Aluminum Sulfate as an Innovative Draw Solute for Forward Osmosis Desalination

  • Dhouha Ben MaouiaEmail author
  • Ali Boubakri
  • Amor Hafiane
  • Salah Bouguecha
Original Article


Forward osmosis (FO) is a membrane technology which has attracted attention for water treatment and desalination. However, selecting an appropriate draw solute is vital to optimize its performance. This study seeks the efficiency of aluminum sulfate as an alternative draw solute in FO desalination with a cellulose triacetate (CTA) membrane. The effects of operating parameters on the performance of the FO were studied such as feed and draw temperatures, concentrations and flow. The experiments revealed that the permeate flux was improved by monitoring draw temperature, with a maximum of 2.5 L/m2 h was obtained at 53 °C. Also, the permeate flux was found to decrease with feed concentration. A maximum permeate flux of 2 L/m2 h was obtained at a draw flow rate of 35 L/h and draw concentration of 1 mol/L. On the other hand, using deionized water as feed solution yielded a reverse aluminum sulfate flux of 1.46 g/m2 h. The plots of the experimental and the modeling water flux displayed analogous trends in all tests, but the results showed a large deviation which was attributed to reverse solute flux, internal polarization concentration (ICP), external polarization concentration (ECP) and membrane fouling. Precipitation reaction using calcium hydroxide served to recover product water from the diluted draw solution. This operation was carried out via a precipitation reaction of aluminum sulfate with calcium hydroxide to eliminate the soluble chemicals like insoluble aluminum hydroxide and calcium sulfate. Eventually, aluminum sulfate draw solution was recovered by the reaction of aluminum hydroxide with sulfuric acid.


Forward osmosis Desalination Aluminum sulfate Draw solution Water flux 



Osmotic coefficient


Number of anions in the electrolyte formula


Number of cations in the electrolyte formula


Charges of the anion


Charges of the cation


Overall molality of the electrolyte (mol/kg)


Third virial coefficient


Solution concentration (mg/L)


Molar mass of aluminum sulfate (kg/mol)


Draw solution density (kg/m3)


Pitzer parameter

βac(0), βac(1)

Second virial coefficients

α, b

Model parameters


Ionic strength


Debye–Hückel constant


Respective molality values of the anion (mol/kg)


Respective molality values of the cation (mol/kg)


Avogadro number

\( \uprho_{{{\text{H}}_{2} {\text{O}}}} \)

Density of water (kg/m3)


Electronic charge


Absolute temperature (K)


Water dielectric constant


Boltzmann’s constant


Osmotic pressure (Pa)


Universal gas constant (J/mol/K)


Density of aluminum sulfate draw solution (kg/m3)


Mass fraction

\( \bar{\nu } \)

Partial molar volume (m3/kg)


Water flux (L/m2 h, LMH)


The change in volume of feed solution (L)


Time interval for the volume change of ΔV (h)


Active membrane surface area (m2)


Membrane water permeability coefficient (m/s/Pa)


Osmotic pressure of the draw solution (Pa)


Osmotic pressure of the feed solution (Pa)


Reverse salt flux (g/m2 h, gMH)


Initial feed concentrations (g/L)


Final feed concentrations (g/L)


Initial volume of the feed solution (L)


Final volume of the feed solution (L)


Initial conductivity of the feed solution (S/m)


Final conductivity of the feed solution (S/m)


Molar electrical conductivity of the feed solution (S m2/mol)


Initial aluminum feed concentration (mol/m3)


Final aluminum feed concentration (mol/m3)


Concentration polarization


Internal concentration polarisation


External concentration polarisation


Feed solution


Draw solution


Forward osmosis


Cellulose triacetate membrane

1 Introduction

Due to the freshwater scarcity and the growing need for water supply, numerous regions of the world have opted for the desalination of seawater and brackish water. Membrane water treatment processes have been developed, including electro-dialysis (ED), nano-filtration (NF), reverse osmosis (RO) and membrane distillation (MD). In fact, RO is the most common desalination technique in the world but is considerably expensive [1]. Notably, all the conventional desalination technologies (thermal or membrane) are energy-consuming, which is costly both to the environment and the economy [2]. Thus, new energy-efficient and cost-effective desalination techniques have been sought.

Since the 1980s, forward osmosis (FO) was suggested as a more efficient technique. In contrast with earlier desalination methods, this simple technique relies on the difference in osmotic pressure for the permeation of a feed solution into a draw solution, which is highly concentrated one, by means of a semi-permeable membrane. The semi-permeable membrane keeps salts in the feed solution. After this operation, an energy saving and efficient technique is used to extract pure water from the diluted draw solution. This technique is preferred because of its removal effectiveness, low pressure operation, low membrane fouling, cost-effectiveness and energy-effectiveness. Compared with adsorption, the flux always declined, since the porous structure got compact and blocked by more carbon nanotubes (CNTs) and the fouling layer composed of dye deposition became thicker, which kept decreasing permeate flux [3].

