Sustainable process for the preparation of potassium sulfate by electrodialysis and its concentration and purification by a nanofiltration process

Abstract

A sustainable approach for the production of high purity potassium fertilizer (K₂SO₄) is highly needed to cope with the food crisis. High value K₂SO₄ produced by the conventional process as well as by metathesis electrodialysis (mED) contains precursor salts as impurities. With the increase of produced K₂SO₄ concentration, the amount of impurity in the product also increases. Hence, a suitable strategy is required for the rapid and effective production of high concentration and relatively pure K₂SO₄. Herein, we report a combination of electrodialysis and nanofiltration (NF) processes for the scalable production, purification and concentration of K₂SO₄. Several parameters such as applied voltage, ratio of the reactants (Na₂SO₄ and KCl) and their concentrations were optimized to obtain K₂SO₄ with relatively high concentration along with good purity via the mED process. The K₂SO₄ obtained by mED was further subjected to NF using a poly(piperazineamide) thin film composite NF membrane to concentrate and to eliminate the chloride (Cl⁻) and associated monovalent cations from the product stream. The NF parameters such as feed pH and applied pressure were adjusted to obtain a high concentration of aqueous K₂SO₄ with purity as high as ca. 99%. This process thus provides a suitable way for the production and recovery of solid K₂SO₄.

---

Introduction

Potassium is one of the three essential elements that plants require for their proper growth.^1,2 Generally, one can supply potassium to the plants by fertilizing the potassic salts to the soil. Potassic fertilizers are usually divided into two categories: one is potassic fertilizers containing chlorine as a component mainly in the form of potassium chloride (KCl). KCl may be harmful or toxic for chloride-sensitive crops (such as potatoes, tomatoes, citruses and so on), moreover, it would increase soil salinity and pH.^1,2 The other type of chlorine-free potassic fertilizers are potassium sulfate (K₂SO₄), potassium carbonate (K₂CO₃) and potassium nitrate (KNO₃). K₂CO₃ and KNO₃ are expensive to be used in agriculture. However, cost of K₂SO₄ is relatively lower and it has excellent physical properties, and thus is more appropriate for chloride-sensitive crops.³ Therefore, the preparation of K₂SO₄ seems to be very important.

K₂SO₄ is rarely found in a pure form since it is mixed with magnesium, sodium and calcium salts. Hence, preparation of relatively pure K₂SO₄ requires additional processing to separate these other salts. K₂SO₄ was first prepared by stoichiometric reaction between KCl and sulphuric acid (H₂SO₄). The hydrochloric acid bi-product was evaporated from the admixture. The resulting solution after cooling yields K₂SO₄ crystal.⁴ Efraim et al. reported the co-production of K₂SO_4, sodium sulfate (Na₂SO₄) and sodium chloride (NaCl) by progressive precipitations in the evaporation step.⁵ The process was time consuming.⁵ Zisner et al. reported a methodology for the production of K₂SO₄ by differential contacting process. First, potash, water and Na₂SO₄ were placed in a differential countercurrent contactor. After a while, Na₂SO₄ source, potash and water were allowed to contact differentially. The product K₂SO₄ was washed in situ to remove NaCl impurities from the mother liquor, which significantly improved product quality without additional capital and operating expenses.⁶ Iwashita et al. reported the preparation of K₂SO₄ from the kneading of KCl and H₂SO₄ followed by heating the mixture at 250–500 °C where the KCl to H₂SO₄ ratio was 1.40 : 1.07.⁷ Preparation of K₂SO₄ by the reaction between KCl and ammonium sulfate was also reported. The product K₂SO₄ contained 12% Na₂SO₄ as impurity.⁸ All the above mentioned processes requires high energy consumption and hence cost of the process increases.

Besides the conventional process, inorganic salts can also be prepared by electrodialysis (ED) process. ED is also used for water desalination, separation and concentration of valuable chemicals from water.^9–15 The conversion of one salt to another salt by the reaction between two different salts in ED process is known as metathesis electrodialysis (mED). In mED process, the separation and purification of the produced salt is relatively easy, as each compartments are separated by cation exchange (CEM) and anion exchange (AEM) membranes alternatively.^16,17 The double salt formation due to close solubility of different salts (reactant and product) is also avoided by employing this process. Typical mED stack contains six compartments such as two feed, two product and two electrode wash compartments.^16,17 After the application of an external electric field, ions starts to move to the respective electrode. Thus, ions from feed salt solutions transport through respective ion-exchange membranes to the adjacent compartment (which is a new product compartment). As a result, concentration of ions in feed solutions will decrease as cations and anions will be transported through the respective ion-exchange membranes. These ions will combine with each other and will form new salts as products in product compartments. The overall result of a mED process is similar to ED, i.e. ion concentration increases in the product streams with a depletion of ions in the feed streams. The efficiency of the process depends on the electrochemical properties, more particularly the transport number of CEM and AEM as well as the design of the mED setup. There are only few reports for the production of salts by mED process. For example KNO₃ had been prepared by mED using commercial AEM (Ralex AM-PP) and CEM (Ralex CM-PP).^16,17 The effect of operating parameters such as current density, feed concentration, composition, and solute and solvent transport phenomenon were reported. The product KNO₃ contained NaNO₃ as impurity. The final salt concentration was limited by electroosmosis. Preparation of potassium carbonate (K₂CO₃) from the reaction of sodium carbonate and potassium sulfate employing mED process using polyethylene–polystyrene interpolymer based CEM and AEM had been reported by researcher from our laboratory.¹⁸ The product K₂CO₃ contains 1.6–4.8% impurity depending on the applied current density and reactant concentration.¹⁸ The power consumption value for this process was reported to be 2.8 kW h kg⁻¹ at 30 mA cm⁻² current density and 1N reactant concentration. Preparation of high purity magnesium sulfate (MgSO₄) from the reaction of magnesium chloride and Na₂SO₄ by mED using commercial CMX and AMX (Neosepta) was also reported.¹⁹ The feed concentration of both the reactants were varied from 0.3 M to 0.5 M. The product (MgSO₄) contained Na₂SO₄ impurity.¹⁹ The power consumption value at 40 mA cm⁻² current density was 1.6 kW h kg⁻¹. Preparation of magnesium chloride (MgCl₂) by the reaction between MgSO₄ and KCl via mED using commercial Neosepta, membranes (CMX and AMX) was also reported.²⁰ The flow of co-ions (Cl⁻ and SO₄²⁻) across the CEM and the flow of K⁺ and Mg²⁺ across the AEM could not be restricted. Therefore, the product MgCl₂ was contaminated with 0.113% SO₄²⁻ and 1.18% K⁺ ions. On the other hand, other product (K₂SO₄) was contaminated with 0.303% Cl⁻ and 0.016% Mg²⁺.²⁰ Recently, Zhang et al. reported the preparation of K₂SO₄ from the KCl and ammonium sulfate mixture by mED process using commercial membranes from Hefei Chemjoy Polymer Materials Co. Ltd.²¹ The process was cost effective but the product K₂SO₄ was ca. 60.54% pure. Therefore this K₂SO₄ could not be used as fertilizer for highly chlorine sensitive crops.²¹

