1 Introduction

Electrodialysis (ED) and electrodeionization (EDI) processes are used to remove dissolved ions from aqueous solutions. These technologies rely on the electric potential difference between a pair of electrodes (anode and cathode), which is obtained by applying a constant electric voltage to the electrodes. ED/EDI is mainly used for the production of ultrapure water (UPW) [1,2,3,4,5,6]. High-purity water is extensively used in various industries, including chemical and pharmaceutical production, laboratories, semiconductor production and the power industry [2, 7,8,9]. The process involves the passage of ions through a selective ion exchange membrane. The cation and anion exchange membranes are arranged alternately in the ED/EDI electrodialyser chambers. Water flows through these chambers, and under the external electric field, ions are removed from the water, creating purified water and a concentrate of ions. The advantage of this method of water purification, in comparison with other methods, is that it produces a relatively small amount of wastewater (approximately 20%) relative to the volume of treated water [10,11,12,13,14,15,16,17,18,19].

The ion exchange resins are also used, in EDI systems (in purified water chambers) allowing for high-quality ion separation and increased energy efficiency. When dilute compartments are filled with a mixed ion exchange resin, they behave as a conductor due to the presence of functional groups that serve as a bridge between the ion exchange membranes [9, 10, 12, 20,21,22]. The advantage of this solution is that it allows the continuous resin regeneration without the need for harsh chemicals like hydrochloric acid and sodium hydroxide [9]. Ion exchange resins are continuously regenerated by H+ and OH− ions produced as a result of water dissociation. This makes EDI an environmentally friendly process [9, 18, 20, 23,24,25,26,27,28,29,30]. However, the key role in electrodialysis processes, regardless of their variety (ED, EDI, etc.), is played by electrodes. Their physical and chemical properties affect the efficiency of the water purification process and the lifetime of the device. Carbon (graphite) or metal materials such as stainless steel, aluminium, lead, copper and titanium are commonly employed as electrodes due to their low cost, good electrical conductivity and chemical inertness [31, 32]. However, the main limitation in use of traditional carbon and metal electrodes is the problem of anode oxidation and its electrochemical dissolution, which in turn leads to the degradation of the electrode material and decrease in process efficiency. Therefore, currently used anodes are made of platinum or titanium, coated with a mixture of noble metal oxides (NMO), such as ruthenium and iridium, or with platinum [9, 25, 31, 33,34,35,36,37]. It is expensive but effective solution, so these materials are used only for anodes, while cathodes are made of cheaper material such as stainless steel, graphite or titanium [32, 38]. Unfortunately, such solutions have their disadvantages. Foremost, it is the inability to change the polarity, which aids the cleaning of the electrodialyser from ions remaining in the membranes or on the surfaces of the electrodes. Secondly, the reduction of production costs of water electrodialysis devices based on this type of electrodes comes down to the use of one pair of cells, placed in the electrodialyser, regardless of its size and efficiency. Such a solution is more energy-intensive than multi-electrode solutions, because electricity consumption in a multi-electrode solution can be reduced by lowering the voltage on the cell [31]. Therefore, solutions are currently being sought that will ensure high durability and electrochemical stability and will allow for low production costs [39]. Carbon materials based on graphene not on graphite can be such solution. In the last few years, a lot of work has been dedicated to the use of reduced graphene oxide (rGO) to produce electrodes used in capacitive deionization (CDI), mainly using its physical and chemical stability [40,41,42,43,44,45,46,47,48,49,50,51,52,53]. However, there are currently no literature reports on the use of rGO or its derivatives in the electrodialysis (ED) or electrodeionization (EDI) of water.

A new material for electrodes used in water electrodialysis (ED) is proposed, in this work. This material is a composite based on carbon fibres bonded with epoxy resin, covered with a coating of nanostructured reduced graphene oxide (n-rGO). Electrodes made of this material, thanks to the use of rGO, the flakes bonded by nitrogen bridges [54], turned out to be electrochemically stable and durable in operation, as demonstrated in the tests. It has been shown that this type of electrodes can be successfully used for electrodialysis of water in multi-electrode devices, using the cyclic process of cleaning the electrodialyser through reverse polarity.

