Capacitive deionization of water using mosaic membrane

Desalination 426 (2018) 1–10

Contents lists available at ScienceDirect

Desalination journal homepage: www.elsevier.com/locate/desal

Capacitive deionization of water using mosaic membrane

MARK

Yu.M. Volfkovich , А. Yu. Rychagov , А.А. Mikhalin , М.М. Kardash , N.А. Kononenko , D.V. Ainetdinovb, S.A. Shkirskayac, V.Е. Sosenkina a,⁎

a b c

a

a

b

c

A.N. Frumkin Institute of Physical Chemistry and Electrochemistry of RAS, Leninskii prospect, 31, 119071 Moscow, Russian Federation Engels State Tehnology University, Svobodi Ave. 17, Engels, Saratov Region 410054, Russian Federation Kuban State University, Stavropol'skaya Str., 149, Krasnodar 350040, Russian Federation

G RA P H I C A L AB S T R A C T

A R T I C L E I N F O

A B S T R A C T

Keywords: Capacitive deionization Mosaic membrane Activated carbon Method of standard contact porosimetry Surface conductivity

Capacitive deionization of water using membrane-electrode assembly was researched. The approach that allows to decrease energy consumptions of water purification was considered. The method provides usage of mosaic membrane containing both cation and anion exchange fragments instead of glass spacer. Counter-ions inside the membrane ensure rather high ionic conductivity even in pure water. Mosaic membranes based on polyethylene (film type) and phenol-formaldehyde (fibrous type) matrices were studied, their electric conductivity and exchange capacity were determined. Deionization in static and dynamic electrochemical cells, which were filled with deionized water and 0.005 M KCl solution respectively, was researched. The mechanism of electric double layer charging inside pores of the electrodes impregnated with pure water has been proposed. Specific energy consumptions for deionization of very diluted solutions are sufficiently lower for the cell containing mosaic membrane than those for the cell with inert glass spacer. Minimum energy consumptions and maximum deionization degree are reached at cell voltage of 1.4 V. The value of specific energy consumptions is 12 Wh mol− 1 for the laboratory cell containing the mosaic membrane, when degree of deionization reaches 50%, cell voltage is 1.4 V, electrode area is 50 cm2, initial concentration of the KCl solution is 0.005 M.

Abbreviations: AC, activated carbon; CDI, сapacitive deionization; ECSC, electrochemical supercapacitor; EDI, electrodeionization; EDL, electric double layer; HDCE, highly dispersive carbon electrode; MCDI, membrane capacitive deionization; MEA, membrane-electrode assembly; MM, mоsaic membrane; MSCP, method of standard contact porosimetry ⁎ Corresponding author. E-mail address: [email protected] (Y.M. Volfkovich). http://dx.doi.org/10.1016/j.desal.2017.10.035 Received 27 July 2017; Received in revised form 6 October 2017; Accepted 19 October 2017 0011-9164/ © 2017 Elsevier B.V. All rights reserved.

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

1. Introduction Capacitive deionization (CDI) is a promising electrochemical method of water desalination. Economically, it is the most attractive technique in comparison with reverse osmosis (3 times cheaper) [1,2], distillation, and electromembrane separation due to lower energy consumptions [1–3] and simpler equipment [4]. As opposed to electrodeionization (EDI), which involves ion exchange membranes and granulated polymer [5–7] or inorganic [7] ion-exchanger between them, electrodes play the key role in CDI processes. CDI involves the passage of aqueous solution through the electrochemical cell between two highly dispersive carbon electrodes (HDCEs) with high specific surface area (500–2500 m2 g− 1) [1,2,8]. Low potential difference (≥ 1.2 V) is applied providing a high safety level. Porous inert spacer is placed between the electrodes to separate them. In a number of works, for instance in [9], glass or polymer fibrous plate is used as a spacer. Adsorption of anions and cations occurs on positively and negatively charged electrodes respectively, electric double layer (EDL) is charged similarly to that in an electrochemical supercapacitor (ECSC) [10,11]. This results in deionization of the solution. When the circuit is closed or polarity is reversed, ions diffuse from the solid-liquid interface back to the solution, increasing the solution concentration and causing energy regeneration. The deionization stage corresponds to the charging of the ECSC, while the regeneration stage is related to discharging. During the regeneration stage, significantly lower amount of water is supplied to the cell. After regeneration of the electrodes, the cell is converted to a deionization cell, whereas CDI device including at least two electrochemical cells operates continuously. In other words, while deionization takes place in the first cell, regeneration occurs in the second cell. Thus, the consumed energy can be partially compensated by electrical energy from the regeneration unit. Electrode materials for CDI have been in the focus of attention in the last years. Different types of both single-component, such as activated carbon (AC), aerogels, nanotubes, graphene [12–15], and AC–based composite materials (AC-AC composite, AC-metal oxide composite, ACpolymer composite and AC-polymer-metal oxide composite) [16–19] have been suggested. CDI application for removal of different salts from water has been investigated in detail [20–22]. In [23,24]., the equations predicting the lowest regeneration time and the highest desalination degree for CDI at constant current have been found using the mathematical model of adsorption cycle. During purification process, the effluent concentration reaches the highest purity level after a certain period of time. The method of membrane capacitive deionization (MCDI) can be considered as a modification of CDI. In this case, the anion exchange membrane is adjacent to the positively charged electrode, and the cation exchange membrane borders on the cathode [25–28]. The anion exchange membrane prevents cation transport to the anode, while the cation exchange membrane makes impossible anion movement towards the cathode. This provides more complete separation of cations and anions in the MCDI cell. When the HCDEs, which are characterized by high specific surface area, are used and no membranes are applied to the CDI process, separation of oppositely charged ions occurs due to EDL charging inside pores of the electrodes. However, the membranes provide additional hydrodynamic resistance (increasing energy consumptions). This is a disadvantage of the MCDI method in comparison with CDI. Deep water purification requires high energy consumptions also in the case of CDI because of high ohmic losses caused by huge electrical resistance of pure water. At the same time, ionic conductivity of AC electrodes is rather high even in pure water [29] due to surface conductivity caused by large amount of ion exchange groups on the surface of carbon materials [30]. Therefore, at the final stage of the deionization process, energy consumptions are determined by practically zero conductivity of water, which fills pores of the spacer between the electrodes.

