Desalination of brackish water by electrodeionization: Experimental study and mathematical modeling

Highlights

• A novel model of EDI in a mixed-bed ion-exchange resin has been developed.

• The effect of cell voltage and concentration changes on current density can be predicted.

• 2D distribution of complex interrelation among variables is described by the model.

• Two mechanisms for desalination are experimentally identified at a high cell voltage.

Abstract

Electrodeionization (EDI) is an advantageous technology for desalination of waters with the range of salinity found in brackish water. Although modeling can help to understand the behavior of EDI cells, no reliable models connecting relevant operational variables have been published so far. Thus, this paper presents the experimental study and the mathematical modeling of an electrodeionization process used for the desalination of brackish water. The desalination process was carried out in a 4-compartment laboratory EDI cell. The model developed proposes an approach to the transport of ions in the liquid flowing in the mixed-bed resin's interstitial space. This transport is driven by the potential gradient in the liquid phase, formed by the different electric potential of cation- and anion-exchange resin particles and the Donnan potential at the interface. The model is used to analyze the current density distribution in the EDI cell and its effect on ion removal and performance to identify ineffective regions of the cell. The comparison of model calculation with experimental data shows the model to be capable of describing the effect of concentration (0.01 and 0.02 M) and cell voltage (7, 10.5 and 14 V) changes on current density and desalination rate.

Introduction

Potable water scarcity is a big problem today, and by 2025, two-thirds of the world's population could be affected by the lack of water [1]. One way to obtain either drinking or ultrapure water is from brackish water, whose salt concentration ranges from 292 mg L⁻¹ [2] to 30,000 mg L⁻¹ [3], depending on the source. There are different methods to desalinate water based on either the thermal or membrane process. The first one is a very mature technology that, however, presents some drawbacks, for example, mechanical vapor compression (MVC) requires a high amount of energy and exhibits scaling issues [4]; another one is related to multi-effect evaporation/distillation (MED), a labor-intensive process of no easy water quality control and low conversion (30%–40%) that requires large amounts of materials and space [5]. One of the most used membrane-based processes is the reverse osmosis (RO), which is beneficial for removing salt from seawater, but not for desalinating low-salinity feedwater [6]. The RO has problems with the membrane fouling that affects the lifetime of the membranes [7], high energy consumption (investment and operational costs) [2,8], and limited water recovery (30–60%) [8,9].

One of the goals of engineering is to make processes more efficient, not only energetically, but also environmentally. The hybrid ion exchange/electrodialysis (IXED) treatment, also called electrodeionization (EDI), has shown to be a very efficient process for brackish water desalination [8,10] with important environmental advantages [8,11]. EDI allows overcoming the limitations of each of the two processes alone: electrodialysis (ED) and ion exchange (IX). ED presents higher electric resistance in the dilute compartment (DC) as water becomes desalted, such that more energy is required to obtain high purity water leading to a dramatically decreased efficiency [3,12]. On the other hand, the chemical regeneration of exhausted resin in the IX process is required to recycle it so that the IX process cannot be operated continuously [12]. However, the synergy of two technologies in the EDI process allows continuous operation under the influence of the electric field to regenerate the resin. Thus, stable water quality can be produced. The electric voltage applied between the electrodes continuously desalinates water with no need for chemical regeneration of saturated ion exchange resins. This important advantage of EDI makes it a more environment-friendly process [12,13].

Different basic configurations of electrodeionization have been analyzed in the literature: separate beds (EDI-SB), beds separated by a bipolar membrane (EDI-BM), simple bed (IXED-SB) and mixed bed (EDI-MB) [14,15]. The last system is typically used for producing high purity water from solutions with a low concentration of ions. Moreover, the mixed ion exchange resin bed (filling the desalted channel) improves mass transfer, facilitates resin regeneration, and reduces the resistance of desalted solution [12,13]. However, there is a need for new methods of reactor analysis that could be applied to EDI cell design and scaling.

A valuable approach that can help understand the complex relationships between the geometric parameters and operating conditions, and their effect on EDI cell performance, is modeling. The mathematical model allows predicting the concentration of ions in a given time, the electrical energy consumption [16], the potential and current distribution, and the water dissociation related to high Donnan potential distribution over the membrane/solution interface [17].

