Elsevier

Desalination

Volume 417, 1 September 2017, Pages 36-51
Desalination

Improvement of MCDI operation and design through experiment and modelling: Regeneration with brine and optimum residence time

https://doi.org/10.1016/j.desal.2017.05.004Get rights and content

Highlights

  • The ion transport model for membrane capacitive deionization(MCDI) was extended.

  • An asymmetric voltage distribution and activity coefficients were included.

  • The feasibility of using a brine stream for MCDI regeneration was assessed.

  • Brine recycling increased water recovery but productivity decreased.

  • Residence time increases in the MCDI unit were ineffective beyond a certain range.

Abstract

Membrane Capacitive Deionization (MCDI) is an energy efficient, electrochemical desalination technology, in which ions are removed from a salty stream upon applying a constant voltage or current. The ions are stored in carbon electrodes and then released back into the stream by reversing the polarity. In this work, we aimed to assess the feasibility of using a brine stream to regenerate the MCDI unit in order to improve water recovery. We further aimed to determine the optimum residence time in the MCDI unit. To achieve these objectives, we first enhanced the ion transport model previously developed for MCDI by independently measuring the counter-ion and co-ion diffusion coefficients in the ion-exchange membranes. These experiments allowed for an asymmetric model of the MCDI unit where the voltage drop across the cation exchange membrane was greater than that across the anion exchange membrane. Using this revised model, we found that in batch operation, a brine to feed water concentration ratio of around two was optimum. In continuous operation, over 40% enhancement in water recovery could be achieved when the regeneration brine was partially recycled, but water productivity dropped. We further showed that the maximum desalination capacity did not increase beyond a critical residence time in the MCDI cell, while the water recovery decreased.

Introduction

Desalination of sea and brackish water has become more common over the past decades to meet freshwater requirements for domestic, agricultural and industrial demands. Desalination capacity, energy demand, regeneration method and water recovery are the main features distinguishing the technologies developed for this purpose, which are generally thermally or membrane based [1], [2]. Capacitive Deionization (CDI) is an alternative, electrochemical water treatment method in which ions are temporarily adsorbed in electrical double layers of two oppositely charged porous carbon electrodes [3]. Therefore the desalination capacity is reliant on the carbon material used in the electrode fabrication [4], [5]. To enhance the performance of CDI, Membrane Capacitive Deionization (MCDI) has also been proposed, where ion-exchange membranes (IEMs) are placed in front of the charged electrodes to inhibit the co-ions, i.e. ions carrying the same charge as the surface, from reaching the electrodes [6], [7]. In CDI or MCDI, ions or other charged species are removed from the concentrated feed. However, this is not the case for most other desalination technologies in which water is separated from the polluted stream. It is thus apparent that at lower concentrations of the charged species in comparison with that of water, the energy consumption of CDI based processes will be significantly lower than the latter [8]. Therefore, CDI or MCDI is mostly employed as an energy-efficient method for brackish water remediation where the salinity is limited to 1000 mg L 1 [9].

Once the carbon pores reach their saturation limit, the CDI or MCDI unit is not capable of adsorbing ions anymore. Therefore, a regeneration step is required to deplete the electrode materials of the charged species. In CDI, the previously adsorbed ions are released back to the liquid phase by dropping the system voltage to zero, while in MCDI it is more common to release these ions by reversing the polarity. As a result the stack undergoes repetitive adsorption/desorption cycles [10]. Ease of regeneration adds to the merits of this promising desalination method [11].

While many attempts have been made towards synthesis of novel carbon materials with improved physico-chemical properties for CDI processes [12], [13], less research has been conducted on the operational aspects of both CDI and MCDI. Zhao et al. [14] were the first to optimize the salt adsorption in MCDI by varying operational variables within both constant voltage and constant current modes. Recently, García-Quismondo et al. [15] considered new operational modes to increase energy efficiency in CDI by varying charge and discharge current densities and also investigated the use of a concentrated brine stream during discharge. Yet more research is needed to improve the design aspects of MCDI. To illustrate, feed water recovery, i.e. the volumetric ratio of desalinated water produced to that of the feed water, is one such operational metric used to evaluate water treatment processes [16]. In the field of MCDI, the higher the feed water recovery, the less feed water volume is wasted during the desorption step. In this work, we specifically consider the potential of regenerating the MCDI unit with a stream of higher salt concentration than the feed to increase the feed water recovery. Since this is a very critical question, more exploration is required to determine its feasibility especially in the presence of ion-exchange membranes. If a more concentrated stream can be utilised in the desorption step, a certain amount of the water for regeneration can be recycled a number of times which then leads to much greater water recovery.