Fig. 1. Structure of MM: 1 - polymer matrix (polystyrene matrix cross linked with divinylbenzene for film type of mosaic membrane or phenol formaldehyde for fibre type of mosaic membrane); 2 - micropores and mesopores; 3 - positively charged fixed groups (amino groups) in anion-exchanger particles; 4 - anions (counter-ions to anion-exchanger); 5 - negatively charged fixed groups (sulfo groups) in cation-exchanger particles; 6 - cations (counter-ions to cation-exchanger); 7 - cathode; 8 - anode.

The aim of the work was to develop and investigate the membraneelectrode assembly (MEA) for CDI processes to obtain pure water. The MEA design provides specially manufactured membrane of mosaic structure (mosaic membrane, MM) instead of inert porous spacer in order to decrease energy consumptions as much as possible. The membrane contains both cation and anion exchange groups. It is also necessary to use HDCEs that are characterized by highly developed surface and contain oppositely charged surface groups. As assumed, the energy losses for the MEA are much less than those for the cell containing a conventional porous spacer. This is due to high ionic conductivity of MM that is caused by ion exchange groups (similarly to any ion exchange membrane). 2. Experimental 2.1. Membranes Mosaic membranes (MM), in which cation- and anion exchange fragments are randomly and homogeneously distributed through the volume of inert polymer matrix (Fig. 1), were applied to investigations. In fact, the whole volume of this membrane is characterized by mosaic structure. According to the review [31], similar membranes are related to amphoteric materials, As for known MM, their cation and anion exchange fragments are continuous and located in parallel to each other. Thus, each fragment provides continuous pathway for ion transport. However, mosaic structure is attributed only to outer surface of the membrane, not to the whole volume. Both film and fibrous MM were used in this work. The film heterogeneous membrane produced by NIIPM Company (Engels, Russia) was manufactured by mixing of powders of AV-17 anion exchange resin (70 mass %) and KU-2 cation exchanger (30%). The resins were produced by Schekinoazot Company (Schekino, Russia). Polyethylene was used as a binder, nylon threads were also applied in order to provide mechanical durability of the membrane. The mixture was formed into sheets under elevated pressure and temperature [32]. Structure and functional properties of MM of different composition were investigated 2

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

(

earlier [33]. Fibrous MM “Polycon” was manufactured with the method developed in the Engels Technological Institute (Engels, Russia). The technique involves polycondensing filling of polymeric composite materials. “Polycon” MM is the composite material containing both strongly acidic cation exchange constituent and weakly basic anion exchange component with mass ratio of 1:1. Synthesis of ion exchangers was performed on the outer surface and inside fibrous phenol-formaldehyde matrix similarly to [34,35]. When 30–35% degree of curing of the polymers was reached, the membrane was thermostated at 130 °С and 2 МPa.

)

r

effective pore radius r ∗ = cos θ , where is the wetting angle , was plotted. The r∗ parameter was determined from the MSCP measurements using octane and water. Octane is used more often, since this liquid wet any material almost perfectly (θ = 0°). Water is applied to study hydrophilic-hydrophobic properties. When materials are wetted with water poorly (for example, carbon materials), pore size distributions that are obtained using water are shifted to higher r∗ values comparing with the curves measured with octane. Based on this shift, the contact angle was r calculated as mentioned above cos θi = r ∗ for each pore volume and pore radius. Total porosity was determined from measurements with octane; hydrophilic porosity was estimated from measurements with water. Hydrophobic porosity is a difference between total porosity and hydrophilic porosity.

(

2.2. Electrodes and inert glass spaсer Two types of AC electrodes were applied to investigations, namely Norit powder (Nederland BV Company, the Netherlands) and СН900 textile (Kuraray Company, Japan). Thickness of the electrode formed from the Norit sample was 150 μm, it contained also 6 mass % of a binder (polytetrafluoroethylene). Thickness of the textile was 500 μm. A glass fibrous plate (Glass fibre prefilter, Millipore, Ireland) was used as an inert spacer. Porosity of the inert spacer was 93% and thickness was 300 μm.

)

2.5. Electrochemical measurements in static cell A static cell (without liquid flow) was applied to study electrochemical characteristics of the MEA that was preliminarily impregnated with deionized water. The cell construction was described in detail earlier [45,46]. The cell contained two electrodes of the same type (made of Norit AC) with similar surface areas (from 2.5 to 3 cm2). These electrodes were more preferable than CH900 textile due to their smaller thickness, which provided faster stabilization of electrochemical characteristics. Graphite current collectors provided good contact with the electrodes. The current distribution layer (foil) was located between the electrode and current collectors. The collector was manufactured by compressing the thermoexfoliated graphite powder followed by impregnation with molten paraffin. As opposed to [45,46], MMs were used instead of inert porous spacer. Preliminarily the electrodes were washed in a stream of deionized water and dried at 170 °C under vacuum. Since the volume of water in pores of the electrodes and membrane was extremely low (0.2–0.5 cm3), oxygen has been removed by exposure of the electrode for 10–15 min at the potential close to that of hydrogen evolution. Then the membrane and electrodes were stored in twice deionized water for long time (> 2 days). During the storage, water was periodically stirred and heated up to 40о C in order to remove soluble impurities. Further the membrane and electrodes were inserted into the hermetic teflon cell containing porous graphite current collectors. Compression pressure for the electrochemical groups was about 50 kg cm− 2. The cell was polarized periodically (several times a day) to stabilize electrochemical characteristics, the diapason of voltage was from − 600 to 600 mV. Two types of MEA were investigated, both with the electrodes produced from Norit AC. The film membrane was used for the first MEA (the electrode mass was 24.3 mg). The second MEA contained the fibrous membrane, mass of the electrodes equaled to 30 mg. Just before electrochemical measurements, both the electrodes and MM were impregnated with twice distilled water, electrical conductivity of which was 1–2 μS cm− 1. Electrochemical studies were carried out using a Voltalab–40 potentiostat (Radiometer Analytical, France). Following electrochemical methods were applied to investigations: cyclic voltammetry, galvanostatic technique, and impedance spectroscopy.