2 Materials and methods

2.1 Preparation of the composite graphene electrode based on n-rGO

The graphene composite electrodes were made according to patent no. PL 236678 [55]. For the construction of electrodes, a carbon pre-preg (CFRP—carbon fibre-reinforced polymer) with a—bidirectional—straight weave was used, with a weight of 200 g m−2, resin saturation—42% by weight, from Engineered Cramer Composites (ECC), and a water suspension of graphene oxide (concentration 0.2% by weight) from Advanced Graphene Products SA. The carbon pre-preg was prepared for coating with rGO suspension by mechanically damaging the surface of the carbon fibres with a scraper blade. This was done in order to increase the active spots responsible for increasing the adhesion of the coating to the composite. The CFRP was then oxidized in concentrated sulfuric acid and an aqueous solution of saturated potassium permanganate, each for about 10 min. After the fibre surface was oxidized, the mat was rinsed with deionized water and coated with hydrazine. It was then immersed in a suspension of graphene oxide to which 1% by volume was added hydrazine. Hydrazine was added to the GO suspension in an amount to ensure a significant, but not complete, reduction of the graphene oxide, which runs in parallel with the pillaring process. Functional groups on the edges of GO (–COOH and –COH) were used as active centres for the formation of pillars based on C–N bonds in the structure of the coating material. Similar active centres on the surface of CFRP fibres were obtained by partial removing the epoxy resin and then oxidizing their surface, as described above.

After evaporation of water from the n-rGO suspension (NTP conditions), the composite was dried at 60 °C for about 60 min. The final stage of production was thermal treatment. First, polymerization of the epoxy resin was performed. Polymerization was carried out in a vacuum sleeve at a temperature of 120 °C and heating the electrodes for 120 min. Then, after the polymerization process, the electrodes were heated in a hydrogen atmosphere at 130 °C for 20 min. This process removed excess of oxygen groups and improved the electrical conductivity of the electrodes.

A high-strength electrode with a nanostructured n-rGO coating of high cohesion and coherence with the CFRP support frame was produced in this way.

2.2 Research of the chemical structure of the composite

2.2.1 FTIR spectroscopy

FTIR analysis was conducted using Nicolet iS50 spectrometer at a 4 cm−1 resolution in the range of 4000–600 cm−1. The DTGS KBr beam splitter was used. The measurement was performed in the absorbance mode, using a DRIFT-type diffusion adapter. The angle of incidence of the beam was 20°, and the number of scans was 128. These studies aimed to determine the chemical structure of the produced n-rGO coating and changes in the structure of this material arising as a result of electrodialysis and during electrodialyser cleaning after polarization reversal.

2.2.2 Raman spectroscopy

Raman spectroscopic studies were conducted to determine the amount of oxygen groups present in the nanostructured reduced graphene oxide (n-rGO), on the electrodes, before, after work and after polarity reversal. These studies were performed using a Renishaw inVia Raman spectroscope, using a 532 nm laser wavelength. The laser power on the sample was 0.3 mW, exposure time was 10 s, and the applied magnification was 50 times.

2.3 Electrode resistivity tests

The surface resistance of the produced electrodes was carried out using a comb electrode in a two-electrode system, at a temperature of 24 °C and humidity of 54%. The average electrode resistance measurement calculated from 12 measurements is 14.0 ± 1.2 Ω. After considering the dimensions of the tested object and conversion, it gives a surface resistivity of 6.2 × 10−5 Ω m.

2.4 Surface morphology studies and elemental composition analysis

The surface morphology studies of the n-rGO-coated composite electrodes were carried out using a scanning microscope (SEM) from JEOL JSM-6610LV equipped with a X-Max80 detector for examining the chemical composition using the EDX (EDS) method and Aztec software. The tests were carried out in the SE mode, and an accelerating voltage = 5 kV was used. The tests were carried out to determine changes in the morphology of the electrode surfaces and their elemental composition resulting from operation in an electrodialyser.

