Ammonia Recovery from Wastewater as a Fuel: Effects of Supporting Electrolyte on Ammonium Permeation through a Cation-Exchange Membrane.

Electrodeionization (EDI) is used to recover ammonia from wastewater as a fuel, but how its performance for ammonia recovery is affected by the supporting electrolyte is not very clear. This study involved experimental tests and theoretical calculations on NH3 recovery, NH4 + permeation, and NH4 + and Na+ interacting with the functional groups in a cation exchange membrane (CEM) using Na2SO4 as the supporting electrolyte. The results demonstrated that a low concentration (≤0.250 mol L-1 of Na2SO4) was conducive to NH4 + permeation, while the a concentration (0.750 mol L-1 of Na2SO4) hindered NH4 + permeation. A maximum recovery efficiency of ammonia of 80.00%, a current efficiency of 70.10%, and an energy balance ratio of 0.66 were obtained at 0.250 mol L-1 of Na2SO4. Numerical results indicated that an increase in Na2SO4 concentration caused severe concentration polarization that resisted NH4 + migration in the CEM. The DFT results demonstrated that competitive adsorption of Na+ to the CEM hindered NH4 + migration. The weaker interacting force between NH4 + and the sulfonate functional group (-SOH3) in comparison to that between Na+ and -SOH3 might be related to the geometric and orientation effects, which generated an additional energy barrier for NH4 + transport. Therefore, this study suggests that the supporting electrolyte concentration should be matched with that of the desalted ions.

Introduction

Ammonia (NH3), occurring in wastewater and liquid waste from industry, municipalities, and agriculture has become the second-largest amount of pollutants after chemical oxygen demand (COD).[1] Overdischarged NH3 entering water bodies results in environmental issues such as dissolved oxygen reduction, corrosion, disinfection inefficiencies, and toxicity to aquatic creatures.[2,3] Nevertheless, NH3 is rich in H (17.60 wt % H) and has 5.2 kWh kg–1 energy intensity, 40% higher than that of methanol (12.60 wt % H, 4.32 kWh kg–1), it is an attractive carbon-free fuel, and it has an essential role in energy storage and transportation.[4] NH3 removal/recovery from wastewaters has attracted great effort by many scholars,[5,6] but it is either not easy to extract or the recovery efficiency is extremely low.[7,8]

Electrodeionization (EDI) is considered an effective approach for salt recovery, and it performs very well in ammonia recovery.[5,9−11] During ion separation, a supporting electrolyte is indispensable, since it provides sufficient electrolytic conductivity, reduces the Ohmic drop,[12−14] and releases the surface phenomena occurring between pollutant species and metal hydroxides.[15,16] It also enhances the ionic strength to form a compressed electric double layer (EDL) and generate a greater electric current in the circuit under a specific applied potential.[17,18] Meanwhile, the supporting electrolyte causes a problem in that it sometimes replaces the transport ion and becomes a determining factor, which greatly influences the concentration polarization of the ion exchange membrane.[19] According to the literature, the an inert or reactive supporting electrolyte concentration changed the limiting diffusion current density and led to an uneven electric current distribution.[20,21]

Additionally, the ions of the supporting electrolyte generally have more significant diffusion coefficients in comparison to the transport particles and form an extended DBL to hinder the target ion’s transport and reduce the selective efficiency. Our study found that the Na2SO4 concentration caused fierce competition between Na+ and NH4+ in the electric double layer (EDL). As a result, Na+ formed a thin and compact layer to prevent NH4+ reduction, thus decreasing the hydrogen recovery.[9] We also found that NH4+ behaviors in the cathode were closely related to the NH4+ concentration.

However, the effects of the supporting electrolyte on NH4+ selectively migrating through cation exchange membrane (CEM) are still not fully understood. Therefore, in this paper, an extended study of the relationship between NH4+ permeation in a cation-exchange membrane and the supporting electrolyte is carried out through experimental tests and theoretical calculations. Na2SO4 is used as the supporting electrolyte because of its excellent performance.[22−25] We discuss NH3 recovery, NH4+ permeation, and molecular interactions of NH4+, Na+, and a functional group in a CEM based on ions permeating through polymeric membranes, an Arrhenius-type equation,[26,27] and density functional theory (DFT).[28,29] We hope that the results are helpful for EDI application on a full scale.