2.3. Exchange capacity and electrical conductivity Ion exchange properties of the membranes and electrodes were analyzed as follows [36]. Total cation exchange capacity (Qc) for Hform of the samples was determined by treatment with alkaline solution (mixture of 0.1 M NaOH and 0.1 M NaCl) followed by titration of the effluent with 0.1 M HCl solution. Regarding analysis of anion exchange properties, the membrane was stored in acidic solution (0.1 M HCl and 0.1 M NaCl), and excess of the acid was titrated with alkaline solution. Membrane conductivity (κm) was determined from the value of membrane resistance measured as a real part of impedance. Mercury electrodes were applied to measurements. The electrodes were in direct contact with the membrane [37]. Since membrane conductivity depends on the solution concentration [37,38], measurements were carried out for diluted solutions (NaCl). Preliminarily the membrane was equilibrated with solution of one or another concentration. The κm value, which corresponds to conductivity in deionized water, was estimated by extrapolation of the dependence of κm on concentration. 2.4. Porosimetric measurements The method of standard contact porosimetry (MSCP) [39–43] was used to study the electrodes and membranes. This technique allows us to determine pores in a very wide interval of pore radius (from 1 nm to 100 μm). Moreover, the MSCP gives a possibility to study hydrophilichydrophobic properties, which are intrinsic for porous carbon materials. The MSCP has been recognized by the IUPAC as a technique that allows one to obtain adequate data [44]. Both the test sample and standard samples (for which the porosimetric curves are known) were dried under vacuum conditions at 170° C (standards and electrodes) or 60° С (membranes), further they were weighed separately. Then the sample was placed between two standards, vacuumized, impregnated with water or octane and dried under vacuum conditions. The set was disassembled periodically, its components were weighed. The state of capillary equilibrium was controlled for each point of the pore size distribution. The equilibrium curve of relative moisture content was determined for the test sample. The curve is the dependence of amount of octane or water in the studied sample on the liquid amount in the standards. Further pore size distributions were plotted as described in [37–41]. It was shown [37,38,40] that it is possible to determine the contact wetting angle directly in pores using various working liquids. For this purpose, the pore volume distribution versus r∗ parameter, which is the

2.6. Electrochemical measurements in dynamic cell Disk electrodes made of CH900 textile were applied to investigations in this case. A diameter of each electrode was 8 cm. The cell has been produced by Samsung Electronics Company (South Korea Republic). Construction of the cell was described earlier [4]. Fig. 2 illustrates a CDI device with dynamic cell, solution of 0.005 M KCl passed according to “once through” scheme. The liquid was supplied from the tank by means of a peristaltic pump, further it passed through the valve and pressure sensor, then entered the cell, where deionization or regeneration occurred. After the cell, the liquid flows 3

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

Fibrous membrane Film membrane

-1

κ (S m )

0.6

0.3

0

0

0.05 C (M)

0.10

Fig. 3. Membrane conductivity vs concentration of equilibrium NaCl solutions for film and fibrous MMs. Fig. 2. CDI device for deionization of aqueous solutions: 1 – electrochemical cell, 2 – peristaltic pump, 3 – potentiostat, 4 – ammeter, 5 – conductivity meter 6 – рН meter, 7 – micro processor and display, where time dependencies of electrical conductivity and pH of aqueous solution are displayed.

low and rather high concentrations were used for preliminary impregnation of MM. The measurement errors were explained by heterogeneities of the membranes. Extrapolation of the curves to zero concentration gives a value of specific conductivity. This is due to conductivity of functional groups that is similar to conventional ion exchange membranes [36,47]. The film MM demonstrates higher conductivity than the fibrous material.

successively through the conductometer, pH meter and the second valve into waste container. Before start of the desalination process, the system was flushed with solution to equality of concentrations at the inlet and outlet of the cell. During these processes, argon was purged through the system in order to remove oxygen from pores of the electrodes and spacer. Concentration of the purified solution was estimated according to data of electrical conductivity using reference data. Proportionality of conductivity to concentration was taken into consideration. Since the regeneration stage provides short-circuit of electrodes, the value of current that passed through the cell is important information. Thus, current was indicated during the process.

3.3. Porous structure of the membranes and electrodes Since practically all porous carbon materials contain both hydrophilic and hydrophobic pores [40–42], octane and water were used as measuring liquids. The measurements with octane allow us to determine all types of pores, while water is applied to estimate hydrophilic porosity. The measurement errors were explained by the accuracy of the weights (1–2%). Integral and differential pore size distributions for Norit and CH900 electrodes are shown in Figs. 4, 5. Fig. 6 shows the curve that reflects the distribution of angle of wetting with water for the СН900 sample. The wetting angle was calculated based on data of Fig. 5a according to the algorithm given in [40–42]. Regarding the Norit sample, the wetting angle was found to be close to 90о. Table 2 gives main characteristics of porous structure and hydrophilic-hydrophobic properties of the AC samples. In comparison with the Norit powder, the СН900 textile possesses more expressed hydrophilic properties (the values of its total and hydrophilic surface area are much higher). Since deionization processes occur inside hydrophilic pores (on hydrophilic surface) of AC electrodes, the parameters characterizing hydrophilic pores are important for CDI. Thus, the СН900 textile is more preferable than the Norit sample. Norit powder contains all types of pores: micropores (a radius of which is < 1 nm), mesopores (with radius from 1 to 100 nm according to M.M. Dubinin classification) and macropores (with radius higher than 100 nm). However, the СН900 AC contains micro- and macropores, but almost no mesopores. Charging and discharging of EDL during CDI processes (as a result, desalination and regeneration) occur mainly on hydrophilic surface of micro- and mesopores. Electrolyte inside hydrophilic pores determines ionic conductivity of AC electrodes, this is especially important under rather high current. Thus, increase of hydrophilic porosity causes a growth of ionic conductivity of AC electrodes. In the case of the Norit sample, all types of hydrophilic pores make a comparable contribution to ionic conductivity. Regarding the СН900 sample, the dominant contribution is made by hydrophilic macropores. However, the total volume of hydrophilic pores is higher than that for the Norit AC. According to parameters of porous structure,