A few previous studies [4] used UF to separate and concentrate crude protein from waste leaf extraction juice. However, during the concentration process, serious flux decline caused by membrane fouling, increasing the operation cost and restricting its sustainable operation and industrial application.

Despite these advantages, FO has some drawbacks, namely the significantly lower water flux obtained compared to the other available processes, the inefficiency of FO membranes and the lack of easily recyclable profitable draw solutes.

Also, the formation of biofilms on spaces hinders the system performance. These mesh-type nets are located in the feed and draw channels as turbulence promotors. Biofouling on the spacers increases the pressure drops along the feed and draw channels [5]. According to Wang et al. [6] particles tend to deposit near the “hydrodynamic dead zones” between the spacer filaments and membrane surface.

An appropriate draw agent is requisite for the success of the FO treatment, notably according to specific draw solution selection criteria [7, 8]. These criteria include an osmotic pressure which outweighs osmotic pressure of the feed water and a diluted draw solution whose separation from the water product is simple. Other parameters, including low cost, eco-friendliness, stability and the intactness of the semi-permeable membrane are equally important [9]. Thus, developing appropriate draw solutes seems a paramount stake for FO desalination. Still, in addition to power consumption, the diluted draw solution recovery has the drawback of high salt leakage. To improve the efficiency of FO, a novel draw solute has been sought to increase water flux, decrease reverse salt diffusion, and facilitate the recovery of the diluted draw solution [10, 11]. Numerous elements have been tried as draw solutions in FO [12, 13]. Batchelder et al. [14] relied on volatile solutes like sulfur dioxide for seawater desalination. They removed the volatile solutes from the product water via a heated gas stripping operation. McCutcheon and Elimelech [15] used highly soluble ammonium bicarbonate to increase water flux. This operation used moderate heating that reached 60 °C. High temperature causes the draw solutes composed of ammonium bicarbonate to decompose into gases (ammonia and carbon dioxide) and leave the solution that turns into pure product water. Adham et al. [16] employed magnetic nanoparticles as FO draw solutions to separate the product water from the draw solution by means of a magnetic separator. Magnetic nanoparticles were also used by Ge et al. [17] as draw solution but product water was recovered by applying a magnetic field. The draw solute by Ling et al. [18] was super hydrophilic nanoparticles and nanoparticles were separated from product water by means of ultrafiltration (UF). Ge et al. [19] reconcentrated polyelectrolytes of a series of poly-acrylic acid sodium salts that they used as draw solutes through a pressure-driven UF process. Stone et al. [20] mixed water, carbon dioxide and tertiary amines to obtain the draw solution. Product water was recovered by heating the draw solution to strip carbon dioxide. Interestingly, these draw solutions exhibited an acceptable water flux. However, the reverse salt flux was high and energy consumption was relatively high. In spite of the numerous attempts, draw solutes that simultaneously offer a stronger water flux, a weaker reverse salt flux, and a simpler recovery are still sought [9, 21].

In this work, we propose aluminum sulfate (Al2(SO4)3) as a novel draw solution in FO desalination because it is highly water-soluble and it can be found in natural sources. Also, it does not require energy for its regeneration. This operation does not rely on electrical energy to collect product water as the draw solution is reacted with the aqueous calcium hydroxide solution to precipitate all the soluble chemicals from the product water. Theoretical and experimental studies were made to show the efficiency of the selected draw solution by studying the influence of operating parameters on the performance of the FO concerning flux and quality of produced water. The impact of draw and feed concentrations, draw and feed temperatures, along with draw and feed flow rates were elucidated.

2 Materials and Methods

2.1 FO Membrane

A commercial flat sheet FO membrane supplied by Hydration Technologies Innovations (HTI OsMem™ CTA-NW Membrane) was used. This asymmetric membrane is made of a thin semi-permeable nonporous active skin layer of cellulose triacetate (CTA) embedded in a nylon mesh (a porous support layer). Table 1 shows its technical specifications.
Table 1

CTA membrane technical specifications

Membrane type

Cellulose triacetate (CTA) nonwoven support

Maximum operating temperature

160 °F (71 °C)

Maximum transmembrane pressure

10 psi (70 kPa), if supported

pH range


Tolerance to chlorine

2 ppm

The membrane characterization relied on the following methods:
  • FEI Quanta FEG 250 scanning electron microscope (SEM) elucidated the CTA membrane morphologies.