2.5 Electrochemical research

The graphene composite electrode material was electrochemically characterized by cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and potentiostatic polarization techniques. The equipment used for the electrochemical measurements consisted of a PAR 273A potentiostat (EG&G Princeton Applied Research Company) and the Solartron 1255A frequency response analyser connected to a computer with CorrWare 2.9, CorrView 2.9, ZPlot 2.9 and ZView 2.9 software (Scribner Associates, Inc.). All the electrochemical measurements were carried out in a conventional three-electrode cell where a graphene composite electrode (8.0 cm2) was used as a working electrode, a platinum sheet as a counter electrode and a saturated calomel electrode in NaCl (SCE) with Luggin capillary as a reference electrode. The electrochemical measurements were carried out in NaCl solution with a concentration of 1 mol L−1. All the measurements were performed in triplicate. The cyclic voltammetry curves were registered in the potential range from − 2.0 to + 2.0 V with the scan rate 20 mV s−1. The impedance characteristics were registered at an open-circuit potential (OCP) within the frequency range from 0.1 to 105 Hz, and the amplitude was 20 mV. The potentiostatic polarization of the graphene composite electrode was performed in potential sequences (+ 2.0 V for 900 s and then − 2.0 V for 15 s) simulating the real work conditions of the ED system.

2.6 Multi-electrode electrodialyser for water deionization

2.6.1 Materials

MEGA a.s. (Czech Republic) RALEX® CMHPP cation exchange membranes (CEM) and AMHPP anion exchange membranes (AEM) were used as ion exchange membranes in the laboratory electrodialyser. For desalination tests, tap water with a total hardness of 125 ppm (municipal water supply in Lodz, Poland) was used as a feed solution. The feed solution was prepared in a tank to avoid changes in water composition.

2.6.2 Stack configuration

The electrodialyser consisted of 4 electrodes, n-rGO coated, placed in 4 chambers, 3 CEM, 3 AEM and 3 water treatment cells. Electrodes of dimensions 170 mm × 80 mm were used. Voltage was applied to the electrodes through the electrode strips led outside the module. The cell for water treatment was made of plastic mesh embedded outside in a rubber seal, and the width of the dilute cell was 4 mm. Figure 1 shows a schematic illustration of the ED stack. The electromembrane system was placed in a measurement and control station, which consisted of a water pump for circulating purified water, solenoid valves, a controller, a DC power supply for the electrodes and a water reservoir with a capacity of 3 L.

Fig. 1
figure 1

Configuration of multi-electrode electrodialyser for water deionization with water flow indication

Full size image

2.6.3 Operation of the ED system

A schematic diagram of ED system operation is shown in Fig. 2. The purified water was circulated in a closed circuit from the reservoir to the electrodialyser and back to the reservoir at a rate of 400 ml min−1 set by the pump. Water was flushed into the electrodialyser through two inlets to each purifying cell, flew parallel through three cells and flowed out of each cell through two holes back to the reservoir (Fig. 1). The water flow through the electrolyte chambers was 12 ml min−1, which allowed to obtain an 80% recovery rate. The concentrate was discharged into the wastewater. Water purification was carried out at a constant voltage of 12 V. The frequency of polarity reversal was set to 15 min (1-min reversal duration). The water conductivity was measured with WTW inoLab 7310 conductometer with LR 325/01 probe. The frequency of the water TDS measurement was 10 s. After an hour, the water in the tank was replaced with a new one. Tests were conducted at room temperature.