3. Results and discussion 3.1. Ion exchange capacity of the membranes and electrodes According to the method described in Section 2.1, the values of cation (Qc) and anion (Qa) exchange capacity were determined both for the AC electrodes and MMs (Table 1). As seen, membranes and electrodes possess both cation and anion exchange ability. This is fundamentally important for the CDI method to obtain pure water. It should be stressed that cation exchange groups dominate in the Norit AC. At the same time, the СН900 sample shows mainly anion exchange properties. Regarding MMs, the film material demonstrates mainly anion exchange ability, cation exchange groups dominate in the fibrous membrane. 3.2. Electrical conductivity of the membranes: Effect of solution concentration Fig. 3 illustrates the dependence of electrical conductivity of film and fibrous MMs on concentration of equilibrium solution. Solutions of Table 1 Cation and anion exchange capacity of electrodes and MMs. Sample АC electrode MM

Norit СН900 Film Fibrous

Qc, mmol g− 1

Qa, mmol g− 1

0.56 0.06 0.36 0.96

0.20 0.70 0.75 0.65

± ± ± ±

0.02 0.01 0.01 0.02

± ± ± ±

0.01 0.01 0.02 0.01

4

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

1.0

2.0

-1 3 -1

3 -1

V (cm g )

1.6

dV/(dlogr) (cm g nm )

Working liquid octane water

1.2 0.8 0.4 0.0

Working liquid octane water

0.8 0.6 0.4 0.2 0.0

0

1

2

3

4

5

0

1

2

3

log r(nm)

logr(nm)

a

b

4

5

Fig. 4. Integral (a) and differential (b) pore size distributions for the Norit AC.

the СН900 electrodes are more attractive for CDI processes. However, thinner electrodes produced from the Norit powder are preferable for measurements in a static cell due to faster stabilization of electrochemical characteristics. Water was used as working liquid for porosimetric investigations of MMs. It was necessary, since ion exchange resins are the membrane constituents: they swell in water and aqueous solutions due to hydration of ion exchange groups. Hydrophilic micro- and mesoporous structure is formed by this manner providing selectivity of the membranes towards counter-ions [39,40,42,43]. Integral and differential porosimetric curves for the fibrous membranes are shown in Fig. 7, main data are given in Table 3. As follows, the fibrous MM is characterized by lower volume of micropores than the film membrane. However, the total porosity (v) and the volume of macropores (r > 1000 nm) are higher for the fibrous material. As a result, hydrodynamic permittivity (K) of this membrane should be also higher according to Kozeni formula [48]:

K=

vr 2 8

Wetting angle (degree)

90

85

80

75

70 0

1

2 logr (nm)

3

4

Fig. 6. Distribution of angle of wetting with water for the СН900 sample.

3.4. Processes in static electrochemical cell (1) Investigations in a static cell (without liquid flow) with MM and AC electrodes were performed in order to ascertain a possibility of charging-discharging processes for the MEA even in pure water, especially to show a contribution of mobile counter-ions of MM to ion transport. The static cell consisted of two electrodes and spacer (see Section 2.5) was firstly washed with distilled water for a long time and then impregnated with water. The measurement errors were caused by the accuracy of weights (1–2%), where masses of the electrodes and membranes were determined. The measurement instrumentation also

which has been obtained for capillary model (straight cylindrical capillary of the same radius were assumed). Since high hydrodynamic permittivity is necessary to provide significant efficiency of CDI processes, the fibrous membrane can be considered as more attractive material than the film MM despite its lower conductivity. MMs that were described in [31] are related namely to film materials, they can hardly be used for CDI due to low hydrodynamic permittivity.

8

4

V (cm3g-1)

dV/d(logr) (cm3g-1nm-1)

Working liquid octane water

3

2

1

Working liquid 6

octane water

4

2

0

0 0

1

2

3

4

0

5

1

2

3

logr(nm)

logr(nm)

a

b

Fig. 5. Integral (a) and differential (b) pore size distributions for the CH900 AC.

5

4

5

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

Table 2 Porosimetric measurements of AC samples. Sample

Norit СН900

Pore volume, cm3 g− 1

Specific surface area, m2 g− 1

Wetting angle, degree

Total

Hydrophilic

Hydrophobic

Total

Hydro-phobic

1.65 ± 0.01 3.56 ± 0.02

0.60 ± 0.01 2.95 ± 0.02

1.05 ± 0.01 0.61 ± 0.02

1260 ± 15 1520 ± 14

567 ± 10 870 ± 16

made its contribution. In general, the error bars reached 3–5%. As found, electrochemical characteristics of the cell reached constant values in one month even for the thinner electrode prepared from the Norit powder. Until stabilization is achieved, capacitance increases gradually, electrical resistance also grows slightly. Slower stabilization made impossible usage of textile AC electrodes for measurements in a static cell. It is assumed that the stabilization process was limited by slow surface diffusion of hydrated counter-ions (both cations and anions) through the bulk of AC electrodes. All data, which are given below, were obtained after finishing of stabilization of chemical characteristics. Fig. 8 illustrates cyclic curves that reflect dependencies of cell capacitance on voltage under different scan rates. The capacitance values were calculated as Iω− 1, where I is the current, ω is the scan rate. Thus, the curves reflect voltammetric dependencies. As shown, the currentvoltage dependencies are characterized by extremes. A shape of the curves is sufficiently different from that for doublelayer capacitor with AC electrodes and concentrated electrolyte (in this case, the shape is close to rectangular [10,11]). This indicates the influence of the MM on electrode charging. The simplest explanation of shape of the curves is as follows: charging of the EDL is complicated by high electrical resistance, since only counter-ions of functional groups provide surface conductivity in pores of AC and MM [29]. Pure water contains very low amount of ionic species, thus, their contribution to surface conductivity can be neglected. Maximal value of the capacitance for AC electrodes (CEDL) reached 56–66 F g− 1. These values have been obtained by dividing the average current by the electrode mass. The calculations were performed for the plateau region and for the slowest scan rate. The impedance spectrum shows gradual increase of capacitance in the region of low frequencies (Figs. 9, 10). It is evidently that the ratedetermining stage of ion transport is diffusion of hydrated counter-ions under these conditions. In the case of the film membrane, the average value of specific conductivity is 0.15–0.2 S m− 1 at 104 Hz. This is close to data obtained for the film MM using the method that involves mercury electrodes (see Fig. 3). The main feature of electrochemical behavior of the cell with the fibrous MM is gradual decrease of resistance