Methods and Materials

Experimental Tests

Electrodeionization Setup

The EDI cell structure and the electrode material were the same as in our previous study.[9] A two-channel EDI reactor was made of an organic glass material. The effective volume of each channel was 20 mL. The anode and cathode were made of platinum plates with an area of 16 cm2. The exchange membranes were IONSEP AM and IONSEP CM (IONTECH Co., Ltd., Shanghai, China). The diluted channel was fed into a 0.25 mol L–1 fixed (NH4)2SO4 solution, and the concentrated channel was fed with 0.125, 0.250, 0.375, 0.500, 0.750 mol L–1 Na2SO4. An extra voltage of 2.0 V DC (PS-305DM, Hong Kong Longwei instrumentation Co., Ltd.) was applied to the EDI cell. All tests were conducted at 24.40–25.20 °C and 1.0 atm.

Chemical Analysis

The total gas generated from the cathode was collected in a gas bag then flowed through three serial cylinders with 1 mol L–1 H2SO4 solution. The ionized NH3 and NH4+ were measured via the Berthelot method. The residual H2 was determined by gas chromatography (Agilent 4890D; J&W Scientific, USA).[30] An HP-MoleSieve column (30 m × 0.53 mm × 50 m) was the detection column. Helium gas, the carrier gas, was injected at a rate of 6 mL min–1. The injection port, column, and thermal conductivity detector temperatures were 200, 35, and 200 °C, respectively. Samples were injected by microsyringes with in 200 μL amounts (Shanghai Anting Scientific., China).

Electrochemical Impedance Spectroscopy and Current Measurements

Electrochemical impedance spectroscopy was used for measurements according to the method devised by Fu et al.[31] First, the membrane was immersed in 0.25 mol L–1 Na2SO4 for 24 h before each set of measurements.[25] Then, the impedances were recorded with frequencies ranging from 50 to 15000 kHz. Each point was repeated three times to minimize noise effects. Each point was repeated three times to minimize noise effects. The resistance was obtained based on fitted results as depicted in Supporting Information.

The current–voltage curve was measured on the basis of a method developed by Choi et al.[24] The cell was divided into two equal-volume (20 cm3) compartments by a CEM. The anolyte (NH4)2SO4 was fixed at 0.250 mol L–1, and the concentration of the catholyte Na2SO4 was increased from 0.125 to 0.750 M. Anodic and cathodic channels were fed at a flow rate of 1.00 mL min–1. A DC power source was added between the two Ag/AgCl electrodes. Then, a stepwise current supply was applied at a scanning rate of 30 mA min–1 through an electrochemical workstation (CorrWare, Scribner Associates Inc., USA). Finally, the electrodes were connected to a Keithley 2700 multimeter (Tektronix, Inc., USA). The potential drops of the two sides of the membrane were recorded and stored every 30 s.

Numerical Simulations and Calculations

The model was developed on the basis of one-dimensional transport phenomena. The initial concentrations of ions, flow rates, applied voltages, and membrane characteristics were input according to the characteristics of mustard tuber wastewater, the operating conditions, and the geometric structure. The principle of ion transport is displayed in Figure a. The assumed geometry of an electrolysis cell pair consisted of three parts (a diluted compartment, a concentrated compartment, and a cation-exchange membrane).[32−35] Since we only focused on the CEM, the numerical results near the membrane were plotted. The geometry of the cell was drawn using COMSOL Multiphysics version 5.4,[33] as shown in Figure b. Details of the calculations are described in the Supporting Information.

Figure 1

Ion transport through the membrane: (a) geometric structure and mesh used in numerical simulation; (b) polymer-sulfonate and solution.

