Simultaneous phosphorous and nitrogen recovery from source-separated urine: A novel application for fertiliser drawn forward osmosis

Re-thinking our approach to dealing with waste is one of the major challenges in achieving a more sustainable society. However, it could also generate numerous opportunities. Specifically, in the context of wastewater, nutrients, energy and water could be mined from it. Because of its exceptionally high nitrogen (N) and phosphorous (P) concentration, human urine is particularly suitable to be processed for fertiliser production. In the present study, forward osmosis (FO) was employed to mine the P and N from human urine. Two Mg2+-fertilisers, i.e. MgSO4 and Mg(NO3)2 were selected as draw solution (DS) to dewater synthetic non-hydrolysed urine. In this process, the Mg2+ reverse salt flux (RSF) were used to recover P as struvite. Simultaneously, the urea was recovered in the DS as it is poorly rejected by the FO membrane. The results showed that, after concentrating the urine by 60%, about 40% of the P and 50% of the N were recovered. XRD and SEM - EDX analysis confirmed that P was precipitated as mineral struvite. If successfully tested on real urine, this process could be applied to treat the urine collected in urban areas e.g., high-rise building. After the filtration, the solid struvite could be sold for inland applications whereas the diluted fertiliser used for direct fertigation of green walls, parks or for urban farming. Finally, reduction in the load of N, P to the downstream wastewater treatment plant would also ensure a more sustainable urban water cycle.


41
Sustainability in wastewater treatment is one of the significant issues of this century (Xie et 42 al. 2016). In particular, rethinking wastewater as a valuable resource is crucial in meeting 43 adequate sanitation, water and fertilier demand to feed a growing population (Elser and 44 Bennett 2011, Xie et al. 2016). For these reasons, the efficient separation, treatment and reuse 45 of human urine have gained increasing attention due to its inherent value potential (Maurer et 46 al. 2006, Udert and Wächter 2012, Zhang et al. 2014). In fact, despite the low volumetric load 47 of urine (i.e., less than 1% of the overall wastewater volume), it accounts for approximately 48 80% of the nitrogen (N), 50% of the phosphorus (P) and 55% of the potassium (K) load in 49 most of the wastewater treatment plants (Liu et al. 2016, Maurer et al. 2006, Udert and 50 Wächter 2012, Zhang et al. 2014). In particular, the amount of P contained in the urine is the 51 single most significant source from urban areas (Zhang et al. 2014). Given the forecasted 52 depletion of minable phosphorous rocks, reusing the P from waste streams could significantly 53 enhance the sustainability of the urban water cycle (Xie et al. 2014, Xie et al. 2016. 54 Additionally, with the increase in the size and densities of modern cities high-rise buildings 55 are already becoming necessary. In these, urine separation and gravity-driven collection 56 might become a feasible choice. After treatment, the recovered nutrients could be reused in 57 several urban applications such as green walls, parks or urban farming. Simultaneously, the 58 load of N, P to the downstream wastewater treatment plants would be reduced, possibly 59 making their operation less energy demanding (Ishii and Boyer 2015, Kavvada et al. 2017). 60 Despite the applicability of raw human urine for direct fertigation, its nutrients imbalance 61 (i.e., mainly N), and low nutrients concentration (i.e., N: 0.9%, P: 0.06%, K: 0.3%) as well as Their study showed that urea is practically not rejected by the FO membrane, therefore, 114 impeding a safe discharge of the diluted RO brine in the environment. 115 We herewith present a novel FDFO concept for concentrating human urine, where both 116 limitations of FO (i.e., RSF of the draw solutes and urea/NH 3 rejection loss), are beneficial to 117 recover both nitrogen and phosphorous from urine. In fact, in this study, the feasibility of 118 using a Mg 2+ -based fertiliser draw solution to dewater fresh (i.e. non-hydrolysed) human 119 urine is investigated. In this concept, the reverse solute diffusion will trigger P-recovery via 120 struvite precipitation, while the rejection loss of urea/NH 3 will enrich the Mg-fertiliser with 121 valuable nutrients. At the same time, the final volume of urine will be reduced thereby 122 improving the efficiency in downstream processes for N-recovery (e.g. ammonia stripping). 123 This initial study will address the following:

132
In Figure 1 the schematic representation of the system is displayed for a better understanding 133 of the process. The synthetic fresh urine feed solution (FS) was prepared according to the 134 recipe of Udert et al. (Udert et al. 2006), and its composition is displayed in 135 urine has a about half the osmotic pressure compared to hydrolysed urine. This means that, 137 theoretically, higher J w and concentration can be achieved when fresh urine is chosen. 138 Secondly the acidic pH of fresh urine (i.e. 5.5 -6.