89 ± 9 77 ± 11

Table 3 Porosimetric measurements of MM. MM

Total porosity,cm3 cm− 3

Microporosity, cm3 cm− 3

Specific surface area, m2 g− 1

Film Fibrous

0.39 ± 0.004 0.53 ± 0.006

0.05 ± 0.001 0.15 ± 0.002

147 ± 2 310 ± 4

and growth of capacity, which are observed during cycling. Fig. 11 shows the curves obtained during galvanostatic cycling of the cell containing fibrous membrane at 50 μА cm− 2 (5 mA g− 1). The measurements were carried out after 1 and 30th days of cycling. Thus, maximal values of specific capacitance for both AC electrodes are rather close to each other and are in the diapason from 56 to 66 F g− 1. Since hydrophilic specific surface area of the Norit AC is 305 m2 g− 1, the value of specific capacitance of these electrodes per unit of hydrophilic surface is rather high (18.4–21.6 F cm− 2). The measurements were carried out in pure water (the effect of diffuse EDL was practically excluded). In this case, it is possible to talk only about Helmholtz EDL. Based on obtained results, following operation mechanism for static cell containing pure water has been proposed (Fig. 12). When electric field is applied, counter-ions (cations) migrate according to relay mechanism through only one AC electrode due to surface conductivity. The species move towards MM, then migrate through the membrane towards the second electrode. At the same time, anions migrate through the counter-electrode towards MM. Then anions move through the membrane towards the electrode that is a source of cations. The EDL of both electrodes is charged by this manner. If counter-ions in the electrodes and MM are different, ion transport through the pathway of electrode-membrane-electrode is accompanied by ion exchange. Inaccuracy of the electrochemical curves obtained for the static cell shown on Figs. 8, 9, 10, 11, 12 was ± 3–5%; it was caused by inaccuracy of electronic scales and depended on absolute weights of electrodes and membranes.

0.6

0.6 Membrane

3 -1

V (cm g )

3 -1

film fibrous

0.4

film fibrous

-1

dV(dlogr) (cm g nm )

Membrane

0.2

0.4

0.2

0.0

0.0 0

1

2 3 log r (nm)

4

0

5

1

2

3

log r (nm)

a

b

Fig. 7. Integral (a) and differential (b) pore size distributions measured with water for mosaic membranes.

6

4

5

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

0.8

1500 1000 500

0.0

C (F)

Capacitance (F)

0.4

-0.4

-1

Scan rate (mV s ) 20 2 1 0.1

-0.8 -1.2 -800

-400

0

400

0 -1

-500

Scan rate (mV s ) 5 2 0.5 0.1

-1000 -1500 -400

800

0

400

U (mV)

U (mV)

a

b

Fig. 8. Capacitance of static cell as a function of voltage at different scan rate. Membranes: film (a), fibrous (b).

600

80

400

0.01 Hz

60

U (mV)

40

//

Z (Ohm)

200 1 day 30 days

0 -200

20

-400 0

-600 0

40

80

120

0

Z/(Ohm)

1000

2000 t (s)

3000

4000

Fig. 11. Galvanostatic curves obtained at 150 μA for the cell containing fibrous membrane, the curves show a growth of capacitance after start of cycling in water (time is shown in the legend).

Fig. 9. Nyquist diagram for the cell containing film MM. The measurements were performed at 0 V in the frequency interval of 0.01–105 Hz. Voltage amplitude was 50 mV.

300 8th day 24th day

0.001 Hz

//

Z (Ohm)

200

100 0.01 Hz

0 0

100

200 / Z (Ohm)

300

Fig. 10. Nyquist diagram for the cell containing fibrous membrane. Legend shows time after start of cycling in water. Frequency diapason (Hz) was: 1 - 0.01–104, 2 - 0.001–104, voltage amplitude was 50 mV, the measurements were performed at 0 V.

3.5. Processes in dynamic electrochemical cell Fig. 12. Scheme of MEA that involves AC electrodes and MM impregnated with pure water.

Electrical conductivity of the solution containing KCl was measured after passage though the dynamic electrochemical cell. The measurements were performed during desalination and regeneration stages according to the method described in Section 2.6. Very low concentrated solutions passed through the cell; the fibrous MM was used as a spacer due to its high hydrodynamic permittivity (see Section 3.3). In

comparison with this material, the film MM possesses higher hydrodynamic resistance, so pumping of the solution through the film membrane was impossible and it was not used in the study. The measurement errors were caused by the inaccuracy of the conductometer 7

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

0.04

0.006

0.03 1.4 V 1.7 V

C(M)

C (M)

1.4 V 1.7 V 2.0 V 0.004

2V

0.02

0.01

0.00 2000

0.002 0

1000

2000

3000

4000

t (s)

t (s)

a

b

5000

Fig. 13. Concentration of KCl solution as a function of time at 1.4–2.0 V for desalination (a) and concentration (b) stages. Initial concentration of the solution was 0,005 М, the flow velocity was 5 cm 3 min− 1.

than the WD magnitudes. This is due to release of energy as explained above (see Section 1). Moreover, all energy magnitudes demonstrate growth with increasing in voltage. The WCDI values are lower for the MEA containing fibrous MM. Fig. 15 shows the effect of voltage on resulting energy per 1 mol of salt (desorbed from the electrodes). As seen, specific energy consumptions for the cells that contains both glass spacer and fibrous MM decrease with decreasing in cell voltage. The curves show plateau at U ≤ 1.4 V. The energy consumptions for the cell containing MM are lower than those for the cell with glass spacer. This is due to mobile counter-ions (cations and anions) of functional groups of the MM and electrodes. These counter-ions provide ion transport, when the content of ions in water inside pores is extremely low (see the explanation to Fig. 13). Alternately, the glass separator contains no mobile ions. In order to compare correctly two types of spacers (MM and glass), it is necessary to take into account different thickness and porosity of the spacers: the thickness and porosity of the fibrous MM are 450 μm and 53% respectively, while those of the glass spacer are 300 μm and 93%. As seen, hydrodynamic permittivity along the glass spacer is higher in comparison with MM (see Eq. (1)). Thus, it is possible to expect higher difference of energy consumptions (under conditions of similar cell voltage and hydrodynamic permittivity) than that shown in Fig. 15. It is possible to confirm that minimal energy consumptions and almost maximal desalination degree are reached at 1.4 V for the cell containing fibrous MM (initial concentration of KCl solution is 0.005 M, solution velocity is 5 cm3 min− 1). Thus, the sufficient advantage of MM over inert spacer for obtaining of pure water can be considered proven. Ceramic [49–51] or inert polymer membranes modified with