Fig. 2
figure 2

Schematic diagram of ED system operation: 1—purified water reservoir, 2—electrolyte reservoir, 3—water pump, 4—DC supply, 5—electrolyte stream, 6—feed stream, 7—purified water stream

Full size image

3 Results

3.1 Electrochemical characteristics of electrodes

The CV curve for the graphene electrode determined in 1 mol L−1 NaCl solution (Fig. 3) allows characterization both the resistivity of the electrode and its capacitance. The shape of the CV curve indicates that the reactivity of this material is very low in NaCl solution, even in high concentrations of the solution. There are no signals other than the capacitive current in the range from − 2.0 to + 2.0 V. The shape of the CV curve, and especially its slope, suggests that, in addition to the charge of the electrochemical double layer, there is also a potential drop across the ohmic resistance. This means that there is a surface layer on the electrode with low electrical conductivity (in comparison with metals), comparable to graphite. Based on the CV curve, the capacity of the electrochemical double layer (CCV) can be estimated using the equation: CCV = i/v, where i—current, and v—scan rate. The capacity of the electrochemical double layer was calculated for several potential values. The calculation results presented in Table 1 show that this electrode has a capacity (309–462 µF cm−2 depending on the potential value) several times higher than the values characteristic for metallic electrodes, such as a platinum electrode (40–75 µF cm−2) [56]. These values are not so high when compared to other materials based on reduced graphene oxide dedicated for supercapacitors, but high capacity is not a required feature for an electrode material for the ED/EDI system.

Fig. 3
figure 3

The cyclic voltammetry curve of the tested electrode in 1 mol L−1 NaCl solution (scan rate: 20 mV s−1)

Full size image
Table 1 The double-layer capacitance calculated from CV data
Full size table

Additional information on the electrochemical features of the graphene electrode was obtained from EIS analysis. The EIS spectrum in the Nyquist plot is given in Fig. 4. The best fitting results were obtained using the equivalent electrical circuit presented in Fig. 5. This system consists of a resistor (R1) corresponding to the ohmic internal resistance of the cell, constant phase element 1 (CPE1) and resistor (R2) connected in parallel to it representing the surface layer of the low conductivity and constant phase element 2 (CPE2) that represents the double-layer capacitance. The values of each element of the circuit are presented in Table 2. Comparing the CPE2 value and CCV values (from Table 1), it can be stated that the EIS results confirm the findings from the CV curve related to the double-layer capacitance. Both techniques give values of the same order of magnitude.

Fig. 4
figure 4

The Nyquist diagram of the tested electrode in 1 mol L−1 NaCl solution. Potential: OCP, amplitude of 20 mV, frequency range from 0.1 to 105 Hz

Full size image
Fig. 5
figure 5

Equivalent electrical circuit

Full size image
Table 2 Parameters of equivalent electrical circuit obtained by fitting the EIS measurement data
Full size table

Under conditions in which the electrode is alternately polarized with the potential of + 2.0 V for 900 s and then with the potential of − 2.0 V for 15 s (Fig. 6), successive cycles of polarization are very repeatable. The short cathodic polarization performed after long anodic polarization brings the electrode back to the starting conditions. The processes of gas evolution on the tested electrode, both in the cathodic and anodic potentials, do not affect the condition of the electrode. The characteristics of the electrode remain unchanged.

Fig. 6
figure 6

Current–time curves (b) for the graphene electrode alternately polarized with potential pulses (a) of E1 =  + 2.0 V, t1 = 900 s and E2 = − 2.0 V, t2 = 15 s, number of sequences: 4, solution: 1 mol L−1 NaCl

Full size image

To sum up, the tested graphene electrode has a surface layer of low electrical conductivity, and it does not show any electrochemical activity in a wide range of potentials (− 2.0 to + 2.0 V). Therefore, it can be described as an electrode close to a perfectly polarizable electrode. Based on the results of electrochemical tests, it can be assumed that this type of electrode can be successfully used as an element of the ED/EDI system working with water of low chloride ions concentration.