We propose to generate relatively pure and concentrated K₂SO₄ fertilizer by combination of mED and nanofiltration (NF) techniques. The later process is the emerging technology which uses thin film composite (TFC) membrane for selective separation of divalent and monovalent ions.^22–26 Ge et al. reported the use of NF membrane as CEM in the ED stack for enhancement of permselectivity i.e. effective separation of monovalent ions from bivalent ions.²⁷ State of art polypiperazineamide [poly(PIP)] membrane is widely used for NF of water for removal of bivalent anionic salt from monovalent ions.^28,29 This membrane has capability of high SO₄²⁻ rejection and low Cl⁻ or Na⁺ rejection. Besides this, the membrane is capable of operating at low pressure. There is so far no report of the use of both mED and NF processes for the preparation and concentration of K₂SO₄. We propose here, mED for the preparation of K₂SO₄ followed by purification and re-concentration of the generated K₂SO₄ by NF process in an aim to use the concentrated K₂SO₄ solution as fertilizer. The purity of the obtained K₂SO₄ should be >99% to be use as an fertilizer.³⁰

Herein, we report the use of mED process to convert low value potassic fertilizer (KCl) to high value potassic fertilizer (K₂SO₄) using indigenous membranes (polyethylene–polystyrene interpolymer based CEM and polyethylene–poly-4-methylstyrene interpolymer based AEM) and commercial membranes (Ionsep). The high concentration of K₂SO₄ has been achieved using interpolymer based membranes. The K₂SO₄ obtained by mED process contains 1.2–4.8% impurity depending the type of membrane used for mED process, which were further purified and concentrated by permeating through the NF membrane. This NF process also increased the purity of K₂SO₄ to about 99% by removing Na⁺ and Cl⁻ ions from the product. NF parameters such as, operating pressure and pH as well as mED parameters such as ratio of reactants, concentration of reactant and applied voltage were adjusted to obtain best results. Suitable, conditions of mED and NF gave highly concentrated and highly pure K₂SO₄ solution. We carried out this work in the scale of 5 L solution and the final product concentration achieved was 8%. The drying cost of K₂SO₄ solution of concentration 8% or more by solar evaporation is reported as an economical process.³¹ Therefore, the product solution can be solidified by evaporating the salt solution in solar concentrator. This approach thus shows promises as sustainable process for the production of purified K₂SO₄ for use as a fertilizer using membrane based technology.

Experimental

Materials

4-Methylstyrene (96%) and divinyl benzene (80%) were purchased from Sigma-Aldrich Chemicals and was used as received. Styrene, N-bromosuccinimide and benzoyl peroxide was purchased from TCI, Japan. Film grade high density polyethylene (F46 grade) and linear low density polyethylene (F19 grade) were purchased from Reliance Industries, India. Ethylene dichloride, xylene, toluene, chlorosulfonic acid, sodium sulfate (Na₂SO₄), potassium chloride (KCl), glycerol, hexane, dimethyl formamide were purchased from S D Fine Chemicals India. Polysulfone (Psf) was purchased from Solvey, India. Piperazine (PIP), trimesoyl chloride (TMC) was purchased from Aldrich. Commercial CEM (CEM_Ionsep) and AEM (AEM_Ionsep) were purchased from Ionsep (China). Nonwoven fabric (TS-100, France) was used as received.

Characterization

Preparation of interpolymer based CEM and AEM. The polyethylene–polystyrene interpolymer based CEM (CEM_inter) and polyethylene–poly(4-methylstyrene) interpolymer based AEM (AEM_inter) have been prepared using the procedure reported earlier by our group.¹³

Physical and electrochemical characterization of CEMs and AEMs. The physical characterization such as water uptake was determined by measuring the change of weight of membrane in water and in dry state. Electrochemical characterization such as ion-exchange capacity (IEC), membrane conductivity (K^m) of CEM_inter, CEM_Ionsep, AEM_inter and AEM_Ionsep were determined using the same process as reported earlier by us. The transport number (t⁺ or t⁻) of the above membranes were determined using KCl, NaCl, Na₂SO₄, K₂SO₄ of concentration 0.1 M and 0.01 M following the same procedure used by us earlier.^10–15 All the detailed characterization processes such as determination of water uptake, IEC, K^m, t has been mentioned in ESI.†

Production of K₂SO₄ by mED process. The conversion of KCl to K₂SO₄ using AEM_inter and CEM_inter was performed by mED using an in-house prepared mED cell. Fig. 1 shows the mED setup and the membrane configuration in the cell. The electrode housing, cathode and anode were made of rigid PVC sheet, stainless steel and titanium tantalum respectively. The membranes are kept inside gaskets made of HDPE sheet of thickness 0.4 mm. The spacer gaskets made of HDPE sheet of thickness 0.8 mm were used to separate AEM_inter and CEM_inter. A parallel-cum series flow arrangement of membranes was used in the mED unit. Peristaltic pumps were used to recirculate the inlet and outlet streams. Six compartments among which two reactant compartments (2 and 4), two product compartments (3 and 5) and two electrode wash compartments (1 and 9) are present in the mED cell. Both the electrode wash compartments were interconnected and 0.1% Na₂SO₄ solution was circulated through the electrode wash compartments. A predetermined DC electrical potential was applied between the electrodes by means of an AC–DC rectifier. The mED experiments were performed in mED unit of effective area 200 cm² using 24 pieces of each type of CEM and AEM. Next, KCl and Na₂SO₄ solution were circulated in compartment 2 and 4, whereas H₂O was taken in compartments 3 and 5. The initial volume of KCl, Na₂SO₄ solution and water taken in compartments 2–5 were 5 L each. After application of voltage to mED unit, NaCl and K₂SO₄ were produced in compartments 3 and 5. The concentration and mol ratio of KCl and Na₂SO₄ (compartments 2 and 4) and also the applied voltage was varied for the standardization experiments.