(2–5%). Deionization of the solution occurred, when voltage was applied to the cell. The regeneration stage started after closure of the electrical circuit. Fig. 13 shows concentration of KCl solution vs time for deionization and regeneration stages. The interval between initial time of these stages was 2500 s. As shown from Fig. 13a, increasing in voltage causes rather slight growth of desalination degree. Similar effect is observed for the regeneration stage (Fig. 14b). During the regeneration stage, energy is partially transferred to the CDI stack. Thus, energy for desalination is compensated in part. The resulting energy (WCDI) is expressed as:

WCDI = Wd − Wc

(2)

where WD is the energy consumed during deionization stage, Wс is the energy released during regeneration. It is the WCDI energy that must be taken into account for the operation of the CDI stack, since there is a gain due to energy during the regeneration stage. The energy of desalination and regeneration was calculated as:

W=U

∫t

t2

1

Idt

(3)

where U is the cell voltage. The values of energy consumptions under different voltage were compared for the MEAs that contain fibrous MM (Fig. 14a) and porous glass separator (Fig. 14b). These data for deionization and regeneration stages as well as the magnitudes of resulting energy were collated. As follows from Fig. 14, the resulting energy is sufficiently lower

Fig. 14. Energy as a function of cell voltage for deionization (1) and concentration (2) stages, resulting energy (3). The cell contained fibrous MM (a) and porous glass separator (b). Initial concentration of the solution was 0,005 М, the flow velocity was 5 cm3 min− 1.

8

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

measurements of electrochemical characteristics in a static cell (without water flow). Cyclic voltammetric curves have been obtained for the static cell containing MM and AC electrodes, which were impregnated with twice deionized water (electrical conductivity of water was 2 μS cm− 1), Rather high values of specific capacitance (56–66 F g− 1) have been obtained for the AC electrodes. Since the measurements were performed in pure water, the effect of diffusive EDL is practically excluded. In this case, it is possible to talk about Helmholtz EDL. As found from the measurements in a dynamic electrochemical cell (with a solution flow), specific energy consumptions for deionization of very diluted solutions are lower for the fibrous MM comparing with the inert glass spacer. The optimal cell voltage (1.4 V), which is necessary to achieve maximal deionization degree and minimal energy consumptions, was found. Thus, the idea about optimal composition of the MEA for obtaining of pure water has been confirmed experimentally. The MEA has to contain MM and AC electrodes. As shown, the requirements to the membrane are high porosity and large volume of macropores in order to provide sufficient hydrodynamic permittivity. Highly dispersive AC electrodes have to possess both cation and anion exchange capacity. The other important requirement is high surface conductivity. It should be stressed that MM application to CDI processes of obtaining pure water is one way to solve the problem of their efficiency and minimizing of the energy cost.

Fig. 15. Specific resulting energy of CDI as a function of cell voltage. The cell contains porous glass spacer (1) and fibrous MM (2). Initial concentration of the solution was 0,005 М, the flow velocity was 5 cm3 min− 1.

inorganic ion-exchangers (hydrated oxides of multivalent metals) [52,53] could be considered as an alternative to polymer MMs. The modifiers possess amphoteric properties in neutral media [54]. However much higher energy consumptions and lower desalination degree are expected due to insignificant electrical conductivity [50] and low charge selectivity [49] of the composites due to considerable thickness of ceramic membranes (about 1 mm) [49–51]. Another reason is rather large incorporated particles of the modifier [52], though size of oxide particles is smaller comparing with phosphate of the same metal [53,55]. As a result, large pores between particles of the modifier provide functional properties of the composites (low transport numbers of counter-ions [49], no rejection of species with mass of 35,000 Da [53]). Similarly to mentioned materials, polymer MMs are able to exchange both cations and anions. An important advantage of the MMs over inorganic membranes (particularly glass spacers) is high ion exchange capacity that provides significant electrical conductivity in pure water and decreases energy consumptions. As a result, application of MMs to CDI processes is more attractive from the economical point of view. In fact, the MMs can be related to practically non-alternative materials for CDI processes for obtaining of pure water.