3.2 Water deionization tests

The composite electrodes coated with n-rGO were used in a multi-electrode electrodialyser for water deionization. The performance of the graphene electrodes in the ED system was evaluated by monitoring the salinity reduction of water in the dilute stream. An example of the characteristics of the desalination process is shown in the chart (Fig. 7). Due to the operating mode used (U = const.), the desalination dynamics are the highest at the beginning of the cycle, which was also reflected in the highest current flowing through the electrodes (I = 0.3 A) at the beginning, 0.1 A at the end of the cycle, resulting in a total energy consumption of 0.8 kWh m−3 of treated water. The duration of one desalination cycle was 60 min. In each cycle, the water was treated to a level of approximately 10 ppm. The desalination curve indicates that a plateau has not been reached and that the conductivity of the water can still be reduced at a slower rate if a new cycle is not started.

Fig. 7
figure 7

Water desalination characteristic in multi-electrode ED module

Full size image

The demineralization rate (DR%) was calculated from the equation \({\text{DR\;}}\% = 100\left[ {1 - \frac{{{\text{TD}}{{\text{S}}_t}}}{{{\text{TD}}{{\text{S}}_0}}}} \right]\), where TDS0 is the water salinity at the start of the cycle and TDSt is the water salinity at the end of the water treatment cycle. DR% was 90% in an hour cycle at the beginning of the electrodialyser operation.

The decrease in the desalination efficiency (Table 3) during the electrodialyser’s lifetime was calculated based on an equation \(\Delta \eta = \frac{{\Delta {\text{TD}}{{\text{S}}_S} - \Delta {\text{TD}}{{\text{S}}_T}}}{{\Delta {\text{TD}}{{\text{S}}_S}}}\;100\), where ΔTDSS and ΔTDST are the changes in the water salinity during a 1-h work cycle at the start and the end of the lifetime, respectively.

Table 3 Desalination efficiency
Full size table

Both the test conducted using the n-rGO electrodes in the ED electrodialyser and observations during work in other systems confirm the electrical stability of the electrodes. During the 60-cycle period of operation, there was no change in current–voltage parameters during desalination, both in desalination and regeneration (polarity reversed) mode.

3.3 Material tests before and after electrodialysis tests

3.3.1 Changes in the chemical structure of the composite electrode

3.3.1.1 FTIR spectroscopy

In the first step, FTIR studies of the chemical structure of graphene oxide before and after reaction with hydrazine were carried out to determine the structure of the n-rGO coating applied for the electrodes. Subsequently, the n-rGO-coated electrodes (anode and cathode) were examined after operation in the electrodialyser in desalination mode (normally polarized) and after polarity reversal to determine the stability of the n-rGO structure under operating conditions.

Figure 8 shows the FTIR spectrum of GO both unmodified and cross-linked by reaction with N2H4. The spectrum of unmodified GO shows bands characteristic for the chemical structure of this material. At the wavenumber of 1730 cm−1, there is a peak corresponding to the stretching vibrations of C=O bonds of the carboxyl group (–COOH). In addition, the presence of the carboxyl group was confirmed by the peak at 3412 cm−1, which comes from the stretching vibrations the O–H bond, being its component. Another maximum at 3285 cm−1 confirms that GO also has a large number of hydroxyl groups (–OH) linked together by hydrogen bonds on the surface of GO. The peak coming from vibrations of C–OH deformation bonds in carboxyl groups is at 1420 cm−1. Another low-intensity peak, characteristic of the epoxy bond (C–O–C), was identified in the range of 1230 cm−1. At 1078 cm−1, there is a band coming from stretching vibrations of C–O bonds. At 1620 cm−1, there is a peak typical for stretching vibrations of C=C bonds occurring in the hexagonal ring of graphene oxide [57,58,59,60]. In the wavenumber range of 3000–2800 cm−1, a wide band appears originating from both symmetrical and asymmetric stretching vibrations of C–H bonds in aliphatic hydrocarbon chains attached to the ring structure of GO [60]. Such groups are impurities, formed during the synthesis of this material. In the GO spectrum after the hydrazine cross-linking process, there are no bands characteristic for vibrations of O–H, C–OH, C–O–C and C–O bonds. In turn, the band originating from C=C vibrations was shifted towards lower values of wavenumbers characteristic of reduced graphene oxide (rGO). In addition, groups typical of C=N connections appeared at 1690 cm−1 and at 1290 cm−1, which originate from vibrations stretching C–N bonds [59, 60].