Fig. 1 Schematic diagram of membrane arrangement and reactants for the preparation of K₂SO₄ using mED.

Determination of concentration of ions (K⁺, Na⁺, Cl⁻ and SO₄²⁻) in compartment 5 after mED process. After mED experiment the composition of ions present in compartment 5 was determined. Concentration of produced K⁺ and Na⁺ was determined in a Digital Flame Analyzer (Cole-Parmer Instrument, Model 2655-00). The concentration of SO₄²⁻ and Cl⁻ concentration was determined by standard titration method.³¹

Determination of power consumption (W) and current efficiency (CE) during mED process. Power consumption and current efficiency are two important parameters for any electricity driven separation process. Power consumption (W in kW h kg⁻¹) during mED process is defined as the amount of energy needed to convert one kg of KCl to K₂SO₄. W has been calculated using the following equation:^10–15where V is the applied voltage; I is the current (amp); dt is the time (h) required for the mED process; and w is the weight of KCl (kg) converted to K₂SO₄.

The current efficiency (CE) is defined as the fraction of the current transported by the specific ion and has been calculated using the following equation:^10–15

where F is the Faraday constant (26.8 amp. h mol⁻¹); M is the molecular weight of KCl (58.5 g mol⁻¹), N is the number of cell pairs used in the mED unit (24 pairs), Q is the amount of electricity passed throughout the system (amp. h).

Preparation and characterization of poly(PIP) TFC NF membranes. Detailed preparation procedure of poly(PIP) TFC NF membrane was reported in our earlier publications.^22–26 Briefly, Psf support membrane of molecular weight cut-off about 100 KDa was prepared on non-woven fabric. Water wet Psf support membrane (20 cm × 20 cm) was then carefully attached on a glass slide using tape and then dipped 2% (w/v) aqueous solution of PIP for 20 s. Next, the membrane was removed from the solution and the membrane was gently rolled with a soft rubber roller to eliminate small bubbles. Then interfacial polymerization was performed by dipping the attached membrane into a 0.125% (w/v) solution of TMC in hexane for 60 s. The membrane was then heat cured at 60 °C for 2 min. Finally the membrane was washed with water and stored in glycerol–water mixture.

The membrane was characterized by attenuation total reflectance infrared spectroscopy (ATR-IR), zeta potential (ξ), atomic force microscopy (AFM) and scanning electron microscopy (SEM) using the process reported by us earlier.^22,26 The permeability coefficient (L_p), effective pore radius (r_p) pore structure factor (l_p/ε_p) (where, l_p = thickness and ε_p = porosity) were calculated from the Hagen–Poiseuille pore flow model as described earlier.^22,26

Concentration and purification of K₂SO₄ solution obtained in mED by NF process. Standard procedure was used for the evaluation of performance.^22–26 The performance of the NF membrane was evaluated in a reverse osmosis kit consisting four numbers of parallel filtration kits. The NF performance of the membrane was first evaluated using 1500 mg L⁻¹ concentration of NaCl, KCl, Na₂SO₄, K₂SO₄ and combination of some of these salts separately as feed solutions at 0.5 MPa after obtaining steady flux through initial pressurization at 0.7 MPa for 1 h. The permeate flux was calculated by using the following equation:where J is the permeate flux (L m⁻² h⁻¹); V is the volume of water permeated (L), A is the membrane area (m²) and t is the permeate time (h). The salt concentrations in the feed and permeate were determined by measuring the electrical conductivity of the solutions using digital conductivity meter (EuTech Instrument, Con 700). The salt rejection (SR) was determined using the following expression:where C_f and C_p stands for the salt concentrations in the feed and permeate, respectively. Averages of 6–8 swatches were taken with standard deviation.

For the concentration of K₂SO₄ obtained in mED process, the solution about (10 L) was taken in bucket as feed solution. The test kit was completely washed with water and the pressure pump was completely evacuated for running the experiment. This feed was then permeated through the membranes in the parallel setup. The fresh membranes were pressurized with the feed for 1 h at initial pressure 500 psi. At this point no permeate was removed. The permeate was connected with feed in circulation mode. After that permeate were collected for different times and the TDS values of the concentrate and permeate were measured. Separate sets of experiments were conducted by varying the applied pressure. The feed pH was also varied at fixed applied pressure. The standardization experiment revealed that 450 psi applied pressure and feed pH about 8.5 is suitable for further concentration and purification of feed obtained from mED process.

Thus finally, separate K₂SO₄ solutions (pH adjusted to 8.5) obtained in mED process were subjected to NF operation at applied pressure 450 psi. The permeates were continuously removed from the four parallel filtration kits. The TDS of feed and permeates after desirable NF time was measured. The permeate flux was also recorded. After certain time interval the concentration of reject reached to a very high value. For example, K₂SO₄ solutions of initial concentration of 46 122 mg L⁻¹ obtained from mED were subjected to NF. After continuous removal of permeates from 46 122 mg L⁻¹ for 16 h, the concentration of this solution was reached to 67 200 mg L⁻¹. After filtration for 37 h, the feed of concentration 46 122 mg L⁻¹ was finally concentrated to 80 000 mg L⁻¹. The compositions (individual ion concentrations) of this concentrate solution was obtained by standard titration process.³²

Determination of X-ray diffraction pattern (XRD) of K₂SO₄ crystal. The XRD study of commercial grade K₂SO₄, KCl, NaCl, Na₂SO₄ and the K₂SO₄ produced by only mED process as well as by combination of both mED and NF process were determined in a X-ray diffraction meter (model Philips X'pert MPD System).