Acknowledgements The work was performed within the framework of the project “Fundamental aspects of capacitive deionization of aqueous solutions” supported by the Russian Foundation of Basic Research (Grant no. 1703-0009). References [1] Y. Oren, Capacitive deionization (CDI) for desalination and water treatment—past, present and future (a review), Desalination 228 (2008) 10–29. [2] F.A. AlMarzooqi, A.A. Al Ghaferi, I. Saadat, N. Hilal, Application of capacitive deionisation in water desalination: a review, Desalination 342 (2014) 3–15. [3] H. Strathman, Ion-Exchange Membrane Processes: Their Principle and Practical Applications, Balaban Desalination Publications, Hopkinton, MA, 2016. [4] Yu.M. Volfkovich, D.A. Bograchev, A.A. Mikhalin, A.Yu. Rychagov, V.E. Sosenkin, D. Park, Capacitive deionization of aqueous solutions: modeling and experiments, Desalin. Water Treat. 69 (2017) 130–141, http://dx.doi.org/10.5004/dwt.2017. 0469. [5] T.J. Welgemoed, C.F. Schutte, Capacitive deionization technology: an alternative desalination solution, Desalination 183 (2005) 327–340. [6] L. Alvarado, A. Chen, Electrodeionization: principles, strategies and applications, Electrochim. Acta 132 (2014) 583–597. [7] J.-H. Song, K.-H. Yeon, S.-H. Moon, Effect of current density on ionic transport and water dissociation phenomena in a continuous electrodeionization (CEDI), J. Membr. Sci. 291 (2007) 165–171. [8] Yu.S. Dzyaz'ko, L.M. Rozhdestvenskaya, A.V. Pal'chik, Recovery of nickel ions from dilute solutions by electrodialysis combined with ion exchange, Russ. J. Appl. Chem. 75 (2005) 414–421. [9] R. Zhao, P.M. Biesheuvel, A. van der Wal, Energy consumption and constant current operation in membrane capacitive deionization, Energy Environ. Sci. 5 (2012) 9520–9527. [10] B.E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals and Technological Applications, Springer, New York, 1999. [11] V.S. Bagotsky, A.M. Skundin, Yu.M. Volfkovich, Electrochemical Power Sources: Batteries, Fuel Cells, Supercapacitors, John Wiley & Sons Inc., New Jersey, 2015. [12] C.J. Gabelich, T.D. Tran, I.H. Suffet, Electrosorption of inorganic salts from aqueous solution using carbon aerogels, Environ. Sci. Technol. 36 (2002) 3010–3019. [13] H. Li, T. Lu, L. Pan, Y. Zhang, Z. Sun, Electrosorption behavior of graphene in NaCl solutions, J. Mater. Chem. 19 (2009) 6773–6779. [14] G. Wang, Q. Dong, Z. Ling, C. Pan, C. Yu, J. Qiu, Hierarchical activated carbon nanofiber webs with tuned structure fabricated by electrospinning for capacitive deionization, J. Mater. Chem. 22 (2012) 21819–21823. [15] H. Li, L. Pan, T. Lu, Y. Zhan, C. Nie, Z. Sun, A comparative study on electrosorptive behavior of carbon nanotubes and graphene for capacitive deionization, J. Electroanal. Chem. 653 (2011) 40–44. [16] M.S. Gaikwad, C. Balomajumder, Polymer coated capacitive deionization electrode for desalination: a mini review, Electrochem. Energy Technol. 2 (2016) 1–5. [17] Y. Liu, C. Nie, X. Liu, X. Xu, Z. Sun, L. Pan, Review on carbon-based composite

4. Conclusions The CDI process using the MEA with MMs that contain both cation and anion exchange regions was investigated. The membranes were used instead of conventional glass spacer, which possesses no ion exchange ability, in order to decrease energy consumptions. Ionic conductivity and ion exchange capacity were determined both for the film and fibrous MM, which were made of polyethylene and polyacrylonitrile fibers respectively. Porous structure of the membranes was researched with MSCP. As shown, the fibrous MM is characterized by higher total porosity and larger volume of macropores in comparison with the film membrane. Thus, hydrodynamic permittivity of the fibrous MM is higher. This gives a possibility to use this membrane for CDI processes under dynamic conditions. AC porous electrodes, which are characterized by high concentration of surface functional groups, were applied to investigations. The groups provide high surface conductivity that is principally important for obtaining of pure water with CDI methods. MSCP was also used for research of porous structure and hydrophilic-hydrophobic properties of AC electrodes, effect of which on CDI processes was shown. The CH900 sample is characterized by higher hydrophilic porosity than the Norit powder. It means that the textile electrode is preferable for CDI, since deionization and concentrating occur namely in hydrophilic pores. However, the thinner Norit electrode is more attractive for 9

Desalination 426 (2018) 1–10

Y.M. Volfkovich et al.

[39] Yu.M. Volfkovich, V.S. Bagotzky, The method of standard porosimetry 2. Investigation of the formation of porous structures, J. Power Sources 48 (1994) 339–348. [40] Yu.M. Volfkovich, A.V. Sakars, A.A. Volinsky, Application of the standard porosimetry method for nanomaterials, Int. J. Nanotechnol. 2 (2005) 292–302. [41] N.A. Kononenko, N.P. Berezina, Yu.M. Vol'fkovich, E.I. Shkol'nikov, I.A. Blinov, Investigation of ion-exchange materials structure by standard porosimetry method, Zh. Prikl. Khim. 58 (1985) 2199–2203. [42] Yu.M. Volfkovich, A.N. Filippov, V.S. Bagotsky, Structural Properties of Porous Materials and Powders Used in Different Fields of Science and Technology, Springer Publisher, 2014. [43] N.A. Kononenko, M.A. Fomenko, Yu.M. Volfkovich, Structure of perfluorinated membranes investigated by method of standard contact porosimetry, Adv. Colloid Interf. Sci. 222 (2015) 425–435. [44] J. Rouquerol, G. Baron, R. Denoyel, H. Giesche, J. Groen, P. Klobes, P. Levitz, A.V. Neimark, S. Rigby, R. Skudas, K. Sing, M. Thommes, K. Unger, Liquid intrusion and alternative methods for the characterization of macroporous materials (IUPAC Technical Report), Pure Appl. Chem. 84 (2011) 107–136. [45] Yu.M. Volfkovich, D.A. Bograchev, A.A. Mikhailin, V.S. Bagotzky, Supercapacitor carbon electrodes with high capacitance, J. Solid State Electrochem. 18 (2014) 1351–1363. [46] Y.M Vol’fkovich, A.A. Mikhalin, D.A. Bograchev, V.E. Sosenkin, Carbon electrodes with high pseudocapacitance for supercapacitors, Russ. J. Electrochem 48 (2012) 424–433. [47] T. Badessa, V. Shaposhnik, Electrical conductance studies on ion exchange membrane using contact-difference method, Electrochim. Acta 231 (2017) 453–459. [48] S. Kozeny. J., Uber die kapillare Leitung des Wassers im Boden (Aufstieg Versickerung und Anwendung auf die Bewasserung. Sitz. Ber. Akad. Wiss, Wien, Math. Nat.(Abt. Iia)), Vol. 11 (1927), p. 271. [49] Y.S. Dzyazko, V.N. Belyakov, S.L. Vasilyuk, N.V. Stefanyak, Anion-exchange properties of composite ceramic membranes containing hydrated zirconium dioxide, Russ. J. Appl. Chem. 79 (2006) 769–773. [50] Y.S. Dzyazko, A.S. Rudenko, Y.M. Yukhin, A.V. Palchik, V.N. Belyakov, Modification of ceramic membranes with inorganic sorbents. Application to electrodialytic recovery of Cr(VI) anions from multicomponent solution, Desalination 342 (2014) 52–60. [51] M.C. Martı-Calatayud, M. Garcıa-Gabaldon, V. Perez-Herranz, S. Salesb, S. Mestre, Ceramic anion-exchange membranes based on microporous supports infiltrated with hydrated zirconium dioxide, RSC Adv. 5 (2015) 46348–46358. [52] R. Pang, X. Li, J. Li, Z. Lu, X. Sun, L. Wang, Preparation and characterization of ZrO2/PES hybrid ultrafiltration membrane with uniform ZrO2 nanoparticles, Desalination 332 (2014) 60–66. [53] V.G. Myronchuk, Yu.S. Dzyazko, Yu.G. Zmievskii, A.I. Ukrainets, A.V. Bildukevich, L.V. Kornienko, L.M. Rozhdestvenskaya, A.V. Palchik, Organic-inorganic membranes for filtration of corn distillery, Acta Periodica Technologica 47 (2016) 153–165. [54] C.B. Amphlett, Inorganic Ion Exchangers, Elsevier, Amsterdam, 1964. [55] Y.S. Dzyazko, L.M. Rozhdestvenskaya, Y.G. Zmievskii, A.I. Vilenskii, V.G. Myronchuk, L.V. Kornienko, S.L. Vasilyuk, N.N. Tsyba, Organic-inorganic materials containing nanoparticles of zirconium hydrophosphate for baromembrane separation, Nanoscale Res. Lett. 10 (2015) 64, http://dx.doi.org/10.1186/ s11671-015-0758-x.