Fig. 8
figure 8

Spectrum of FTIR GO used for research: unmodified (a), modified with hydrazine (b)

Full size image

The FTIR spectra presented in Figs. 9 and 10 illustrate the structural changes of the anode and cathode that occur as a result of work in the electrodialyser after water deionization and polarization reversal. The following characteristic bands appear in the spectra: 3660 and 3430 cm−1 (O–H bond stretching vibrations), 3000–2800 cm−1 (asymmetric and symmetric C–H bond stretching vibrations), 1753 cm−1 (C=O bond stretching vibrations), 1615 cm−1 (C=C bond stretching vibrations in the aromatic ring), 1460 cm−1 (N–O bond stretching vibrations), 1370 cm−1 (N–O bond stretching vibrations), 1290 cm−1 (C–N bond stretching vibrations), 1190 cm−1 (asymmetric stretching C–O–C bonds, 1030 cm−1 (vibrations stretching asymmetric C–O bonds) [59, 60].

Fig. 9
figure 9

FTIR spectra of the anode after operation in the water electrodialyser without (a) and after polarity reversal (b)

Full size image
Fig. 10
figure 10

FTIR spectra of the cathode after operation in the water electrodialyser without (a) and after polarity reversal (b)

Full size image

In the spectrum of the anode after work in the electrodialyser, as well as after reversed polarization, the presence of a band with a maximum of 3660 cm−1 indicates the chemical attachment of –OH groups to carbon. For the cathode, a band appears at the maximum of 3430 cm−1 in its spectra, which confirms the presence of vibrations from water molecules physically bound to the surface; in such a configuration hydrogen bonds are formed between hydroxyl groups. As in the case of the anode, the intensity of this band increased after applying reversed polarization. In the spectrum of the anode after the deionization process, there is a peak of considerable intensity coming from the C=O bonds, in the spectrum from the surface of the cathode it is absent. In all spectra presented in Figs. 9 and 10, the C=C bond peak is at the typical position for reduced graphene oxide at 1615 cm−1 [57, 58]. In addition, in the spectrum of the anode after the water deionization process, several bands are originating from the newly formed C–O and N–O bonds, which significantly reduce in intensity after the polarization reversal. Several characteristic peaks originating from calcium carbonates and bicarbonates, which are scale components, are also identified in the spectrum of the cathode. These peaks are at 2560, 2512, 1790, 1445, 850, 730 cm−1. In addition, the intensity of the peak at 1670 cm−1 from the C=O groups increased, which indicates the oxidation of the cathode, which after reversing the polarity became the anode.

3.3.1.2 Raman spectroscopy

Examples of the spectra of the analysed samples obtained by Raman spectroscopy are shown in Fig. 11. All the spectra of the tested samples have a shape characteristic for rGO [61]. The obtained spectra were subjected to the deconvolution procedure using the Lorenz function in the spectral region of 1000–2000 cm−1 (Fig. 12) using the method proposed by King et al. [61]. According to it, the Gapp band located at an altitude of ~ 1600 cm−1 was deconvoluted into two modes: G and D′. Then, the difference in the positions of the Gapp and D′ bands was calculated. The value of this difference (D′ − Gapp) allows indication the reduction degree of oxygen groups in the carbon structure. Three measurements were carried out in randomly selected places on the electrode surface. The obtained, averaged results are presented in Table 4.