In this paper, we explore this concept using both experimental techniques and a theoretical model. The modified Donnan Theory developed by Biesheuvel et al. [17], [18] is the most well developed model available in the literature to describe the storage of ions in the EDLs of micropores (< 2 nm) of carbon electrodes. The same research group has also proposed a dynamic ion transport model to describe the performance of the MCDI unit [17], [19], [20]. Tang et al. [21] has recently used a similar approach to model the removal of fluoride in CDI. This ion transport model includes all the mass and charge balances throughout the MCDI unit and is combined with modified Donnan theory to describe the ion storage in the electrical double layers (EDLs). Nonetheless, this theoretical method is based on various simplifying assumptions. In this work we avoid some of these assumptions by (i) measuring the diffusion coefficients in the ion-exchange membranes; (ii) using these values to adjust the voltage distribution across the two half-cells and (iii) including the ion activity coefficient in the solution and the membrane phases. The novel features introduced into the mathematical approach strengthen the model, as estimation of some parameters is replaced with direct measurements. The improved ion transport model was then utilised to determine the impact of partially recycling the regeneration brine, and further to determine the optimum residence time in the MCDI unit.

Section snippets

MCDI model

To describe the deionization process the modified Donnan (mD) Theory proposed by Biesheuvel et al. [18] is used to describe the ion storage in the electrical double layers (EDLs) of the carbon electrode (Section 2.1). Mass transfer equations then enable the prediction of the ion flux from the flow channel formed by a spacer through the ion-exchange membranes to access these EDLs (Section 2.2). In this approach, the effect of faradaic reactions and also the contribution of hydronium and hydroxyl

Materials

The analytical grade of all chemicals was utilised. Polyvinylidene fluoride (PVDF, Mw ~ 530,000 g mol 1, Sigma-Aldrich), Nsingle bondN dimethylformamide (DMF, 99.8%, Merck Millipore) and activated carbon (AC Norit SA 4, Cabot Norit Activated Carbon, USA) were utilised for electrode preparation. Sodium chloride (NaCl, 99.7%, ChemSupply) was used to prepare salt solutions at different concentrations. Sodium hydroxide (NaOH, 97.0%, ChemSupply), hydrochloric acid (HCl, 37.0%, ChemSupply), caesium chloride

Characterization of the MCDI cell

Initial experiments focused on determining the parameters needed for the MCDI model to be effective. The parameters determined experimentally for the carbon electrodes are summarized in Table 1. Details of the experimental procedures used to determine these parameters and further characterization details are provided in the Appendix D.

The permeability coefficient of the counter-ions (Na+ in case of CMX and Cl in case of AMX) was measured using the method explained in Section 3.2. The observed

Conclusion

In this work, we have investigated whether or not a brine stream, which is more concentrated than the feed solution, is capable of regenerating a MCDI unit and whether there is an optimum residence time. To date, no other research group has tackled these important research questions.

To approach these problems, we first extended the MCDI model available in the literature, to specifically account for the activity coefficients of the counter-ions and co-ions within the membranes themselves.

Acknowledgements

Armineh Hassanvand acknowledges The University of Melbourne for the IPRS (International Postgraduate Research Scholarship) and APA (Australian Postgraduate Awards) scholarships, which are funded by the Australian Government. George Chen and Sandra Kentish acknowledge research funding from the Australian Research Council's Industrial Transformation Research Program (ITRP) scheme (Project Number IH120100005). The ARC Dairy Innovation Hub is a collaboration between The University of Melbourne, The

References (39)

Cited by (0)

View full text