  • The membrane hydrophobicity character was elucidated through the measurement of the membrane contact angle (CA) with an Attension Theta optical tension-meter

  • Zygo new-view 7100 profilo-meter was used to investigate the surface roughness of the membrane and determine the three-dimensional membrane shape surface.

  • Perkin Elmer spectrum 100 FTIR spectrophotometer determined the chemical composition of the membrane.

The CA measurement (Fig. 1a) made for virgin membrane surface was 76°. The used membrane is thus hydrophilic because a contact angle of 0° entails to an optimal hydrophilic membrane surface. The rise of this contact angle improves its hydrophobicity [22, 23]. This hydrophilic character is beneficial to the permeation of water and can enhance the antifouling tendency of the membrane.
Fig. 1

a CA, b SEM, c Profilo-meter, d FTIR spectrum membrane characterization

Figure 1b illustrates the SEM image of the membrane with magnification of 40 μm, where the microstructure of the membrane surface can be easily observed. Here, the active layer of CTA-NW seems to be dense and less porous. According to Fig. 1c, the characterized membrane maximum and average surface roughness values are 10.117 and 1.560 μm, respectively. This roughness favors for a drop of fixed volume to spread by capillary action into the crevices on the surface. The dispersion supported by the hydrophilicity of the polymer. It enhances the water mass transfer by capillary in the crevices and the pores within the membrane. This consequently favors the appearance of a lower contact angle. At the same scale, the surface of asymmetric CTA-NW membranes was rather smooth, which implies lower potential of fouling propensity of the membrane.

Fourier transform infrared spectroscopy (FTIR) analysis is shown in Fig. 1d. The peaks at 925.23 cm−1 and 871.93 cm−1 are attributed to the pyranose cycle of cellulose. The CTA-NW is confirmed with the appearance of peaks for C=O, CH(CH3), C–O of acetyl groups of wave numbers 1712 cm−1, 1462.9 cm−1, 1240 cm−1, respectively.

2.2 Lab Scale Setup Front

The experiments were conducted using a flat sheet FO laboratory setup as displayed in Fig. 2. The cell is composed of two channels and a flat membrane sandwiched in between. The hydrophilic membrane has an area of 32 × 10−4 m2. The active layer of the membrane was fixed to the feed solution (NaCl) while the supported layer faced the draw solution (Al2(SO4)3) in all experiments. Both the feed and the draw solutions were pumped at a flow rate of 20 L/h on either side of the membrane using a variable speed peristaltic pump (Master flex L/S). A constant temperature bath served to keep the feed and draw solutions at constant heat during the experiment.
Fig. 2

Schematic diagram of the lab-scale FO system

2.3 Chemicals and Materials

All solutions used deionized water. Sodium chloride (NaCl, AR grade, 99.5%) and hydrated aluminum sulfate (Al2(SO4)3·18H2O, AR grade, 59% anhydrous mass) were made at the Laboratory of Reagents and Fine Chemicals. Calcium hydroxide (Ca(OH)2·8H2O, ACS grade, 98.0%) was obtained from Ismae Merck.

3 Physical Laws

3.1 Osmotic Pressure of Aluminum Sulfate Draw Solution

In FO desalination, the osmotic pressure of draw solute is among the most important water flux parameters. It can be assessed by the Van’t Hoff equation [24]. Generally, high osmotic pressure yields a higher water flux. However, in FO, the draw solution is very concentrated and, hence, the non-ideal solution behavior is important.