Although there are several modeling studies of electrodialysis [16,18,19], and some of them have included complex phenomena such as electro-migration, diffusion, electro-osmosis, osmosis, and convection [20]. The concepts used in ED models have not been applied for describing analogous mechanisms of species transport in the interstitial solution and resins, and the phenomena at the resin-solution interfaces in ion exchange resin bed of EDI. The mathematical model of the hybrid process of electrodialysis with a mixed bed of anion and cation-exchange resin (AR and CR) might be more complex. Thus, a model for electrodeionization has not yet been clearly established since the behavior of a mixed bed ion-exchange resin is not entirely elucidated. The way the resin-filled compartment works is the key issue in the behavior and success of the EDI cell. This compartment is able to remove ions from the dilute solution, transfer them to the resin and from there to the membranes (anion-exchange, AM, or cation-exchange membranes, CM) and finally, to the solution in the concentrate compartment, (CC). Different mechanisms might be present in each step of the transport of species, ion exchange at the resin/electrolyte interfaces, and resin regeneration. Some models have been proposed for specific EDI cell configurations, for example, parallel conductance in resin and electrolyte [21,22], and an ion-exchange process controlled by film diffusion on the resin particles [23,24]. In the first case, configurations (a) and (d) of Fig. 1 were used, where water splitting took place on the DC/AM interface in configuration (a) and on both DC/AM and CM/AR interfaces in configuration (d). The authors reported a discrepancy between the model and the experimental data, which was due to the lack of understanding of the water dissociation mechanism and the assumption of negligible water dissociation on the surface of cation resins. In the second case, configurations (d) and (c) of Fig. 1 were used, but only in the latter, reactions in the electrodes produce H⁺ and OH⁻ ions to regenerate the resin.

Although resin regeneration by H⁺ and OH⁻ ions generated in water splitting can explain the EDI process, it has been experimentally shown that water splitting occurs at high current densities. Jiang et al. [25] removed traces of Cs⁺, Sr²⁺, and Co³⁺ ions with a continuous electrodeionization. They showed that the percentage of removal of metal ions increases with the current intensity, but an excessive current gives rise to hydrolysis forming unwanted compounds, such as Co(OH)₂. Lopez et al. [10] found that current efficiency decreased as the cell voltage increased in brackish water desalination in mixed resin beds of EDI stack since water splitting occurred at a higher voltage. In addition to resin electroregeneration associated with water splitting, there is another operating regime of EDI cell, the enhanced transfer, where the resin remains in the salt form [26,27].

Regarding the transport of ions in the dilute compartment and ion exchange at the resin/liquid interface, experimental work has shown the process to be mass transfer controlled. Mahmoud et al. [28] studied the EDI of copper solutions in a simple bed like that of Fig. 1(b), but with cation exchangers in place of anion exchangers, i.e., they used a bed of cation-exchange resin between two cationic membranes. EDI was represented by n continuous stirred tank reactors (CSTR) in series, using a mass transfer coefficient to characterize the flux of ions. On the other hand, in our research group, electrodeionization of sodium arsenate solutions was performed in a 5-compartment simple bed configuration, IXED-SB (Fig. 1(b)), in a recirculation mode batch system [11]. Measured changes of pH, arsenic concentration, and conductivity at different cell voltages and initial arsenic concentration indicate that the main source of H⁺ and OH⁻ to regenerate the resin takes place at the anion-exchange membrane/concentrate interface, in the lower potential side (left-hand side in Fig. 1(b)). A model was also developed [16] coupling the mass transfer controlled ion exchange with the resin regeneration by OH⁻ generated by water splitting and Nernst-Planck equations, including the Donnan potential, in the resin-free compartments in order to describe complex interactions among variables (potential, the concentration of each species and pH). However, in the case of mixed-bed EDI, required for brackish water desalination, the mechanisms taking place in the dilute compartment can be very different owing to the liquid contact with both types of resin particles, AR and CR. Therefore, it is essential to study experimentally and theoretically, the way the mixed-bed ion exchange works in order to evaluate the effect of design and operating variables on its performance.

For the desalination of brackish water by the EDI process, a robust methodology to design and scale up efficient EDI cells is necessary. Mathematical modeling is a powerful tool to understand the behavior of EDI cell and can be used to develop this methodology. Thus, in the present work, an experimental study is carried out in a laboratory EDI cell with EDI-MB configuration for desalination of artificial brackish water in order to develop a model capable of describing the effect of concentration and cell voltage changes on the tertiary current and potential distribution. To our knowledge, such a model has not been developed before, and the present work can thereby fulfill this gap.