This study examined total ammoniacal nitrogen (TAN) rejection by two reverse osmosis (RO) and two nanofiltration (NF) membranes as a function of pH for three ammonium salts to optimize conditions for a hybrid membrane system that can produce high-purity TAN streams suitable for reuse. The results showed that TAN rejection was significantly influenced by membrane type, feed pH, and the ammonium salt used. This study represents the first attempt to simulate real manure wastewater conditions typically found in pig manure. TAN rejection for (NH
4)
2SO
4 and NH
4HCO
3 reached up to 95% at pH values below 7, with the SW30 membrane showing the highest performance (99.5%), attributed to effective size exclusion and electrostatic repulsion of SO
42− and HCO
3− ions. In contrast, lower rejection was observed for NH
4Cl, particularly with the MPF-34 membrane, due to its higher molecular weight cut-off (MWCO), which diminishes both exclusion mechanisms. TAN rejection decreased markedly with increasing pH across the BW30, NF90, and MPF-34 membranes as the proportion of uncharged NH
3 increased. The lowest rejection rates (<15%) were recorded at pH 11.5 for both NF membranes. These results reveal a notable shift in separation behavior, where NH
3 permeation under alkaline conditions becomes dominant over the commonly reported NH
4+ retention at low pH. This novel insight offers a new perspective for optimizing membrane-based ammonia recovery in systems simulating realistic manure wastewater conditions. TAN recovery was evaluated using a hybrid membrane system, where NF membranes operated at high pH promoted NH
3 permeation, and the SW30 membrane at pH 6.5 enabled TAN rejection as (NH
4)
2SO
4. This hybrid system insight offers a new perspective for optimizing membrane-based ammonia recovery in systems simulating realistic manure wastewater conditions. Based on NH
3 permeation and membrane characteristics, the NF90 membrane was operated at pH 9.5, achieving a TAN recovery of 48.3%, with a TAN concentration of 11.7 g/L, corresponding to 0.9% nitrogen. In contrast, the MPF-34 membrane was operated at pH 11.5. The NF90–SW30 system also achieved a TAN recovery of 48.3%, yielding 11.7 g/L of TAN with a nitrogen content of 1.22%. These nitrogen concentrations indicate that both retentate streams are suitable for use as liquid fertilizers in the form of (NH
4)
2SO
4. A preliminary economic assessment estimated the chemical consumption cost at 0.586 EUR/kg and 0.729 EUR/kg of (NH
4)
2SO
4 produced for the NF90–SW30 and MPF-34–SW30 systems, respectively.
Full article