To simplify the numerical simulation, the process of the whole system maintained electroneutrality. Diffusivities and mobilities of the cation and anion are functions of the ion concentrations. The resistance of the ion exchange membrane was a function of both the salt concentration and time. The current densities of the ions are assumed to be equal to the current densities of the solutions at the interface of the membranes. No chemical reactions or water splitting were assumed even though the applied voltage reached 2.0 V.

Mass Transfer Phenomenon

Concentration diffusion and electrical migration are two forces that generate a concentration difference.[32] In this model, the Nernst–Planck–Poisson equation (NPP) is used in a two-dimensional mode to describe the ion flux and charge transport, as expressed by eq (33−35)where Di is the diffusion coefficient, ci is the concentration of the ions, –Di▽ci is the diffusion flux, zi is the charge, ui is the species mobility, F is Faraday’s constant, and φ is the electrolyte potential. In the convection term, v denotes the fluid velocity vector. i is NH4+, Na+, or SO42–.

Current Density

All of the molar fluxes should be used to calculate the electrolyte current density, but we ignored the convection term, and so the electrolyte current density is expressed by eq where the summation is made over all n ions. If ρ is the space charge density, then the current is given by eq :

The mesh building and solution study are available in the Supporting Information.

Current utilization (CU) and energy consumption (E) are two important parameters for EDI. CU is expressed through a control volume analysis for general electrolytes (eq )[36]where D is the diffusion coefficient, c is the concentration of the ions, and z is the charge carried by a cation/anion, respectively.

Substituting the electroneutrality equation (eq ), current utilization is presented in eq .

Energy consumption (Es) is used to compare the energies of different desalination devices. In electrochemical desalination systems, energy consumption is expressed by eq :where I is the current, V is the voltage, and Qdesalted is the flow rate of desalted water.

Since energy consumption is a widely used metric determining the economic viability of a desalination technique, we consider energy consumption to remove a single pair of ions to more exactly describe the energy sonsumption. The energy per ion removal (Esp) is given by eqs and 9:

In unipolar EDI, EPIR/V* is shifted to the opposite direction of the CU shift. z, F, kB, and T indicate the ion valence, Faraday’s constant (=9.65 × 104 C mol–1), thermal energy (=2.479 kJ mol–1; kB, the Boltzmann constant) and temperature, respectively. V is the voltage, V* is the nondimensionalized voltage, I is the current, N is the number of membrane pairs, C0 is the initial ion concentration, Cdesalted is the ion concentration of the desalted flow, and Qdesalted is the total desalted flow rate.

Energy Barrier

The energy barrier was calculated through an Arrhenius-type equation using single-salt solutions,[26] as given by eq where J is the ion flux through the membrane, R is the universal gas constant, T is the absolute temperature, Ea the energy barrier, and A is the pre-exponential factor.

Then the natural logarithm of the ion flux, ln J, was plotted as a function of 1/T, and the energy barrier (Ea) and pre-exponential factor (A) were determined from the linearized form of the Arrhenius-type equation (eq ):

Computational Details

The polymer sulfonate-ion unit was constructed in Materials Studio version 8.0 software, as shown in Figure b. The optimized geometry, electronic structure, and energy changes were calculated using the Dmol3 module.[37] GGA+PBE was used as the exchange correlation functional. A double-numerical basis set containing a polarization function (DNP) was employed for the calculations. The core treatment was set as semicore pseudopots to manage the interaction between the nucleus and valence electron. During the structure optimization, the k-point set of the Brillouin zone was 2 × 2 × 1 and the force convergence error was 0.02 hartree nm–1. After the optimized geometric structure was obtained, the physical properties, including the density of states, frequency, and Fukui function, were calculated on the basis of the optimized geometric structure. The k-point set was 24 × 24 × 1. Charges were analyzed using the Hirshfeld method.[29,38]