Pitzer model predicts the thermodynamic properties at high concentrations [25]. It is based on an ameliorated Debye–Hückel method with a virial expansion. It elucidates binary and ternary ionic-strength-dependent interactions between ions. For single electrolyte solutions, deviations from ideal solution behavior is determined as an osmotic coefficient (ϕ) defined by Pitzer as follows [26]:
$$ \phi = 1 + \left| {{{\text{Z}}_{\text{a}}} {{\text{Z}}_{\text{c }}} } \right|{\text{f}} + {\text{m}}\left( {\frac{{2\upnu_{\text{a}} \upnu_{\text{c}} }}{{\upnu_{\text{a}} + \upnu_{\text{c}} }}} \right)\upbeta_{\text{ac}} + {\text{m}}^{2} \left( {\frac{{2\left( {\upnu_{\text{a}} \upnu_{\text{c}} } \right)^{3/2} }}{{\upnu_{\text{a}} + \upnu_{\text{c}} }}} \right){\text{C}}_{\text{ac }} $$
where νa and νc refer to the numbers of anions and cations in the electrolyte formula, respectively, Za and Zc refer to the charges of the anion and the cation, respectively, m is the overall molality of the electrolyte (mol/kg), Cac is third virial coefficient. The terms f and βac are defined as follows [26]:
$$ m = \frac{C}{{M\left( {10^{3} \rho - C} \right)}} $$
Where C depicts the solution concentration (ppm), M refers to the molar mass of aluminum sulfate (kg/mol) and ρ is the draw solution density (kg/m3).
$$ C_{ac} = C^{\phi } \left( {2\left| {{\text{Z}}_{\text{a}}} {{\text{Z}}_{\text{c}} } \right|^{1/2} } \right) $$
where Cϕ is the Pitzer parameter.
$$ f = - A_{\phi } \frac{\sqrt I }{1 + b\sqrt I } $$
$$ \beta_{ac} = \beta_{ac}^{\left( 0 \right)} + \beta_{ac}^{\left( 1 \right)} exp\left( { - \alpha \sqrt I } \right) $$
where β ac (0) and β ac (1) are the second virial coefficients, α and b represent model parameters, the ionic strength (I) and the Debye–Hückel constant (Aϕ) are expressed as follows [26]:
$$ I = \frac{1}{2}\left( {m_{a} Z_{a}^{2} + m_{c} Z_{c}^{2} } \right) $$
$$ A_{\phi } = \frac{1}{3}\left( {\frac{{2\pi N_{A} \rho_{{{\text{H}}_{2} {\text{O}}}} }}{1000}} \right)^{1/2} \left( {\frac{{e^{2} }}{\varepsilon kT}} \right)^{3/2} . $$

In the above equations, ma and mb refer to the respective molality values of the anion and the cation, NA is Avogadro number, \( \rho_{{{\text{H}}_{2} {\text{O}}}} \) is the density of water, e is the electronic charge, T is the absolute temperature, ε is the water dielectric constant, and k is the Boltzmann’s constant. At 25 °C, the value of Aϕ is 0.392 [26].

From Eq. (3), osmotic pressure can be expressed as [26]:
$$ \pi = \phi \left( {\nu_{a} + \nu_{c} } \right)RTm = \left( {\nu_{a} + \nu_{c} } \right)RTm \left[ {1 + \left| {{{\text{Z}}_{\text{a}}} {{\text{Z}}_{\text{c }}} } \right|{\text{f}} + {\text{m}}\left( {\frac{{2\upnu_{\text{a}} \upnu_{\text{c}} }}{{\upnu_{\text{a}} + \upnu_{\text{c}} }}} \right)\upbeta_{\text{ac}} + {\text{m}}^{2} \left( {\frac{{2\left( {\upnu_{\text{a}} \upnu_{\text{c}} } \right)^{3/2} }}{{\upnu_{\text{a}} + \upnu_{\text{c}} }}} \right){\text{C}}_{\text{ac }} } \right] $$
where R is the universal gas constant, π is the osmotic pressure and T is the absolute temperature.
The density (ρ) of aluminum sulfate draw solution as a function of temperature and concentration is calculated as follows [27]:
$$ \rho \;({\text{kg}}/{\text{m}}^{3} ) = \frac{1}{{\frac{{w_{{{\text{H}}_{2} {\text{O}}}} }}{{\rho_{{{\text{H}}_{2} {\text{O}}}} }} + w_{{{\text{Al}}_{2} ({\text{SO}}_{4} )_{3} }} \bar{\nu }_{{{\text{Al}}_{2} ({\text{SO}}_{4} )_{3} }} }} $$
where, w represents the mass fraction and \( \bar{\nu } \) is the partial molar volume (m3/kg). The density of water (\( \uprho_{{{\text{H}}_{2} {\text{O}}}} \)) is given by the following expression [27]:
$$ \rho_{{{\text{H}}_{2} {\text{O}}}} = \frac{{\begin{array}{*{20}c} {(( - 2.8054253 \times 10^{ - 10} T + 1.0556302 \times 10^{ - 7} )T - (4.6170461 \times 10^{ - 5} )T - (0.0079870401)T} \\ { + (16.945176)T + 999.83952)} \\ \end{array} }}{1 + 0.01687985T}. $$
The partial volume of aluminum sulfate in the draw solution is given as follows [27]:
$$ \bar{\nu }_{{{\text{Al}}_{2} ({\text{SO}}_{4} )_{3} }} = \frac{{w_{{{\text{Al}}_{2} ({\text{SO}}_{4} )_{3} }} - 713.10 - 25.569T}}{{(0.47444w_{{{\text{Al}}_{2} ({\text{SO}}_{4} )_{3} }} - 0.64624)e^{{0.000001(T + 4023.2)^{2} }} }}. $$

In Eqs. (12) and (13), T is the temperature in °C.