Fig. 11
figure 11

Exemplary Raman spectra for analysed samples

Full size image
Fig. 12
figure 12

Exemplary deconvolution process of Raman spectrum of anode after reversed polarization

Full size image
Table 4 Raman spectrum analysis of graphene electrodes with n-rGO coating
Full size table

The lowest values (D′ − Gapp) were obtained for the samples: anode and cathode after polarity reversal, and they are 3.53 and 3.51, respectively. The closer is the result of the difference (D′ − Gapp) to zero, the more oxygen groups the n-rGO contains. This is the evidence of oxidation of electrode surfaces during the electrodeionization process. The highest value (D′ − Gapp) of 4.99 (near the output state) was obtained, in the case of the cathode, which may indicate the smallest number of oxygen groups in the n-rGO structure. The high value of the standard deviation in this case equal 1.25 is related to the measurement error resulting from the presence of the scale on the electrode. When the polarity is reversed, the amount of oxygen groups on the anode is reduced and virtually returns to its initial state, while the cathode is oxidized to the same level as the anode during electrodialyser work.

3.3.2 Assessment of electrode surface morphology changes

The tests performed on the surface morphology (Fig. 13) of the composite graphene electrode prepared for use in the electrodialyser indicate its uniform coverage with the n-rGO coating. The produced coating tightly covers the support substrate made of carbon fibres bonded with epoxy resin over the entire surface of the electrode under study. The elemental composition of the surface (Fig. 13) confirms the results of the analysis of the chemical structure, revealing that in this coating, apart from carbon and oxygen, there is also nitrogen content in the amount of about 1.5% by mass. Trace amounts of sulphur and chlorine are impurities from the production of graphene oxide used for research.

Fig. 13
figure 13

SEM testing of graphene electrode—surface morphology and elemental analysis

Full size image

After the tests of water deionization, the morphology of electrodes surface was tested again. The tests were performed after the final stage of electrodialyser regeneration, that is, after reversing the polarity on the electrodes. Based on the conducted tests, it can be concluded that the anode surface has no structural changes, and the n-rGO coating still tightly covers the carbon fibres (Fig. 14a). Small, point losses were observed at the edge of the anode. The surface of the cathode is unevenly covered with calcium carbonate. No degradation of the n-rGO coating was found in the CaCO3-free areas (Fig. 14b). The elemental composition indicates the content of the main elements in the coating, i.e. C, O and N, and Ca on the cathode (Fig. 14a, b). Compared to the output electrode, there is a slight increase in oxygen content at the anode (approx. 1% by mass), and at the cathode, despite the presence of calcium carbonate, there is a decrease in oxygen content of approx. 2% by mass.

Fig. 14
figure 14

SEM image of the surface morphology and analysis of the elemental composition of: a and b cathode after work in the electrodialyser

Full size image

4 Discussion

It has been shown that the concept of the composite carbon electrode based on a carbon fibre-reinforced polymer (CFRP) skeleton and n-rGO coating presented in this work can be used in the construction of multi-electrode dialysers using reversed polarity for regeneration. The key conceptual issue for the production of such the electrode was the assumption of the possibility of producing an electrochemically resistant coating based on graphene oxide. The electrochemical and mechanical stability of this material is related to the formation of molecular pillars that connect the flakes of reduced graphene oxide.

As a substrate for the pillaring reaction, hydrazine was added to the GO suspension in such amounts and in such a way to significantly, but not completely, reduce the graphene oxide flakes simultaneously with the pillaring process. Natural functional groups on the edges of rGO, were used as active centres for the formation of pillars in the structure of the coating material, while similar active centres on the surface of the load-bearing skeleton were produced by partial removing the epoxy resin from the surface of the fibres and then oxidizing. The final reduction of oxygen groups excesses in the structure was made by heating the already polymerized electrode in hydrogen at a temperature slightly higher than the polymerization temperature of the epoxy resin used. The effectiveness of pillaring was confirmed by identifying peaks from C–N-type bonds and C=N-type bonds in the FTIR spectra.

