crystals ranging between 850 and 950 μm2 and less water solubility.

SUPPLEMENTARY MATERIAL

The Supplementary Material for this paper is available online at http://dx.doi.org/10.2166/wst.2019.371.

REFERENCES

Adnan
A.
2003
Pilot-scale Study of Phosphorus Recovery Through Struvite Crystallization
.
Masters thesis
,
Department of Civil Engineering, University of British Columbia
,
Vancouver
,
Canada
.
Barbosa
S. G.
,
Peixoto
L.
,
Meulman
B.
,
Alves
M. M.
,
Pereira
M. A.
2016
A design of experiments to assess phosphorous removal and crystal properties in struvite precipitation of source separated urine using different Mg sources
.
Chemical Engineering Journal
298
,
146
153
.
doi:10.1016/j.cej.2016.03.148
.
Chauhan
C. K.
,
Vyas
P. M.
,
Joshi
M. J.
2011
Growth and characterization of Struvite-K crystals
.
Crystal Research and Technology
46
,
187
194
.
doi:10.1002/crat.201000587
.
Dhakal
S.
2008
A Laboratory Study of Struvite Precipitation for Phosphorus Removal From Concentrated Animal Feeding Operation Wastewater
.
Masters thesis
,
Department of Civil, Architectural and Environmental Engineering, University of Missouri-ROLLA
,
USA
.
Edahwati
L.
,
Sutiyono
S.
,
Perwitasari
D. S.
,
Muryanto
S.
,
Jamari
J.
,
Bayuseno
A. P.
2016
Effects of the optimised pH and molar ratio on struvite precipitation in aqueous system
.
MATEC Web of Conferences
58
,
01018
.
doi:10.1051/matecconf/20165801018
.
Ezquerro
A.
2010
Struvite Precipitation and Biological Dissolution
.
TRITA LWR Degree Project Department of Land and Water Resources Engineering, Royal Institute of Technology
,
Stockholm
,
Sweden
.
Gadekar
S. M.
2011
Process Development for Recovery of Nutrients as Struvite and Struvite Based Products
.
PhD thesis
,
University of Florida
,
Florida
,
USA
.
Hao
X. D.
,
Wang
C. C.
,
Lan
L.
,
Van Loosdrecht
M. C. M.
2008
Struvite formation, analytical methods and effects of pH and Ca2+
.
Water Science and Technology
58
(
8
),
1687
1692
.
doi:10.2166/wst.2008.557
.
Kozik
A.
,
Hutnik
N.
,
Piotrowski
K.
,
Matynia
A.
2014
Continuous reaction crystallization of struvite from diluted aqueous solution of phosphate(V) ions in the presence of magnesium ions excess
.
Chemical Engineering Research and Design
92
,
481
490
.
doi:10.1016/j.cherd.2013.08.032
.
Liu
Y.
,
Kumar
S.
,
Kwag
J. H.
,
Ra
C.
2013
Magnesium ammonium phosphate formation, recovery and its application as valuable resources: a review
.
Journal of Chemical Technology and Biotechnology
88
,
181
189
.
doi:10.1002/jctb.3936
.
Malila
R.
,
Lehtoranta
S.
,
Viskari
E.
2019
The role of source separation in nutrient recovery – comparison of alternative wastewater treatment systems
.
Journal of Cleaner Production
219
,
350
358
.
doi:10.1016/j.jclepro.2019.02.024
.
Martins
G.
,
Peixoto
L.
,
Brito
A. G.
,
Nogueira
R.
2014
Phosphorus-iron interaction in sediments: can an electrode minimize phosphorus release from sediments?
Reviews in Environmental Science and Bio/Technology
13
,
265
275
.
doi:10.1007/s11157-014-9343-5
.
Mendes
A. J.
,
Rufino
S.
,
Ferreira
R.
,
Brito
A. G.
2019
Resources Recovery and Decentralized Sanitation Fostered by a Urine Diversion Toilet Approach
.
European Water Association
,
Lisbon, Portugal
.
Nelson