materials for capacitive deionization, RSC Adv. 5 (2015) 15205–15225. [18] D. Zhang, X. Wen, L. Shi, T. Yan, J. Zhang, Enhanced capacitive deionization of graphene/mesoporous carbon composites, Nano 4 (2012) 5440–5446. [19] M.T.Z. Myint, J. Dutta, Fabrication of zinc oxide nanorods modified activated carbon cloth electrode for desalination of brackish water using capacitive deionization approach, Desalination 305 (2012) 24–30. [20] R. Broséus, J. Cigana, B. Barbeau, C. Daines-Martinez, H. Suty, Removal of total dissolved solids, nitrates and ammonium ions from drinking water using chargebarrier capacitive deionisation, Desalination 249 (2009) 217–223. [21] S.-J. Seo, H. Jeon, J.K. Lee, G.-Y. Kim, D. Park, H. Nojima, J. Lee, S.H. Moon, Investigationon removal of hardness ions by capacitive deionization (CDI) for water softening applications, Water Res. 44 (2010) 2267–2275. [22] H. Li, L. Zou, L. Pan, Z. Sun, Using graphene nano-flakes as electrodes to remove ferric ions by capacitive deionization, Sep. Purif. Technol. 75 (2010) 8–14. [23] Y.A.C. Jande, W.S. Kim, Predicting the lowest effluent concentration in capacitive deionization, Sep. Purif. Technol. 115 (2013) 224–230. [24] Y.A.C. Jande, W.S. Kim, Desalination using capacitive deionization at constant current, Desalination 329 (2013) 29–34. [25] M.D. Andelman, Charge barrier flow-through capacitor, CA Patent 2444390, 2002. [26] J.-B. Lee, K.-K. Park, H.-M. Eum, C.W. Lee, Desalination of a thermal power plant wastewater by membrane capacitive deionization, Desalination 196 (2006) 125–134. [27] P.M. Biesheuvel, A. van der Wal, Membrane capacitive deionization, J. Membr. Sci. 346 (2010) 256–262. [28] M.A. Anderson, A.L. Cudero, J. Palma, Capacitive deionization as an electrochemical means of saving energy and delivering clean water. Comparison to present desalination practices: will it compete? Electrochim. Acta 55 (2010) 3845–3856. [29] Yu.M. Volfkovich, A.A. Mikhalin, A.Y. Rychagov, Surface conductivity measurements for porous carbon electrodes, Russ. J. Electrochem. 49 (2013) 594–598. [30] Yu.M. Volfkovich, D.A. Bograchev, A.Yu. Rychagov, V.E. Sosenkin, M.Yu. Chaika, Supercapacitors with carbon electrodes. Energy efficiency: modeling and experimental verification, J. Solid State Electrochem. 19 (2015) 1–9. [31] T. Xu, Ion exchange membranes: state of their development and perspective, J. Membr. Sci. 263 (2005) 1–29. [32] G.Z. Nefedova, Z.G. Klimova, G.S. Sapozhnikova, Ion-Exchange Membranes. Granules. Powders, Catalogue, Moscow, (1977) (in Russian). [33] N.P. Berezina, N.A. Kononenko, Yu.M. Volfkovich, Yu.G. Freidlin, L.G. Chernoskutova, Phisico-chemical properties of aniono-cationo exchange membranes of mosaic structure, J. Soviet Electrochem. 29 (1989) 912–915. [34] N.B. Fedorchenko, M.M. Kardash, Effect of the textile structure of polyacrylonitrile fibres on formation of a three-dimensional polymer matrix network in ‘Polikon’ materials, Fibre Chem. 36 (2004) 263–266. [35] M.M. Kardash, N.B. Fedorchenko, O.V. Epancheva, Structural features of composite chemisorption fibre materials from polycondensation filling, Fibre Chem. 34 (2002) 466–469. [36] N.P. Berezina, N.A. Kononenko, O.A. Dyomina, N.P. Gnusin, Characterization of ion-exchange membrane materials: properties vs structure, Adv. Colloid Interf. Sci. 139 (2008) 3–28. [37] Membranes and Membrane Technologies, A.B. Yaroslavtsev (Ed.), Nauchnii Mir, Moscow, 2013 (in Russian). [38] N.P. Gnusin, N.P. Berezina, O.A. Dyomina, N.A. Kononenko, Physicochemical principles of testing ion-exchange membranes, Russ. J. Electrochem. 32 (1996) 154–163.

10