In this study, the Pitzer parameters (β ac (0) , β ac (1) , α, b and Cϕ) for aqueous aluminum sulfate solutions at 25 °C are presented in Table 2.
Table 2

Optimized Pitzer parameters for aqueous aluminum sulfate solutions at 25 °C [28]

Pitzer parameters


β ac (0)


β ac (1)








3.2 Equation of Mass Transfer

3.2.1 Water Flux

Water flux through the membrane is determined according to the changes in the feed solution volume:
$$ J_{w} = \Delta V/\left( {S \times \Delta t} \right) $$
where, Jw is the water flux (L/m2 h, LMH), ΔV is the change in volume of feed solution (L), Δt is the time interval (h) for the volume change of ΔV, and S is the active membrane surface area (m2).
Theoretically, as concentration polarization (CP) has no effects, the optimum water flux in FO obtained as follows [21]:
$$ J_{w} = A \times (\uppi_{D} - \uppi_{F} ) $$
where A is the membrane water permeability coefficient (m/s/Pa), πD is the osmotic pressure of the draw solution (Pa), πF is the osmotic pressure of the feed solution (Pa) obtained as:
$$ \pi_{D,F} = \phi \left( {\nu_{a} + \nu_{c} } \right) RTm = \left( {\nu_{a} + \nu_{c} } \right) RTm \left[ {1 + \left| {{\text{Z}}_{\text{a}}} {{\text{Z}}_{\text{c }} } \right|{\text{f}} + {\text{m}}\left( {\frac{{2\upnu_{\text{a}} \upnu_{\text{c}} }}{{\upnu_{\text{a}} + \upnu_{\text{c}} }}} \right)\upbeta_{\text{ac}} + {\text{m}}^{2} \left( {\frac{{2\left( {\upnu_{\text{a}} \upnu_{\text{c}} } \right)^{3/2} }}{{\upnu_{\text{a}} + \upnu_{\text{c}} }}} \right){\text{C}}_{\text{ac }} } \right] $$
where R is the universal gas constant, π is the osmotic pressure, T is the absolute temperature, νa and νc are the number of anion and cation in the electrolyte formula, Za and Zc are the charges of the anion and the cation, respectively, m is the overall molality of the electrolyte (mol/kg), Cac is third virial coefficient and βac defines the second virial coefficient elucidated in the Sect. 3.1.

3.2.2 Solute Flux

Pure deionized water served as a feed solution to determine the reverse draw solute flux. The input and output concentrations of the feed solution were measured. The reverse salt flux was calculated as follows [21]:
$$ J_{s} = ((C_{f} V_{f} - C_{0} V_{0} )/S\Delta t) $$
where, Js is the reverse salt flux (g/m2 h, gMH), C0 and Cf are the initial and final feed concentrations (g/L), respectively, V0 and Vf are the initial and final volumes of the feed solution (L), respectively, Δt is the time interval (h), and S is the active membrane surface area (m2).
According to Kohlrausch’s law, Cf and C0 were related to conductivity (σ) by the following equation:
$$ C_{f} = \upsigma_{f} / \uplambda $$
$$ C_{0} = \upsigma_{0} /\uplambda $$
where σf and σ0 are the initial and final conductivities of the feed solution (S/m), λ is the molar electrical conductivity of the feed solution (S m2/mol), C0 and Cf are the initial and final aluminum feed concentrations (mol/m3) which were calculated based on conductivity measurement with a portable conductivity meter.
From Eqs. (16) and (17), the expression for salt flux can be defined as follows:
$$ J_{s} = ((\upsigma_{f} /\uplambda )V_{f} ) - ((\upsigma_{0} /\uplambda )V_{0} )/S\Delta t. $$

4 Results and Discussion

4.1 Impact of FS Concentration

The performance of the implemented FO process using a CTA membrane in accordance with feed NaCl concentration is displayed in Fig. 3. Experimental and theoretical flux vs. feed solution concentration are plotted. The experimental permeate flux diminished slightly in the range of 2.63–1.31 L/m2 h by rising NaCl concentration from 5 to 45 g/L at constant DS concentration of 1 mol/L and a constant temperature of 298 K. The flux decrease can be assigned to the increase of the feed solution’s osmotic pressure, which reduces the pressure variance across the membrane and consequently the driving force (gradπ) of water flux [29].
Fig. 3

Water flux and water conductivity vs. concentration of FS

Yang et al. [30] studied the influence of FS concentration on the water flux and the removal of organic micro-pollutants. The water fluxes of phenol and aniline decreased from 17.9 to 15.2 L/m2 h and from 18.1 to 14.7 L/m2 h, respectively, as their concentrations increased from 500 to 2000 ppm. Moreover, Nguyen et al. [9] studied brackish water with 5, 10, and 20 g/L NaCl and obtained water fluxes of 7.68, 6.78 and 5.95 L/m2 h respectively in a study.