Results and discussions

Combination of mED and NF process for production and concentration of K₂SO₄

The stream of KCl and Na₂SO₄ were circulated in mED stack as mentioned in Fig. 1. Different KCl to Na₂SO₄ (mol mol⁻¹) ratio were used for the mED process. The product streams obtained in compartment 5 by mED were then subjected to NF process to obtain relatively pure and concentrated product. Scheme 1 summarizes the whole process for the production, purification and further concentration of K₂SO₄. Scheme 1 also shows the different optimized parameters for the production process.

Scheme 1 Summarized processes for the production and purification of K₂SO₄.

Physical and electrochemical characterizations of ion-exchange membranes

Two different pairs of membranes viz. (i) CEM_inter/AEM_inter and (ii) CEM_Ionsep/AEM_Ionsep were employed in mED unit for the conversion of KCl to K₂SO₄. AEM_inter is polyethylene based cross-linked network of poly(4-methyl styrene) and polydivinyl benzene in which the 4-methyl moieties were converted to quaternized amine groups by benzylic bromination with N-bromosuccinimide and benzoyl peroxide followed by reaction with trimethylamine using the process reported earlier by us.¹³ The detailed procedure for the preparation of the interpolymer based membranes starting from polymer preparation has been described as flow chart in ESI (Fig. S1, ESI†). The characterization of CEM_inter and AEM_inter by FT-IR spectroscopy has also been mentioned in the ESI (Fig. S2, ESI†).¹⁵ We have earlier reported the comparative water uptake, IEC, K^m and t values of all the membranes which are used in this work.¹³ Table 1 additionally summarizes the t values of different ion-exchange membranes in different electrolyte solutions which influence the mED process.

Table 1 Physical and electrochemical properties of CEMs and AEMs

Membrane	Thickness (mm)	Water uptake (%)	IEC (meq g^−1)	K^m (mS cm^−1)	t^+/t^− in different electrolyte
					KCl	K2SO4	Na2SO4	NaCl
CEMinter	0.2	29	2.45	2.86	0.95	0.94	0.96	0.96
AEMinter	0.2	20	1.30	1.12	0.96	0.83	0.83	0.96
CEMIonsep	0.4	40	2.00	3.70	0.90	0.90	0.89	0.90
AEMIonsep	0.4	40	2.20	3.20	0.87	0.80	0.79	0.88

---

The average water uptake by the inter-polymer-based membrane pair CEM_inter/AEM_inter (24.5%) is less than that of CEM_Ionsep/AEM_Ionsep (40%).¹⁵ On the other hand, the average IEC values for these membranes are 1.875 meq g⁻¹ and 2.1 meq g⁻¹ respectively whereas average t [(t⁺ + t⁻)/2] values were 0.955, 0.885 when KCl was used as an electrolyte.¹⁵ This suggests that the transport property of these membranes is different. The average t values obtained with Na₂SO₄ were 0.895 and 0.84 for interpolymer and Ionsep membranes. Hence, inter-polymer membrane pair shows better electrochemical properties in terms of average t value which is extremely important for mED application. The higher average t value of inter-polymer membrane pair compared to Ionsep membrane pair is usually attributed to the lowering of water uptake by the former membrane pair which reduces the back diffusion of ions during transport. This phenomenon was reported by us and others previously.^{10–12,33,34} The membranes pairs were used for mED process for the production of K₂SO₄ as discussed below.

Production of K₂SO₄ from KCl and Na₂SO₄ by mED process

K₂SO₄ was prepared by mED process by the reaction between KCl and Na₂SO₄ using two different pairs of ion-exchange membranes (indigenous polyethylene interpolymer based membrane and commercial Ionsep membrane). The KCl to Na₂SO₄ (mol mol⁻¹) ratio and applied potential were optimized for mED process. KCl and Na₂SO₄ were separately circulated in compartments 2 and 4 whereas ultrapure water was separately circulated in compartments 3 and 5. During mED process, NaCl and K₂SO₄ are produced in compartments 3 and 5 respectively (Fig. 1). The Ionsep and interpolymer membranes are not totally permselective, some co-ion leakage occurs through the membranes. The mixing of Cl⁻ ion in product compartment (compartment 5) occurred due to passage of Cl⁻ ion through CEM in the form of a co-ion leakage from compartment 6 (where KCl concentration decreases with time) to compartment 5. Cl⁻ ion also passes from compartment 3 (where NaCl concentration increases with time) to compartment 4 (which contains diluted Na₂SO₄). Owing to higher t⁻ value, Cl⁻ ion passes from compartment 4 through AEM at a faster rate than SO₄²⁻ to compartment 5. Therefore, the product compartment (compartment 5) was contaminated with Cl⁻ ion. Such co-ions leakage in mED process was reported earlier.²⁰ Such a leakage of small amount of co-ions reduces the purity of the targeted product.

Effect of applied potential on K₂SO₄ production

Now we focus our attention to optimize the applied potential to obtain best yield of K₂SO₄. The required suitable applied potential for maximum conversion of KCl to K₂SO₄ was evaluated using commercial Ionsep membrane (24 cell pairs) in a mED unit of effective membrane area 200 cm² using mED stack, designed as discussed earlier (Fig. 1). The concentration of KCl and Na₂SO₄ were 0.5 M and 0.25 M which is the stoichiometric ratio for the chemical conversion of KCl to K₂SO₄. The volume of electrolyte solutions and water taken were 1 L in each compartment. The applied voltage was varied from 1.2–2 volt per cell pair to safely conduct the mED experiments. Fig. 2 A shows the current density vs. time plots at three different applied potentials using commercial Ionsep membrane.

Fig. 2 (A) Current density vs. time plots and (B) variation of produced SO₄²⁻ to Cl⁻ concentration ratio with applied potential during mED process using commercial Ionsep membranes. The KCl : Na₂SO₄ concentration ratio is 2 : 1. Number of each type membranes (CEM and AEM) is 24, effective membrane area 200 cm². Initial volume of electrolyte and water taken was 1 L in each compartment.