Results

Ammonia Recovery from Solution

Driven by the applied electricity and concentration difference, NH4+ is transported from the anodic channel to the cathodic channel. As shown in Figure , when the concentration of Na2SO4 increases from 0.125 to 0.750 mol L–1, the removal efficiencies of NH4+ climb to 54.25%, 88.80%, 89.41%, 65.01%, and 59.64% at 160 min but slightly decrease to 64.76%, 79.77%, 77.75%, 59.17%, and 54.03% at 180 min, respectively. The maximum removal efficiency of NH4+ is 79.77% at 0.250 mol L–1 of Na2SO4. This trend is thought to be caused by the different deionization rates, which increase to 5.83, 10.88, 10.25, 6.66, 5.94 mol L–1 h–1 within 30 min but gradually decrease to 1.171, 2.66, 2.01, 0.44, and 1.22 mol L–1 h–1 at 180 min, respectively (Figure a).

Figure 2

Ammonium conversion: (a) ammonium removal; (b) total ammonia nitrogen.

Since NH4+ migrating to the cathode was accompanied by water splitting, the generated OH– combined NH4+ to produce NH3·H2O and further generated NH3 under 2.00 V extra voltage. Therefore, the nitrogen element in the whole system existed in the forms of anodic NH4+, cathodic NH4+, NH3 gas, and undetectable nitrogen, as discussed in our previous studies.[9,39] As shown in Figure b, the anodic amounts of NH4+ are 6.12, 1.96, 1.85, 5.06, and 8.00 mmol, respectively, and the cathodic amounts of NH4+ are 8.44, 11.01, 9.55, 9.01, and 7.03 nmol, respectively. With the anodic NH4+ migrating to the cathode, the converted NH3 reaches 2.91, 5.93, 4.47, 1.41, and 2.43 mmol, respectively. The maximum NH3 mass (5.93 mmol) is obtained at 0.25 mol L–1 Na2SO4. According to the nitrogen mass balance, it can be seen that a concentrated supporting electrolyte harms harms ammonia recovery.

Ammonium Permeation through a Cation Exchange Membrane

The above phenomena were thought to be caused by ion transport in the CEM and were further analyzed through numerical simulations. Since the conversion flux was ignored to simplify the calculations, the total flux consisting of the diffusion flux and migration flux in the direction along with the main component was analyzed. As plotted in Figure a, when the Na2SO4 concentration (0.125 mol L–1) is lower than that of (NH4)2SO4 (0.250 mol L–1), the diffusion flux of NH4+ is positive, which means that the concentration potential transport predominantly drives NH4+. Conversely, when the Na2SO4 concentration equals the (NH4)2SO4 concentration, the diffusion flux is close to zero, indicating that the electric force replaces the concentration potential as the dominant driving force. When the Na2SO4 concentration (>0.250 mol L–1) is higher than the (NH4)2SO4 concentration, the diffusion flux becomes negative, signifying that the concentration potential generated by Na+ hinders NH4+ transport.

Figure 3

NH4+ fluxes and energy barriers: (a) diffusive flux; (b) migrative flux; (c) total flux; (d) energy barriers.

On consideration of electric migration, when the Na2SO4 concentration is lower than that of (NH4)2SO4, the migration flux of NH4+ (absolute value) decreases to zero, as shown in Figure b). In contrast, with an increase in the Na2SO4 concentration, the migration flux of NH4+ dramatically increases from 0.004 m2 s at 0.250 mol L–1 to 0.023 m2 s at 0.750 mol L–1, while its absolute value in EDL remains stable at 0.004 m2 s as the Na2SO4 concentration (0.25 mol L–1) equals that of (NH4)2SO4 (0.25 mol L–1). When the Na2SO4 concentration (0.375–0.750 mol L–1) is higher than that of (NH4)2SO4 (0.250 mol L–1), the migration flux of NH4+ in absolute value in EDL decreases. The sum of diffusion and migration referring to total flux is illustrated in Figure c. The variations of NH4+ fluxes in EDL and DBL demonstrate that the diffusion flux of NH4+ has a positive increase with Na2SO4 concentration when the concentration of NH4+ is higher than that of the supporting electrolyte. The differences in fluxes finally result in changes in concentration and potential along with Na2SO4 concentration (Figures S1 and S2). It should be noted that the trend of NH4+ fluxes matches that of the measured potential.