As a result of the reaction of hydrazine with GO, the O–H, C–OH, C–O–C and C–O bonds were reduced. The high degree of reduction of graphene oxide is also indicated by the difference in peak positions (D′ − Gapp) in the Raman spectrum. In this way, the electrode was obtained with a nanostructured graphene coating with high cohesion with the CFRP backing substrate and comparable electrical conductivity to graphite. Based on the results of electrochemical tests, it can be concluded that an electrode of this type can be successfully used as an element of the ED/EDI system working with water with a low concentration of chloride ions. The tested graphene electrode does not show any electrochemical activity in a wide range of potentials. The reactivity of this material is very low so it can be described as an electrode similar to a perfectly polarizable electrode. The electrode has an order of magnitude higher electrochemical capacitance than metal electrodes, but from the point of view of application in ED/EDI systems, high capacitance is not a required feature, unlike capacitive deionization (CDI). The ED water deionization technique is often associated with the need to operate in reverse polarization mode to support the purification of the electrodialyser from the ions removed. Tests carried out on n-rGO coated electrodes have shown that a short cathodic polarization carried out after a long anodic polarization restores the electrode to its initial conditions, so the electrode is well suited to this mode of operation. The processes of gas evolution on the tested electrode, both in the cathodic and anode potentials, do not affect the condition of the electrode. The characteristics of the electrode remain unchanged.

The functionality and exploitation durability of the electrodes were verified by installing them in a multi-electrode test stand for water electrodialysis. Each of the electrodes worked alternately as a cathode and anode in the work and regeneration cycles, respectively. The tested electrodialyser effectively removed ions from water in each of the 60 work cycles with full stability of current and voltage parameters. At the start of the test, the ion removal rate was around 92%, and at the end of the work (after treating 180 L of water), it decreased by approximately 5%. After the completion of the test cycle, the electrodes were dismounted and subjected to testing using FTIR and Raman spectroscopy, to examine their condition both after the end of the work cycle and after the end of the regeneration cycle. In the FTIR spectrum of the electrode working as an anode, additional peaks appeared, from the C=O, C–O and N–O bonds, which are not present in the cathode spectrum, indicating oxidation of the anode during operation. However, when the polarization is reversed, the intensity of these peaks decreases or they almost disappear, indicating the restoration of the initial n-rGO structure. The effectiveness of the oxygen reduction process on the anode after polarization reversal was also demonstrated by Raman spectroscopy results, where the (D′ − Gapp) intensity difference for the anode after polarization reversal is close to the initial value. This confirms the reversibility of the oxidation/reduction phenomena at the electrode surface. Elemental composition studies by EDS after electrode regeneration also indicated that the oxygen content on the anode surface returned to its initial state. Furthermore, analysis of the surface morphology by SEM showed no change in the n-rGO coating. After the deionization tests, the coating adhered as well to the substrate as before the test. No signs of mechanical degradation, no cracks, no delamination of the coating, and no cavities were observed. It should be pointed out that the fact that there is a drop in the performance of the electrodialyser at the end of the test related to the precipitation of calcium carbonate, both on the cathodes and on the cooperating membranes.

After conducting an extensive literature review, it was determined that there are no similar material solutions for electrodialyser electrodes used in ED or EDI processes. Systems for producing ultrapure water typically use titanium electrodes coated with noble metal oxides. These solutions are comparable to the ones presented in this article but much more expensive. This may be problematic, especially in large industrial electrodialysers. The proposed graphene electrode is more durable than cheap graphite or metal electrodes, also cheaper than titanium electrodes. Graphene, graphene oxide or its derivatives have previously only been used for capacitive electrodes in water purification using the CDI/MCDI method. The proposed solution is therefore original and has not been used before.

5 Conclusions

  1. 1.

    This paper presents the concept of a unique carbon electrode, made of carbon fibres bonded with epoxy resin and coated with nanostructured reduced graphene oxide (n-rGO).

  2. 2.

    The n-rGO coating created by the reaction of hydrazine with graphene oxide gives the carbon electrode electrochemical stability and strength.

  3. 3.

    The electrodes presented in this article maintain constant current–voltage parameters with variable duty cycles and high-water deionization efficiency.

  4. 4.

    The electrodes can be used in a multi-electrode electrodialyser operating in cycles with regeneration based on the principle of changing polarity.