It is clear from the Fig. 2A that up to 10 minutes of mED process, with the increase of applied potential from 1.2 to 2.0 volt per cell pair current density increases then current density decreases with time. At the start of the mED experiment, the concentrations of produced K₂SO₄ and NaCl were very low. Therefore, the resistance of mED unit was high. Hence, the initial current density was low. K⁺ and SO₄²⁻ ions are transported by the diffusion and migration although the extent of ions transferred by diffusion is much lower than that transferred by migration into the compartments 3 and 5 which were initially filled with water. With the progress of time, the concentration of ions in compartment 3 and 5 increases therefore concentration difference in reactant and product compartment decreases and hence current density again increases (up to 10 minutes). After that, current density again decreases due to dissociation of reactant electrolytes and migration of the ions into the product compartments. Therefore, increase in concentration in product compartment and decrease in concentration in reactant compartment was observed. When the reactant solutions were very diluted, current was very less and the operation of mED unit was stopped.

The purity of K₂SO₄ was calculated by measuring of SO₄²⁻ to Cl⁻ concentration ratio. For example, final concentration of SO₄²⁻ and Cl⁻ were 0.18 M and 0.004 M, and the SO₄²⁻ to Cl⁻ concentration ratio was 45 : 1 (Cl⁻ ion contamination is 2.17%) at 1.2 volt per cell pair applied potential. The concentration of SO₄²⁻ and Cl⁻ were 0.195 M and 0.003 M and the SO₄²⁻ to Cl⁻ concentration ratio was 65 : 1 (Cl⁻ impurity is 1.54%) at 1.5 volt per cell pair applied potential (Fig. 2B). Presumably, with the increase of applied potential from 1.2 to 1.5 volt per cell pair, the current density increases, the movement of ions becomes faster, hence the SO₄²⁻ to Cl⁻ ratio increases. After increasing the applied potential from 1.5 to 2 volt per cell pair, SO₄²⁻ to Cl⁻ concentration ratio was reduced from 62 : 1 to 33.3 : 1 (0.20 M SO₄²⁻ and 0.006 M Cl⁻). At this applied potential, concentration of Cl⁻ ion also increased (1.54–2.91%). At relatively higher applied potential (2 volt per cell pair) the current density increases (11.2–11.8 mA cm⁻²). Hence, transportation of ions throughout the membrane from one compartment to other compartment was relatively faster. Since, the t_Cl⁻ > t_SO4²⁻ for the membranes (Table 1), the product compartment contains high quantity of Cl⁻ as impurity with SO₄²⁻ at applied potential 2 volt per cell pair. Hence, mED experiments were conducted at 1.5 volt per cell pair applied potential to obtain best result.

Effect of KCl to Na₂SO₄ ratio (mol mol⁻¹) on the mED performance

KCl and Na₂SO₄ are used as reactants for the production of K₂SO₄ by mED. NaCl is produced as byproduct as shown below.

2KCl + Na₂SO₄ → K₂SO₄ + 2NaCl

Thus the required KCl to Na₂SO₄ ratio is 2 : 1 (stoichiometric ratio) for completion of the process. However, in the membrane based production process there is change of ideal situation owing to different extent of ion transport which alters the reactant concentration accordingly. Hence, three different KCl to Na₂SO₄ mol ratios (2 : 1, 1 : 1 and 3 : 1) were used for evaluating the best mol ratio of reactants for obtaining maximum production of K₂SO₄ at standardized applied voltage viz. 1.5 volt per cell pair.

The obtained current density was almost constant (Fig. 3A) in all three different salt ratio used for the preparation of K₂SO₄ by mED process. It is observed from Fig. 3B that when KCl to Na₂SO₄ concentration ratios was 1 : 1, the obtained SO₄²⁻ to Cl⁻ ratio was 85 : 1 (impurity 1.16%). On the other hand, obtained SO₄²⁻ to Cl⁻ ratios were 62 : 1 (impurity 1.58%) and 50 : 1 (impurity 1.96%) when used KCl to Na₂SO₄ concentration ratios in mED process were 2 : 1 and 3 : 1 respectively at applied potential 1.5 volt per cell pair. Therefore, it can be concluded that 1 : 1 (KCl : Na₂SO₄) salt ratio produced maximum purified SO₄²⁻ ion (1.16% impurity). The Cl⁻ ion impurity reduced from 1.58% to 1.16% when 1 : 1 salt ratio of KCl and Na₂SO₄ was used. This type of complex observation may be attributed due to higher t⁻ value of Cl⁻ ion than that of SO₄²⁻ ion (Table 1). Therefore, when high KCl to Na₂SO₄ salt concentration ratio in feed mixture was used, the concentration of contaminated Cl⁻ in produced SO₄²⁻ was maximum. The membranes used for mED process are not 100% cation or anion selective. Some co-ion leakage is also occurring during mED process as discussed earlier. Cl⁻ ion from compartment 6 passes through CEM and enters into the compartment 5. Similarly, Na⁺ ion from compartment 4 also passes through AEM and comes into compartment 5. Therefore, compartment 5 contains Na⁺ and Cl⁻ ions as impurity.²⁰

Fig. 3 (A) Current density vs. time plots and (B) bar diagram showing variation of produced SO₄²⁻ to Cl⁻ concentration ratio with variation of feed KCl to Na₂SO₄ concentration ratio used during mED using Ionsep membranes at applied potential 1.5 volt per cell pair. Number of each type membranes (CEM and AEM) is 24, effective membrane area 200 cm². Initial volume of electrolyte and water taken was 1 L in each compartment.

Comparative mED results with Ionsep and interpolymer membranes

The mED experiments were carried out using commercial Ionsep and indigenously developed interpolymer based membranes at standardized 1.5 volt per cell pair applied potential and 1 : 1 molar ratio of salts for the production of K₂SO₄. The initial concentration of KCl in feed solution was increased from 0.4 M to 1.4 M. Fig. 4 A shows the concentration of produced K₂SO₄ with time during mED process using interpolymer based membrane. Similarly, Fig. 4B shows the concentration of produced K₂SO₄ with time during mED process using commercial Ionsep membrane.

Fig. 4 Dependence of K₂SO₄ concentration with time during mED process at different salt concentrations. Plot (A) obtained with indigenous interpolymer based membrane and plot (B) obtained with commercial Ionsep membrane. Applied potential = 1.5 volt per cell pair. Number of each type membranes (CEM and AEM) = 24, effective membrane area = 200 cm². Initial volume of electrolyte and water taken was 5 L in each compartment.