For all FO experiments, the permeate conductivity increased in parallel as a function of feed NaCl concentration. In fact, a feed solution concentration increase from 5 to 45 g/L can result in a slight increase of permeate conductivity from 120.1 to 160.5 μS/cm; which respects the drinking water regulations issued by the United States Environmental Protection Agency (USEPA). The National Primary Drinking Water Regulations (NPDWR) sets the conductivity limit value at 750 μS/cm [21]. So, the FO process guarantees a high quality recovered fresh water.

Figure 3 also shows the experimental and theoretical water flux. The theoretical water flux is almost greater than the experimental one in all the tests carried out. The results show that the theoretical water flux dropped from 4.59 to 3.88 L/m2 h, while the experimental flux recorded initial values of 2.62 L/m2 h decreasing to 1.31 L/m2 h.

This discrepancy between the model and the experimental data is attributed to the reverse solute flux interference with the osmotic pressure gradient. As a result, the driving force of the process decreased. Also, the high salts content may have been stuck to the membrane’s active layer and support layer surface. External concentration polarization (ECP) and internal concentration polarization (ICP) resulted in fouling [31]. Notably, spacers are allegedly connected to membrane sheets as they play the role of maintaining a distance between them. The spacer adjacent to one side of the membrane causes a decrease of the dimensionless driving force to an almost null value, irrespective of the role of feed or draw spacers. Hence, the spacer configuration in FO process requires additional optimization to control spacer attachment effect and optimize the performance of the process [32].

This difference between theoretical and experimental water flux was also reported by Qasim et al. [21], who found an experimental water flux 15 times lower than the theoretical one.

4.2 Effect of DS Concentration

Figure 4 shows the flux vs. conductivity for variable DS concentration from 0.7 to 1 mol/L at a constant FS concentration (20 g/L) and ambient temperature (T = 298 K). It shows a water increased flux at higher draw concentrations because the latter enhances the driving force, which results in an increase in water flux [33]. The asymptotic trends yield the highest permeate flux value (1.59 L/m2 h) for a 1 mol/L DS concentration.
Fig. 4

Water flux vs. concentration of DS

The influence of draw solution concentration was reported by Nguyen et al. [9], who showed that the water flux was raised significantly from 2.65 to 8.8 L/m2 h in parallel with EDTA-2Na concentration increase from 0.1 to 1 mol/L.

At different draw solution concentrations, the measured flux was varied between 1 and 1.59 L/m2 h; whereas the values of theoretical water flux fluctuated between a minimum of 2.35 and a maximum of 4.12 L/m2 h; which is around 2.5 times the experimental flux. This divergence is assigned to CP that considerably reduces the effective osmotic pressure gradient across the FO membrane [34].

In FO mode where the FS fronts to active membrane layer, this deviation essentially attributed to concentrative ECP and dilutive ICP [34]. The concentrative ECP stems from the accretion of feed solutes on the active layer of the membrane. However, dilutive ICP is attributed to the dilution of the draw solution with the membrane support layer and at greater draw solution concentration, it dominates progressively. This phenomenon decreases water flux because of increased osmotic pressure. The latter must be reversed via hydraulic pressure [35].

In addition, due to mass transfer mechanism restrictions, along with the membrane technology impact, an important reduction in permeate water flux may stem from the membrane fouling. Gao et al. [36] observed this phenomenon after a direct fouling via optical coherence tomography, which established the correlation between the preferential foulant deposition and the hydrodynamic shadow zones attributed to the presence of spacer.

As well, the geometry of the spacer, notably its thickness and opening ratio, affects the FO efficiency and potentially the fouling behavior. She et al. [37] asserted that thicker spacers reduce flux during biofouling in FO.

Reverse solute flux (Js) results from concentrated draw solution. This phenomenon occurs as the highly concentrated draw solution improves CP, and decreases the driving force and the permeate water flux [30]. Although several recent studies have elucidated the fouling of FO [38, 39], very few have attributed the effects of reverse solute diffusion to FO fouling [40]. She et al. [41] concluded that the occurrence of reverse draw solutes (RSD) was due to fouling. According to this work, the reverse diffusion of Mg2+ from MgCl2 draw solution into FS causes the growth of microalgae fouling on the FO membrane during the microalgae separation by FO membranes.