On the basis of the total flux of NH4+, energy barriers of interaction between NH4+ and CEM were calculated and are plotted in Figure d. The energy barriers between the bulk solution and CEM for five concentration gradients are 708, 699, 664, 618, and 525 kcal mol–1. A higher energy of the CEM indicates that the membrane shows earlier elution to the cation at equilibrium. The process of NH4+ exchange is faster than the cation in the membrane with lower energy.

Additionally, it can be seen that the zigzags of migration flux become increasingly evident with the supporting electrolyte concentration. We imagine that this phenomenon might be related to the mesh and the supporting electrolyte concentration. During calculations, it was found that a finer mesh gave smaller zigzags. However, the finer the mesh was set, the harder it was for the model to converge. Therefore, the mesh was adjusted to make sure that the model was converged. However, zigzags could not be completely eliminated. This phenomenon indicates that the model needs to be improved when it is used to study concentrated ion migration.

Ammonium Interaction with Functional Group

The interaction between permeable NH4+ and the CEM was studied via theoretical calculations. The structures of polymer sulfonate-acid, polymer sulfonate-ammonium, and polymer sulfonate-sodium units were built and optimized in Materials Studio software (Figure ). The DFT-calculated length information on more bonds at the minimum energy state is given in Table . In detail, the original O4–N and O4–Na bond lengths in the functional group (−SO3H) were 1.480 and 2.401 Å, respectively. After geometric optimization, the bond lengths decreased to 1.473 and 1.168 Å, respectively. These obtained lengths are in good agreement with the results reported in the literature.[29]

Figure 4

Polymer sulfonate-ion unit: (a) original polymer sulfonate-NH4+; (b) optimized polymer sulfonate-NH4+; (c) original polymer sulfonate-Na+; (d) optimized polymer sulfonate-Na+.

Table 1

Most Relevant Bond Lengths (Å)

bond	polymer sulfonate-acid	polymer sulfonate-ammonium	polymer sulfonate-sodium
C–N	1.140	1.082	1.110
C–N	1.140	1.079	1.111
H17–C	1.337	1.354	1.409
C15–H17	1.447	1.419	1.471
C13–C15	1.432	1.395	1.441
C14–C15	1.377	1.458	1.441
F2–C14	1.326	1.333	1.397
N8–C12	1.320	1.332	1.393
O3–N8	1.350	1.392	1.427
O3–N7	1.356	1.418	1.428
N7–C11	1.315	1.308	1.390
S1–C10	1.753	1.771	1.771
S1–O5	1.438	1.444	1.714
S1–O4	1.604	1.506	1.635
	1.110 (O4–H18)	1.473 (O4–N)	1.168 (O4–Na)
		1.010 (N–H)

Bond lengths reflect the polymer sulfonate-ion bound structure’s mechanical and chemical stability.[40,41] A shorter bond signifies that more energy to break it is required. According to Table , the bond length of the −SO3H group interacting with Na+ is much shorter than that with NH4+, indicating that Na+ is more accessible than NH4+ in the membrane. The simulated chemical structures at the minimum energy state, along with their HOMO and LUMO frontier orbitals, are shown in Figure . The energies of the highest occupied molecular orbital for NH4+ and Na+ combined with sulfonate are −5.946 and −5.190 eV, and the energies of the lowest unoccupied molecular orbital are −4.160 and −4.370 eV, respectively. Polymer sulfonate-NH4+ and polymer sulfonate-Na+ units exhibit band gaps of 1.786 and 0.820 eV, respectively. Generally, the larger the energy gap between the frontier orbitals, the more difficult it is to transfer electrons from the HOMO to the LUMO because the activity increases as the energy gap decreases. The energy gap demonstrates that Na+ combined with SO3H groups has a higher chemical and mechanical stability in comparison to NH4+.