Fig. 4A and B shows that maximum 0.37 M and 0.46 M K₂SO₄ was produced with commercial Ionsep and indigenous interpolymer based membranes respectively. After the mED experiment was stopped, 20% increase in volume in the product compartment was observed. This increase in volume change indicates electro-osmostic water transport along with potassium ions through CEM. Therefore high water transport diluted the produced K₂SO₄ solution.¹⁸ Subsequently, co-ion transport decreases the purity of the product. The produced K₂SO₄ contained 1.2–4% Cl⁻ ion impurity when mED experiments were carried out with interpolymer membranes as confirmed by ion analysis. On the other hand when the mED experiments were conducted with Ionsep membrane the produced K₂SO₄ contains 2.1–4.8% Cl⁻ ion as impurity. The amount of impurity increases with increasing feed KCl concentration from 0.4 M to 1.4 M with both types of membranes. The higher concentration of K₂SO₄ with higher purity obtained with interpolymer membranes than Ionsep membrane may be due to low water uptake and high t values of this membrane than Ionsep membrane. As discussed earlier, t_Cl⁻ > t_SO4²⁻ hence, transportation of Cl⁻ becomes faster through the membranes when high KCl concentration was used in the feed solution. pH of all the compartments (2–5) were acetic (3–4). This is due to the non selective transportation of H⁺ ion from electrode wash compartments throughout the mED stack to all the compartments.¹⁹

Determination of W and CE (%) during mED process with Ionsep and interpolymer membranes

W and CE are two important parameters to determine the suitability of mED process. Therefore, W and CE (%) values have been determined during K₂SO₄ production from KCl by mED process using interpolymer and Ionsep membranes. Table 2 shows the calculated W and CE (%) values for the different experiments.

Table 2 W and CE vales obtained during production of K₂SO₄ by mED process using Ionsep and interpolymer membranes at 1.5 volt per cell pair applied potential

Types of membrane used	Final concentration of K2SO4 (M)	% of Cl^− impurity	W (kW h kg^−1)	CE (%)
Interpolymer	0.175	1.20	0.80	75
Interpolymer	0.335	2.45	0.92	74
Interpolymer	0.410	3.75	1.15	72
Interpolymer	0.460	4.00	1.28	70
Ionsep	0.140	2.10	0.97	74
Ionsep	0.202	2.50	1.18	73
Ionsep	0.265	3.00	1.28	70
Ionsep	0.370	4.80	1.42	67

---

The above results indicate that W value increases from 0.80 kW h kg⁻¹ to 1.28 kW h kg⁻¹ and CE value decreases from 75 to 70 as the final concentration of K₂SO₄ increases from 0.175 M to 0.46 M during mED process with interpolymer membranes. Similarly, W value increases from 0.97 kW h kg⁻¹ to 1.42 kW h kg⁻¹ and CE value decreases from 74 to 67 as the final concentration of K₂SO₄ increases from 0.14 M to 0.37 M during mED process with Ionsep membranes. The obtained W value was lower and CE value was higher during mED process using interpolymer membrane. This is due to low water uptake and higher counter ion t number of the interpolymer membrane than that of Ionsep membrane.

Purification and concentration of solutions obtained from compartment 5 via the NF process using PIP-based TFC NF membrane

The K₂SO₄ (0.265 M) obtained by mED process contains 3% Cl⁻ ion as impurity when the mED was carried out Ionsep membrane. Therefore, the produced K₂SO₄ may not suitable as a fertilizer for non chlorine tolerating plants with this high level of impurity. As explained above (Fig. 4) the amount of impurity increases with the increase in concentration of produced K₂SO₄ by mED process due to co-ion leakage and electro osmotic drag through both types of CEM and AEM. Not only the Cl⁻ impurity, W value increases and CE (%) decreases when high K₂SO₄ concentration (Table 2) was prepared by mED process. Hence, we restricted the preparation of K₂SO₄ to concentration 0.265 M (4.6%) by mED process. On the other hand, the solar drying of K₂SO₄ solution is not economical when the solution concentration is <8% w/v (0.46 M K₂SO₄). Hence, to make the process sustainable and produced fertilizer as plant tolerant, further concentration and purification of produced K₂SO₄ was carried out by further subjected to NF operation using suitable poly(PIP) based TFC NF membrane. Several parameters such as conditions of poly(PIP) membrane preparation, feed solution pH and applied pressure were adjusted to obtain best permeate flux as well as concentrate with high content of K₂SO₄.

Preparation and characterization of NF membrane

The TFC NF membrane should contain high surface negative charge, high rejection of bivalent anionic salt and low rejection of monovalent salts. TFC NF membrane was prepared by interfacial polymerization between 2% (PIP) in water and 0.125% TMC in hexane. This is the condition to obtain NF membrane with high rejection of bivalent anion and low rejection monovalent ions. When the concentration of PIP or TMC was higher than the above mentioned concentration then the rejection of monovalent ions enhanced. For example, our PIP-based membranes can effectively remove the SO₄²⁻ salts (92–93%) whereas it shows low rejection of K⁺, Na⁺, Cl⁻ (30–35%). This membrane also exhibited good separation ability of monovalent to divalent salts during seawater NF. For example the rejection of SO₄²⁻ was ca. 80% whereas the rejection of Na⁺ and K⁺ was as low as ca. 11% during seawater NF at operating pressure 1.4 MPa at normal pH of seawater (pH = 7.4).²⁶ Poly(PIP) TFC NF membrane was characterized by SEM, AFM and IR spectroscopy (Fig. S3A, ESI†). The SEM image shows typical globular morphology of the poly(PIP) membrane. Such type of morphology was discussed previously.^22,26 The AFM image of the membrane (Fig. S3B, ESI†) also show typical ridge-valley type of structure as commonly observed for the TFC membrane. Characteristic IR band (Fig. S3C, ESI†) for amide I (–C=O) stretching vibration appears at 1620 cm⁻¹. The band at 1585 cm⁻¹ appears due to C=C aromatic vibration of Psf base membrane. The pore radius of the membrane was 0.6 nm and surface zeta potential was −20 to −25 mV at pH 7–8.