4.3 Effect of DS Temperature

Temperature difference was created between the DS and FS by increasing the temperature of the draw solution, while maintaining the temperature of FS constant (25 °C). Figure 5 plots the water flux (experimental and theoretical) versus the DS temperature. This figure shows the correlation between water flux and DS temperature at a constant FS temperature of 25 °C. In fact, as the DS temperature increased from 25 to 53 °C, the water flux rose from 1.59 to 2.50 L/m2 h. These results show the positive impact of temperature in enhancing the water flux and therefore the performance of FO. This is attributed to the Wilke–Chang equation where the diffusion coefficient is proportional to the absolute temperature divided by the viscosity of the solvent. At a higher temperature, the solution’s viscosity decreases and the diffusion coefficients increase; and consequently, the water flux increases [29].
Fig. 5

Water flux vs. DS temperature at FS temperature fixed at 25 °C

The water flux known to improve with temperature in FO [12, 29]. Xie et al. [42] investigated the effects of temperature on FO membrane flux with rainwater as feed solution and cooling water in the role of draw solution, the average membrane flux was 3.09 L/m2 h at a draw solution temperature of 50 °C and decreased considerably to about 0.28 L/m2 h at 3 °C.

The theoretical and experimental water flux values evolved similarly. In general, the experimental flux followed a similar trend as the theoretical one, the latter being greater in value. Indeed, as the concentration polarization effects were absent, the optimum water flux was obtained between 4.12 and 5.30 L/m2 h, when the experimental flux was between 1.59 and 2.5 L/m2 h.

The experimental water flux was observed to be about 2.5 times weaker than the theoretical one, irrespective of the concentration polarization effects. A cross-flow FO cell can enhance the flux when the feed and the draw solution flow are tangent to the membrane surface. By increasing the tangential velocity or the Re number of the DS flow, the impact of external concentration polarization falls and the mass transfer coefficient augments [43].

4.4 Impact of FS Temperature

FO experiments tests were done at different feed temperatures while keeping the draw temperature constant (25 °C). Figure 6 illustrates the impact of FS temperature on the water flux while maintaining the draw temperature constant (25 °C). The highest value of the flux was 2.18 L/m2 h at 53 °C and the smallest value was 1.59 L/m2 h at 25 °C.
Fig. 6

Water flux and water conductivity vs. FS temperature

Many studies attribute the rise of water flux to the raising of water viscosity, which increases the self-diffusivity of water, and consequently the mass transfer coefficient of FS [44].

Wang et al. [45] studied the impact of FS and DS temperature and transmembrane temperature difference on the rejection of 12 organic contaminants (TrOCs) by two FO membranes. They found an important impact of the feed temperature on the increase of the permeate flux. In fact, increasing the FS temperature from 20 to 40 °C engendered a water flux increase from 5.5 to 7.99 L/m2 h.

Figure 6 shows a difference between the model and the experimental flux at different feed temperatures. The increased flux curves for different feed temperatures exhibited a nearly identical trend. But the theoretical flux recorded higher values that increased from 3.02 to 4.39 L/m2 h, while the experimental flux increased from 1.59 to 2.19 L/m2 h.

The plot of the permeate conductivity according to feed temperature in Fig. 6 shows that the FS temperature increase from 25 to 53 °C can increase the permeate conductivity slightly from 152.2 to 182.2 μS/cm; which confirms the high quality of water produced by FO.

4.5 Effect of Flow Rate

The effect of feed and draw flow rates on the water flux was studied in the range of 5–35 L/h. When one of the two flow rates was varied, the second was fixed at 20 L/h. The chosen inlet feed and draw concentrations were, respectively, 35 g/L and 1 mol/L. The feed and draw temperatures were fixed at 25 °C. Figure 7 presents the variation of water flux versus the feed and draw flow rates. It proves that higher flow rates of both feed and draw solutions increase the water flux. In fact, the highest value of the water flux was 2 L/m2 h when the draw flow rate reached 35 L/h and the lowest value was 1.09 L/m2 h when feed flow rate was 5 L/h. Increasing the flow rates of feed and draw solutions involved the increment in terms of velocity. This is related to the rise of the mass transfer coefficient as a result of the rise of the Reynolds number which in turn causes a rise of the permeate flux [46]:
$$ Re = \left( {L \times \nu \times \rho } \right)/\mu $$
where L is the length of FO test cell (m), ν is the velocity (m/s), ρ is the density of the solution (kg/m3), and μ is the dynamic viscosity (Pa s).
Fig. 7

Water flux vs. draw flow rate and feed flow rate

The influence of feed and draw flow rates was reported by Touati et al. [47], who found that the water flux evolved from 4.5 10−6 to 6.3 10−6 m/s with the rise of the EDTA-2Na concentration from 0.00535 to 0.0321 m/s at draw and feed solution concentrations of 1.026 mol/L and 8.55 mmol/L, respectively.