Forward osmosis experimental set-up 161
An FO set-up, similar to the one used in our previous study (Phuntsho et

Modelling of water flux, reverse salt flux and achievable P-recovery 177
To better understand and comment the experimental results, the water flux ( ), specific Where σ is the reflection coefficient, assumed as unity (complete rejection of the solute), ‫ܭ‬ Where β is the van't Hoff coefficient (i.e., 2 for MgSO 4 and 3 for Mg(NO 3 ) 2 ), R g is the 201 universal gas constant, and T is the absolute temperature (Tang et al. 2010). 202  solution. This step is not necessary when real urine is used since the remaining urea in the 234 feed would spontaneously hydrolyse causing pH to rise to 9.5. Afterwards, the FS was then 235 stirred for 2 hours and later filtered using a 0.45 µm pore-size filter (Merck, Millipore).

Experiments with fresh urine as feed solution 282
Once the model was validated with DI water as FS, long-term tests with fresh synthetic urine 283 were conducted to measure J w , FS up-concentration, Mg 2+ -SRSF, P and N recovery data. NaCl as DS. Another explanation could be due to the stripping of NH 3 during the experiment, 296 due to the relatively fast feed recirculation rate (i.e. 300 mL/min). However, ammonia 297 volatilisation should not significantly affect the system since at pH ≤ 5, the equilibrium 298 ‫ܪܰ‬ ଷ ೌ ୀଽ.ଶହ ሯልልልልሰ ‫ܪܰ‬ ସ ା should be heavily shifted on ‫ܪܰ‬ ସ ା which is non-volatile. Besides, the 299 NH 3 /NH 4 + concentration is relatively low comapred to urea. Finally, 60% FS volume 300 reduction was achieved without any sign of membrane damage. In fact, decrease a in the M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT 20 bar to around 30 bar) and the decrease in the osmotic pressure of the DS (due to dilution). 303 Flushing the membrane with DI-water was generally enough to clean the membrane surface 304 and restore the initial performances ( Figure S3). Nonetheless, experiments having real urine 305 as FS are necessary to better assess membrane fouling in this process. 306 Overall, the model showed an acceptable agreement with the measured data and the 307 experimental tests confirmed that at least 60% FS concentration was achievable without any 308 sign of scaling on the membrane. 309  this process, the urea concentration after the FDFO process is about the same as in the diluted 320 urine (i.e. 50% urea removal was achieved but also 60% volume reduction). Therefore, it is 321 expected that, after the urea hydrolysation process, the concentrated urine would yield a 322 similar final pH. Phosphorous and magnesium concentrations were measured before and after 323 the alkalinisation of the feed, and the insoluble minerals produced were analysed via XRD 324 and SEM -EDX. Figure   The results, plotted in Figure 7, showed that this hypothesis was, at least partially, correct. In 358 fact, at 60% FS concentration, up to 50% of the urea was recovered in the process. In 359 particular, the flux of urea to the DS was found to be much higher than the RSF of NO 3 when 360 Mg(NO 3 ) 2 was used.

367
To conclude, at this stage, not all the P and N in the urine were recovered. However, the 368 process still can be further optimised. For instance, the effect of transmembrane pressure 369 (TMP) in the RSF and urea rejection could also be investigated. Additionally, the 370 performances of real urine should be investigated as it will likely behave differently due to its 371 very heterogeneous concentration. Nonetheless, the simplicity and low cost of the process 372 could incentivise further investigations to reach higher P, N recovery. 373 374

375
This study investigated a novel application for FO to concentrate human urine while 376 simultaneously recovering the phosphorous and nitrogen in it. Nitrogen recovery is achieved 377 by urea transport over the FO membrane, enriching the fertiliser draw solution with urea.

M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT FO (i.e. RSF and poor urea/NH 3 rejection) are desirable and contribute to the simultaneous 381 recovery of P/N from the urine, while reducing its volume. Process modelling, as well as 382 experimental tests, were used to understand better and critically analyse the results. Two 383 Mg 2+ -fertilisers (i.e., MgSO 4 and Mg(NO 3 ) 2 ) were identified as the most promising for this 384 application. Among the two Mg(NO 3 ) 2 displayed a much higher water flux and osmotic 385 pressure, achieving equal P and N recoveries as MgSO 4 . Overall, the FDFO process enables 386 to obtain, simultaneously, the following outcomes: 387 • Reducing the volume of urine by more than 60% thereby possibly improving the 388 efficiency in downstream processes for N-recovery (e.g. ammonia stripping), 389 • Recovering 50% of the nitrogen in the urine, 390 • Recovering 40% of the phosphorous as struvite fertiliser. 391 To conclude, this low cost and robust treatment process enable a unique way to integrate 392 urine volume reduction and P and N recovery. The economic feasibility of this technology, 393 to produce fertiliser for green walls, parks or urban farming applications, should be further 394 investigated in view of enhancing the sustainability of the urban water cycle.

Research highlights
• FDFO could be used for simultaneous urine dewatering, urea and phosphorous recovery; • About 50% of the nitrogen and 40% of the phosphorous were recovered while 60% of urine concentration was reached; • Phosphorous precipitated as mineral struvite;