Effect of applied pressure on NF performance

The rejection efficiency of this NF membrane can be tuned by adjusting the pressure and feed pH (vide infra). The product obtained in compartment 5 not only contains K₂SO₄ but may contain Cl⁻ ion as impurity (Table 2). Therefore, NF experiments were carried out to remove the monovalent ions from the mixtures. Standardization experiments were conducted based on applied pressure to obtain both high flux and high rejection of salts. The pH of the feed solution was kept at 7.4. Fig. 5A shows change of permeate flux with applied pressure for different feed solutions obtained in mED process with varying K₂SO₄ concentration. Fig. 5B shows increase of total rejection (reduction of total dissolved solid TDS) with applied pressure.

Fig. 5 Variation of (A) permeate flux and (B) total rejection with applied pressure during NF of feed obtained from mED process. The feed pH was 7.4 during NF operation. The membranes were pressurized with the same feed for 40 min at 500 psi before collection of NF at 450 psi.

It is seen that maximum flux and rejection were obtained at operating pressure ca. 450 psi. Hence, 450 psi pressure is suitable for concentrating the feeds obtained in mED process. The enhanced flux is due to enhanced rate of permeation. The higher rejection with applied pressure is attributed to the high rate of permeation of water than the salts. This behaviour is well documental in the literatures.³⁵

Effect of feed pH and feed concentration on NF performance during concentration of product obtained in mED process

The NF experiments were also conducted at three different feed pH at applied pressure 450 psi with feed K₂SO₄ solution of concentration 0.265 M prepared with Ionsep membranes. The flux remained almost unaffected whereas the total rejection increased with increasing pH of the feed solution (Fig. 6A). The poly(PIP) membrane is negatively charged. With increasing feed pH the surface negative charge of poly(PIP) membrane increases as confirmed by zeta potential measurement. Hence, with increasing surface charge, the charge–charge repulsion between membrane surface and SO₄²⁻ enhances. Thus, enhanced Donnan exclusion with enhanced surface charge occurred. Since, SO₄²⁻ is the major components in the feed, the total rejection was also enhanced.

Fig. 6 Bar diagram (A) variation of permeate flux and rejection with variation of feed pH during NF of feed solution obtained from mED experiment (0.265 M K₂SO₄ and 3% Cl⁻ ion as impurity). Bar diagram (B) variation of permeate flux and rejection with variation of concentration of feed (pH 8.4). The applied pressure was 450 psi. The permeates were collected after 10 min of filtration. The membranes were pressurized with the same feed for 40 min at 500 psi before collection of NF at 450 psi.

The feed concentration also affects the NF performance. Fig. 6B shows the effect of feed concentration (starting concentration 0.265 M) on NF performance. The continuous NF operation shows that the concentrate reached up to about 0.46 M after 37 h of filtration. The flux and rejection at different time was measured. The decreased permeate flux with time or concentration of NF process was seen. As a result, about 2 fold lowering of permeates flux was observed after 37 h of NF operation (Fig. 6B). The amount of Cl⁻ impurity was also lowered (1%). The rejection remained more or less similar. The lowering of flux with increasing feed concentration is simply due to enhanced osmotic resistant on the membrane against fixed applied pressure. It is important to mention here that K₂SO₄ concentration of 0.37 M and 0.46 M obtained with Ionsep and interpolymer membranes by mED process were not used to further purify and concentrate by NF process. This is due to the fact that at very high feed concentration the permeate flux was decreased.

Determination of the purity of K₂SO₄ by XRD analysis

Solid K₂SO₄ was obtained by the evaporation of water. The XRD spectra of solid K₂SO₄ as obtained by evaporation of solution obtained by dual mED and NF process (Fig. 7, spectrum d) and evaporation of solution obtained by single mED process (Fig. 7, spectrum c) were compared. Fig. 7 also shows the XRD patterns of commercial grade K₂SO₄ (spectrum b) and pure KCl crystals (spectrum a) for comparison purpose.

Fig. 7 XRD pattern of K₂SO₄ produced by only mED process (c) and by combination of mED and NF processes (d). XRD patterns of commercial grade K₂SO₄ (b) and KCl (a) have also been included. Inset shows the diffraction pattern in enlarged scale (28–34°) of purified K₂SO₄ produced by combined mED and NF processes.

Clearly, Fig. 7a indicates that the diffraction peaks appears at 2θ values 28.5, 40.90, 50.37 and 66.67° for commercial grade KCl whereas for commercial grade K₂SO₄ (Fig. 7b), the peaks appear at 21.6, 30, 30.9 and 43.4° respectively. On the other hand peaks appear at 2θ values (Fig. S4a ESI†) 19.39, 23.63, 28.37, 29.59, 32.52, 34.25, 38.91, 48.92 and 59.82° for pure Na₂SO₄. The 2θ value of pure NaCl (Fig. S4b ESI†) appeared at 31.83, 45.64 and 66.38°. The XRD patterns (c) and (d) show characteristic peaks at 2θ values 21.6, 30.39, 31.6, 43.4°, due to presence of K₂SO₄. On the other hand, in the spectrum (c), small intensity peaks at 2θ values 28.5 and 40.9° appeared due to presence of KCl which was further reduced in spectrum (d) where additional purification was conducted by NF operation. Therefore, in the final product (both prepared by only mED and by combination of mED and NF) very little contamination with KCl was observed. Additionally, in the spectrum (c), the intensity ratio of peaks at 2θ value 31.5° to 28.5° is 5.96. On the other hand, in the pattern (d), the intensity ratio of peaks at 2θ values 31.6° to 28.5° is 10.18. This clearly proves the purity of K₂SO₄ produced by the combination of mED and NF processes than produced by only mED process.