Table 2

Energy Balance Evaluation

Na2SO4 concn, mol L–1	current utilization, %	current efficiency, %	energy potential (NH3), J	energy consumption (total), J	energy balance
0.125	78.53	58.23	931.21	1629.39	0.57
0.250	81.78	70.10	1898.44	2886.92	0.66
0.375	83.51	62.17	1430.36	2302.64	0.62
0.500	84.58	46.84	450.42	1629.30	0.28
0.750	85.85	37.65	776.60	1921.34	0.40

Figure 5

Chemical structures simulated in this work, along with the spatial distributions of their frontier HOMO and LUMO orbitals: (a) polymer sulfonate-NH4+; (b) polymer sulfonate-Na+.

Moreover, Na+ has a smaller radius but a larger diffusion coefficient than NH4+. Its interaction force with CEM is much stronger than that of NH4+, indicating that more favorably adsorbed ions with higher affinity to the polymer matrix have lower diffusion within the membrane: e.g., Na+ was more favorably adsorbed than NH4+ but diffused more slowly through the membrane. Na+ had stronger attractive bonds with the fixed groups in the polymer matrix that inhibited its diffusion within the membrane in comparison to NH4+. Moreover, NH4+ has a central nitrogen atom surrounded by four hydrogen atoms in a tetrahedral geometry. This structure consists of four equally spaced sp3 hybrid orbitals forming a tetrahedral geometry with bond angles of 109.5°, which is an asymmetrical 3D structure. A rotation around any axis may result in a molecule indistinguishable from the original. The differences in electron densities around the molecule can render some areas on the surface sterically less approachable than others. Therefore, the geometric and orientation effects may impose an additional energy barrier for NH4+ transport.[26] These results further explain why the excess concentration of supporting electrolyte hinders the target ion transport in an ion exchange membrane.

Discussion

Ammonia Recovery Related to Current Utilization and Energy Balance

On the basis of the recovered ammonia mass and consumed energy, the current utilization, current efficiency, and energy balance were analyzed and are summarized in Table . Apparently, the current utilization slightly increase with an increase in Na2SO4 concentration, signifying that the concentration of the supporting electrolyte has almost no effect on the Coulombic efficiency when the conductivity of the bulk solution is high enough.[42] However, the current efficiency had an upward trend to 70.10% at 0.25 mol L–1 Na2SO4 and then a dramatic downward trend to 37.65% at 0.75 mol L–1 Na2SO4. The decrease in current efficiency indicates that there is a lower actual mass of a substance liberated from an electrolyte by the current in comparison to the theoretical mass, which is generally thought to be affected by increasingly severe concentration polarization.[43] In this study, the recovered NH3 increases with Na2SO4 concentration when the real current is under the limiting current density. Moreover, Na+ migration is an ignorable factor due to supporting ions sharing the current, explaining the reason the current efficiency decreases with Na2SO4 concentration. Due to the differences in current efficiencies and consumed energies, it can be seen that a maximum energy balance of 0.66 is obtained at 0.25 M, which signifies that the recovered NH3 cannot achieve an offset of energy consumption and still needs 34% of energy input.

Figure a,b display the current density and potential crossing the membrane. Specifically, the current density increases from 2.98 mA cm–2 at 0.125 mol L–1 to 5.63 mA cm–2 at 0.750 mol L–1. In contrast, the potential decreases from 0.0086 to −0.016 V at the corresponding Na2SO4 concentration. Since the changes in the current and potential further resulted in resistance differences, the electric resistance of the whole system was analyzed with the aid of EIS. For EDI, the total resistance tot equals the sum of the resistance of the solution (sol), the resistance of the membrane (mem), the resistance of the diffusion layer (diff), the resistance of the Donnan interfacial layer (don), and the resistance of the electrode (el)[25] (eq ).

Figure 6

Electric properties: (a) real current density; (b) membrane potential; (c) electric resistance.