N. O.
,
Mikkelsen
R. L.
,
Hesterberg
D. L.
2003
Struvite precipitation in anaerobic swine lagoon liquid: effect of pH and Mg:P ratio and determination of rate constant
.
Bioresource Technology
89
(
3
),
229
236
.
doi:10.1016/S0960-8524(03)00076-2
.
Rice
E. W.
,
Baird
R. B.
,
Eaton
A. D.
,
Clesceri
L. S.
2012
Standard Methods for the Examination of Water and Wastewater
, 22nd edn.
American Public Health Association/American Water Works Association/Water Environment Federation
,
Washington, DC
,
USA
.
Ronteltap
M.
,
Maurer
M.
,
Gujer
W.
2007
Struvite precipitation thermodynamics in source-separated urine
.
Water Research
41
,
977
984
.
doi:10.1016/j.watres.2006.11.046
.
Ronteltap
M.
,
Maurer
M.
,
Hausherr
R.
,
Gujer
W.
2010
Struvite precipitation from urine – influencing factors on particle size
.
Water Research
44
,
2038
2046
.
doi:10.1016/j.watres.2009.12.015
.
Shalaby
M. S.
,
El-rafie
S.
2015
Struvite precipitation and phosphorous removal from urine synthetic solution: reaction kinetic study
.
Bulletin of Chemical Reaction Engineering and Catalysis
10
,
88
97
.
doi:10.9767/bcrec.10.1.7172.88-97
.
Simha
P.
,
Ganesapillai
M.
2017
Ecological sanitation and nutrient recovery from human urine: how far have we come? A review
.
Sustainable Environment Research
27
,
107
116
.
doi:10.1016/j.serj.2016.12.001
.
Su
C. C.
,
Abarca
R. R. M.
,
de Luna
M. D. G.
,
Lu
M. C.
2014
Phosphate recovery from fluidized-bed wastewater by struvite crystallization technology
.
Journal of the Taiwan Institute of Chemical Engineers
45
,
2395
2402
.
doi:10.1016/j.jtice.2014.04.002
.
Taddeo
R.
,
Honkanen
M.
,
Kolppo
K.
,
Lepistö
R.
2018
Nutrient management via struvite precipitation and recovery from various agroindustrial wastewaters: process feasibility and struvite quality
.
Journal of Environmental Management
212
,
433
439
.
doi:10.1016/J.JENVMAN.2018.02.027
.
Tran
A. T. K.
,
Zhang
Y.
,
De Corte
D.
,
Hannes
J. B.
,
Ye
W.
,
Mondal
P.
,
Jullok
N.
,
Meesschaert
B.
,
Pinoy
L.
,
Bruggen
B. V.
2014
P-recovery as calcium phosphate from wastewater using an integrated selectrodialysis/crystallization process
.
Journal of Cleaner Production
77
,
140
151
.
doi:10.1016/j.jclepro.2014.01.069
.
Villa Gomez
D. K.
,
Enright
A. M.
,
Rini
E. L.
,
Buttice
A.
,
Kramer
H.
,
Lens
P.
2015
Effect of hydraulic retention time on metal precipitation in sulfate reducing inverse fluidized bed reactors
.
Journal of Chemical Technology and Biotechnology
90
,
120
129
.
doi:10.1002/jctb.4296
.
Wei
S. P.
,
van Rossum
F.
,
van de Pol
G. J.
,
Winkler
M. K. H.
2018
Recovery of phosphorus and nitrogen from human urine by struvite precipitation, air stripping and acid scrubbing: a pilot study
.
Chemosphere
212
,
1030
1037
.
doi:10.1016/j.chemosphere.2018.08.154
.
Xu
S.
,
Luo
L.
,
He
H.
,
Liu
H.
,
Cui
L.
2015
Nitrogen and phosphate recovery from source-separated urine by dosing with magnesite and zeolite
.
Polish Journal of Environmental Studies
24
(
2
),
269
275
.
doi:10.15244/pjoes/43611
.
Zhang
X.
,
Hu
J.
,
Spanjers
H.
,
van Lier
J. B.
2016
Struvite crystallization under a marine/brackish aquaculture condition
.
Bioresource Technology
218
(
1
),
151
156
.
doi:10.1016/j.biortech.2016.07.088
.

Supplementary data