Conclusion

Best condition for the production of high value potassic fertilizer (K₂SO₄) by metathesis electrodialysis has been identified. The optimization experiments revealed that 1 : 1 mol ratio of reactant salts (KCl and Na₂SO₄) concentration and 1.5 volt per cell pair applied potential gave best results in terms of purity of obtained K₂SO₄ and product concentration. Among the two sets of ion-exchange membranes, the interpolymer-based cation exchange and anion exchange membrane provided better results than the commercial Ionsep membrane pair. The better performance with former set of membranes is attributed to the high transport number of the membranes in reactant electrolytes compared to later membranes. The co-existed impurities e.g. Cl⁻ and Na⁺ ions were further removed by nanofiltration of the obtained product solution with poly(piperazineamide) membrane under suitable experimental conditions such as at applied pressure 450 psi and feed pH ca. 8.5. The nanofiltration of the product not only enhanced the purity of the K₂SO₄ (>99%) but also enhanced its concentration for further obtaining the product in solid state by solar induced evaporation step. Therefore, the produced K₂SO₄ can be used as fertilizer to crops. The process is scalable and the process can be an efficient process for the production of K₂SO₄.

Acknowledgements

Uma Chatterjee acknowledges DST, Govt of India (YSS/2015/000653) for financial support under Young Scientist scheme. V. Bhadja and J. S. Trivedi acknowledge CSIR network projects (CSC0104 and CSC0105) for providing fellowship. Mr M. N. Parmar is acknowledged for helping in packing the mED stack. We acknowledge centralized analytical facility of CSMCRI for analytical support. CSIR-CSMCRI registration number 051/2016.

Notes and references

1. J. S. Trivedi, V. Bhadja, B. S. Makwana, S. K. Jewrajka and U. Chatterjee, Making chloride-free potash fertilizers, Phosphorus Potassium, 1988, 156, 8.
2. B. Grzmil and B. Kic, Chem. Pap., 2005, 59, 476.
3. M. Tomaszewska, J. Membr. Sci., 2008, 317, 14.
4. R. E. Worthington, A. Magdics and D. B. Stain, US Pat., No. 4,588,573A, 1986.
5. I. Efraim, S. Lampert and C. Holdengraber, US Pat., No. 5,552,126 A, 1996.
6. T. Zisner, C. Holdengraber and S. Lampert, US Pat., No. 5,549,876A, 1996.
7. H. Iwashita and K. Hayashi, US Pat., No. 4,342,737 A, 1982.
8. R. Phinney, US Pat., no 6315976 B1, 2011.
9. O. Arar, E. Yavuz, Ü. Yüksel and N. Kabay, Sep. Sci. Technol., 2009, 44, 1562.
10. V. Bhadja, U. Chatterjee and S. K. Jewrajka, RSC Adv., 2015, 5, 40026.
11. U. Chatterjee and S. K. Jewrajka, J. Mater. Chem. A, 2014, 2, 8396.
12. U. Chatterjee, V. Bhadja and S. K. Jewrajka, J. Mater. Chem. A, 2014, 2, 16124.
13. V. Kulshrestha, U. Chatterjee, S. Sharma, B. S. Makwana and P. D. Maru, Macromol. Symp., 2015, 357, 194.
14. V. Bhadja, B. S. Makwana, S. Maiti, S. Sharma and U. Chatterjee, Ind. Eng. Chem. Res., 2015, 54, 10974.
15. V. Bhadja, J. S. Trivedi and U. Chatterjee, RSC Adv., 2016, 6, 67118.
16. H. Jaroszek, A. Lis and P. Dydo, Bull. Pol. Acad. Sci.: Tech. Sci., 2014, 119, 351.
17. H. Jaroszek, A. Lis and P. Dydo, Sep. Purif. Technol., 2016, 158, 87.
18. S. K. Thampy, B. S. Joshi and K. P. Govindan, Indian J. Technol., 1985, 12, 454.
19. C. Alhéritière, W. R. Ernst and T. A. Davis, Desalination, 1998, 115, 189.
20. B. Pisarska, MgSO₄ + 2KCl → K₂SO₄ + MgCl₂, Desalination, 2008, 230, 298.
21. X. Zhang, X. Wang, X. Liu, X. Han, C. Jiang, Q. Li and T. Xu, Ind. Eng. Chem. Res., 2015, 54, 11937.
22. R. M. Gol, A. Bera, S. Banjo, B. Ganguly and S. K. Jewrajka, J. Membr. Sci., 2014, 472, 154.
23. S. U. Hong and M. L. Bruening, J. Membr. Sci., 2006, 280, 1.
24. V. Silva, M. Montalvillo, F. J. Carmona, L. Palacio, A. Hernández and P. Prádanos, Desalination, 2016, 382, 1.
25. Y. Lv, H.-C. Yang, H.-Q. Liang, L.-S. Wan and Z.-K. Xu, J. Membr. Sci., 2016, 500, 265.
26. A. Bera and S. K. Jewrajka, RSC Adv., 2016, 6, 4521.
27. L. Ge, B. Wu, Q. Li, Y. Wang, D. Yu, L. Wu, J. Pan, J. Miao and T. Xu, J. Membr. Sci., 2016, 498, 192.
28. R. J. Peterson, J. Membr. Sci., 1993, 83, 81.
29. N. Misdan, W. J. Lau, A. F. Ismail, T. Matsuura and D. Rana, Desalination, 2014, 344, 198.
30. C. Zörb, M. Senbayram and E. Peiter, J. Plant Physiol., 2014, 171, 656.
31. A. Sadan, US3966541 A, 1976.
32. A. I. Vogel, Textbook of quantitative chemical analysis. 5th edn, 1989, ch. 10.84, p. 355, chapter 11.72, page 490.
33. T. Rottiers, G. De la Marche, B. Van der Bruggen and L. Pinoy, J. Membr. Sci., 2015, 492, 263.
34. M. A. Izquierdo-Gil, V. M. Barragán, J. P. G. Villaluenga and M. P. Godino, Chem. Eng. Sci., 2012, 72, 1.
35. Y. Dai, X. Jian, S. Zhang and M. D. Guiver, J. Membr. Sci., 2002, 207, 189.

---

Footnote

† Electronic supplementary information (ESI) available: Preparation of the PE–PSt and PE/P4-MS inter-polymer film, FT-IR spectra of CEM_inter and AEM_inter, physical and electrochemical characterizations of different membranes, SEM, AFM images and ATR-IR spectra of TFC NF membrane, XRD spectrum of commercial grade Na₂SO₄ and NaCl. See DOI: 10.1039/c6ra14303b

---