The perturbation frequency ranges are usually divided into three regions (high-, medium-, and low-frequency ranges). Each region reflects its corresponding resistance. In the high-frequency ranges, the total impedance of the membrane system is expressed as eq :[44,45]

In low-frequency ranges, the total resistance is expressed as eq :el and don are negligible; thus, eq is simplified to eq where j = √−1.

The fitting resistances for the different parts are illustrated in Figure c. The electric resistance of a solution apparently decreases from 25.9 Ω at 0.125 mol L–1 to 13.93 Ω at 0.75 mol L–1, while that of the membrane slightly increases from 8.22 to 8.54 Ω. The resistance of the diffusion layer experiences a dramatic drop to 2.5 Ω and then an increase to 16.0 Ω. The total resistances are 41.70, 32.00, 32.40, 37.29, and 38.50 Ω, respectively. The variation of electric resistance indicates that increasing the supporting electrolyte concentration indeed reduces the Ohmic resistance for the dilute condition, which is related to the ionic conductivity and viscosity of the bulk solution. In contrast, the concentrated supporting electrolyte makes concentration polarization much easier to occur. This finding is critical for practical engineering, since the concentrated channel undergoes serious concentration polarization, leading to much more energy consumption and poorer desalination capability. Therefore, it is significant to keep a rational concentration of the supporting electrolyte to improve desalination efficiency.

Ammonia Recovery Hindered by Concentration Polarization of SO42–

According to the achieved results, NH4+ transport through the CEM was closely related to the Na2SO4 concentration. Ammonium ion exchange mainly occurred at the interfaces, and the exchange process was short and fast. As a result, the aggregate amount of ammonium at the interface was much larger than the amount in the membrane. Therefore, the concentration of ammonium in the membrane seems to be unchanged. On consideration of SO42–, Na2SO4 dissociated in the solvent and formed the electrochemically stable complex Na+ and SO42– to provide current transport through the solution and CEM. SO42– has a different interaction with the CEM in comparison to Na+. As plotted in Figure , SO42– forms a thick diffusion layer at the interface of the CEM and the cathode bulk solution. The thickness of the diffusion layer increases with Na2SO4 concentration, caused by concentration and electric potential gaps. An increase in the difference of the concentration between the anolyte and the catholyte serves as a force to drive SO42– diffusion. Although SO42– migrates in the direction of the anode in the electric field, it cannot go through the CEM. As a result, the accumulation of SO42– forms a thick diffusion layer. SO42– clearly shows a Donnan exclusion effect due to permanent negative charges at the CEM surface. SO42– has slower migration and effective diffusion coefficient than NH4+.[46,47] Even though the electric force cannot drive Na+ migration to the CEM, its concentration potential is large enough to impose a strong force so that Na+ carrying a positive charge is adsorbed by the anion in CEM.

Figure 7

Concentration distribution near CEM: (a) SO42– concentration profile; (b) NH4+ concentration profile.

Conclusion

This study investigated Na2SO4 as the supporting electrolyte affecting ammonia recovery from nitrogenous wastewaters. Experimental tests and theoretical simulations demonstrated that Na2SO4 conditionally improved ammonia recovery. Increasing the Na2SO4 concentration in the dilute range (≤0.250 mol L–1) accelerated NH4+ transport. As the Na2SO4 concentration was further increased, the current efficiency and energy balance ratio decreased. The maximum recovery efficiency of 80.00%, current efficiency of 70.10%, and energy balance ratio of 0.66 were obtained at 0.250 mol L–1 of Na2SO4. Thus, the feasible concentration of Na2SO4 in this study was 0.250 mol L–1. The exchange process of ions in the membrane was determined by the diffusion layer formed by SO42– and the interaction energy. The thick diffusion layer resisted NH4+ transport through the CEM. Na+ competitively interacted with −SO3H groups to produce a high chemical and mechanical stability, hindering NH4+ migration. The geometrt and orientation of NH4+ might generate an additional energy barrier for NH4+ transport. Therefore, this study suggests that the supporting electrolyte plays an essential role in EDI and controlling a rational concentration is conducive for NH3 recovery via EDI.