Recovery of ammonia (NH₃) from residual waters offers various reuse opportunities, such as the production of fertilisers and the generation of electricity and heat. However, simultaneous evaporation of water (H₂O) during NH₃ stripping under vacuum results in diluted recovered NH₃ gas with high H₂O contents. Whereas porous gas-permeable membranes are already used for vacuum NH₃ stripping, the use of non-porous silica-based pervaporation (PV) membranes showed promising results in recent literature, with respect to more selective transfer of NH₃ compared to H₂O. In this study, we assessed the selectivity of NH₃ over H₂O transfer (S_NH3/H2O) for different types of membranes, under various hydraulic conditions and feed water compositions. The three following membranes were tested: a porous gas-permeable polytetrafluoroethylene (PTFE) membrane, a hydrophilic (Hybrid Silica PV) membrane and a hydrophobic polydimethylsiloxane PV (PDMS PV) membrane. For the PTFE and the Hybrid Silica PV membrane, S_NH3/H2O ranged between 0.1 and 0.4, indicating that the transfer of NH₃ was consistently less preferred compared to the transfer of H₂O. The preference for H₂O over NH₃ transfer through the membranes at various hydraulic conditions and feed water compositions can be assigned to the similarity in polarity and kinetic diameter of NH₃ and H₂O and the low relative concentration of NH₃ in the used feed waters (approximately 0.1–1.0 wt%). The PDMS PV membrane showed negligible NH₃ transfer and deteriorated rapidly during the NH₃ stripping experiments. The S_NH3/H2O of both gas-permeable and PV membranes was higher for unsteady than for steady hydraulic conditions. Furthermore, the S_NH3/H2O of the both PTFE and the Hybrid Silica decreased when the ionic strength of the feed water increased from 0.0 to 0.8 mol∙L⁻¹ and when the NH₃ feed water concentration increased from 1 to 10 g∙L⁻¹. According to the results, the used PV membranes did not show selectivity of NH₃ over H₂O transfer. In fact, the used PV membranes consistently had a lower S_NH3/H2O than the PTFE membrane. Hence, the dense silica-based PV membranes did not allow for the recovery of gaseous NH₃ from water, with lower H₂O content in the recovered gas, compared to porous PTFE membranes.

Traditionally, industrial processes produce wastes that, even though often containing useful materials, are discarded, contributing to environmental pollution and depletion of natural resources. An example of such wastes are brines, flows of concentrated salts, produced in water treatment processes, which are now routinely discharged into receiving water bodies. Brines however can also be considered as flows of reusable materials which should be recovered, and the Zero Brine cooperation project aims to develop processes for that purpose. For a demineralized water production plant in the port of Rotterdam (the Netherlands), a closed water processing cycle was proposed to treat the large volume of spent Ion Exchange (IEX) regenerant brine which, apart from recovering demineralized water, is also intended to produce magnesium (Mg2+) and calcium (Ca2+) salts, with the highest purity possible, from the otherwise discharged brine. The process scheme includes nanofiltration (NF) for separating mono- and multivalent ions, followed by sequential chemical precipitation of Mg2+ and Ca2+ ions from the NF concentrate, and production of demineralized water by evaporation of the NF permeate. The concentrate of monovalent ions produced in the evaporator, essentially a concentrated sodium chloride solution, in its turn might be reused for IEX regeneration. Part of the supernatant of the sequential precipitation may be fed to the evaporator as well, but bleeding the other part of this supernatant is essential in order to maintain process stability, avoid accumulation of minor pollutants, and reduce scaling. In this study, various scenarios to operate the process were modeled, using PHREEQC and Excel. According to the simulation results, recovery of ≈97% of Mg2+ and Ca2+ is possible, the latter with a higher purity than the former. The main factors affecting the results are the concentration of carbonate present in the spent IEX regenerant, as well as characteristics of the NF membrane and the dosing of sodium hydroxide in the sequential precipitation steps. The results of the simulations were used for the design and operation of a pilot plant, comprising all mentioned process steps.