Separation and recovery of a complexing agent in the production of lithium hexfluorophosphate.
MetadataShow full item record
A novel process for the production of lithium hexafluorophosphate (LiPF6) has been developed by NECSA (Nuclear Energy Corporation of South Africa). The starting materials include hexafluorophosphoric acid (HPF6) aqueous solution and a complexing agent, pyridine. This study focused on the separation and recovery of the waste material from this production process (pyridine, ethanol, water and nitrogen), with high recovery rates and final mole purities for each component such that they can be recycled and reused in the process. Two separation schemes were investigated. Process A treats the waste stream firstly via a flash vessel to remove the nitrogen, then a conventional distillation column to separate the pyridine from the mixture followed by an extractive distillation column to separate the ethanol from water. The solvent chosen for extractive distillation was ethylene glycol. Process B follows the same scheme as process A for the nitrogen and pyridine separation; however it uses a pervaporation membrane modular setup, incorporating a polyvinylalcohol membrane in a spiral wound configuration, to separate ethanol from water as opposed to the use of the extractive distillation column in process A. Modifications were made to each process where an additional flash vessel was used to further purify the nitrogen stream. The Aspen Plus® process simulator was used to design the proposed separation schemes. The design and development of separation processes rely on accurate vapour liquid equilibrium (VLE) data. The ethanol/pyridine (40 kPa, 100 kPa, 313.15 K) and water/pyridine (40 kPa, 333.15 K) phase data were measured since the VLE data found in literature for these systems were inconsistent. A modified recirculating VLE still was used to measure the phase equilibrium data with sample analysis performed using a Shimadzu GC-2014. The measured data was regressed using various thermodynamic models in Aspen Plus®. The model that best suited the components of interest was the Peng-Robinson equation of state incorporating the Mathias Copeman alpha function with the Wong-Sandler mixing rules and the NRTL activity coefficient model (PR-MC-WS-NRTL). The binary parameters obtained with the use of this model was then incorporated into the Aspen Plus® computer software where it was used to simulate the proposed process schemes. The purpose of this is to provide a more accurate representation of the industrial process. A sensitivity analysis on the proposed process schemes was performed in order to obtain the optimal solution. The economic analysis showed that while process A met the specified purity and recovery of 99.9 mol %, it had a high total annual cost (approximately R46 million) due to the large amount of solvent required for extractive distillation. Process B had a significantly lower total annual cost (approximately R32 million). However, this is due to the area restriction imposed on the membrane (960 m2) due to the pyridine loss and slow separation process as the water composition decreased. The ethanol-rich stream (retentate) leaving the membrane unit contained a water content of 0.21 mol %; had this stream been v dehydrated to a water content of 0.01 mol % (similar to the purity obtained with extractive distillation), the total membrane area would have increased to a value of 2000 m2. The area required for the membrane then becomes practically and economically unfeasible. Furthermore, increasing the membrane area, increases the flow rate of the recycle stream (permeate) thereby decreasing the allowable recovery and purity of the core component, pyridine. Thus, process B does not meet the purity and recovery specifications for pyridine and ethanol. If it is attempted to achieve the ethanol purity specification, the membrane process becomes impractical due to the large membrane area required and the recovery and purity of the core component pyridine decreases. Furthermore, the additional flash vessel used in this process significantly increases the total annual cost (R4 million) and does not improve the product purities. Process A meets the required purity and recovery specifications for all components and is practically feasible. The additional flash vessel used in this process recovers a further 6.18 kg/hr of ethanol whilst slightly increasing the total annual cost (R0.14 million) and hence is the optimal solution.