4.6 FO Performance Evaluation: Water and Reverse Solute Fluxes

The FO experiment results were in favor of using aluminum sulfate as a draw solution for desalination. Interestingly, 1 mol/L aluminum sulfate draw solution at 53 °C draw temperature and 35 L/h draw flow rate while keeping FS temperature at 25 °C and feed flow rate 20 L/h should be used for further desalination experiments.

Water flux and the reverse salt flux (Js) were quantified with aluminum sulfate as a draw solution. The reverse aluminum sulfate flux was found using Eq. (2) and quantified at 1.46 g/m2 h. The ratio of reverse salt flux to water flux (Js/Jw) was 0.47 g/L. Thus, we can assert that for 1 L of water permeated in the FO process, 0.47 g of aluminum sulfate draw solute diffuses across the membrane to the feed solution. The draw solute loss during the FO process of desalination is identified via the ratio Js/Jw. This ratio is also useful for the choice of appropriate FO membrane and the draw solute, notably when high product water purity is sought [19].

4.7 Product Water Recovery and Regeneration of Draw Solution

The precipitation reaction served to recover pure water from the obtained diluted draw solution. A stoichiometric amount of aqueous calcium hydroxide (Ca(OH)2) solution was supplemented to a sample of the draw solution in accordance with the concentration of aluminum sulfate in the diluted draw solution. The following chemical reaction was obtained:
$$ {\text{Al}}_{2} ({\text{SO}}_{4} )_{3} ( {\text{aq) }} + 3{\text{Ca(OH)}}_{2} ({\text{aq}}) \to 2{\text{Al(OH)}}_{3} ( {\text{s) }} + 3{\text{CaSO}}_{4} ({\text{s}}). $$

The chemical reaction resulted in two insoluble products, aluminum hydroxide (Al(OH)3) and calcium sulfate (CaSO4); which proves the removal of soluble aluminum sulfate draw solute from the draw solution and its conversion into insoluble chemicals that were settled by gravity afterward. After settlement, we filtered the top layer of water by means of a standard laboratory filter paper. Then, the filtrate was recovered as pure water product.

The described FO process for desalination favors the regeneration of the aluminum sulfate draw solution. The reaction of the obtained precipitates, including aluminum hydroxide and calcium sulfate, with sulfuric acid can regenerate the aluminum sulfate draw solution as presented by chemical reaction:
$$ 2{\text{Fe(OH)}}_{3} ( {\text{s)}} + 3{\text{H}}_{2} {\text{SO}}_{4} ({\text{aq}}.) \to {\text{Fe}}_{2} ({\text{SO}}_{4} )_{3} ( {\text{aq}}. )+ 6{\text{H}}_{2} {\text{O(lq}} . ). $$

However, as water is produced through this reaction, the regenerated draw solution is dilute in comparison with the fresh draw solution. This consequently entails the supplementation of make-up aluminum sulfate to the regenerated draw solution to restore the initial draw solution concentration.

After the regeneration of the draw solution, calcium sulfate is obtained as a secondary product of the process. Calcium sulfate is a useful industrial chemical, most commonly as an ingredient in chemical yeasts, a firming agent for fruit. It is also an ingredient of building materials such as cement, plaster or tiles and present in cosmetics, toothpastes and other pastes. Also, calcium sulfate is commonly used as filler for the paper and in pharmaceutical products.

5 Conclusion

We evaluated the effects of temperature, concentration and flow rate on FO performance with aluminum sulfate as a novel draw solution and sodium chloride as feed solution. The obtained results indicated that water flux increases with DS concentration at low FS concentration in the FO mode. We proved experimentally that the impact of the DS temperature is more significant than the FS temperature on water flux. The tests performed under FO application standard conditions indicated that increased temperature and concentration of DS are the most suitable conditions to improve the FO performance. In addition, aluminum sulfate tested as DS showed that it is highly worthwhile in FO desalination. The described DS is highly water-soluble and it can be found in natural sources. Also, its anhydrous form is a rare mineral. FO performance evaluation indicated that aluminum sulfate solution is capable to extract water from saline feed solution and to produce good quality water. The diluted DS can be simply separated from high-purity produced water and regenerated using simple chemical reactions with no need for electrical energy. Although the experimental results demonstrated the efficiency of aluminum sulfate as DS for water desalination, the FO process was severely affected by concentration polarization as the experimental flux was 2.5 times lower than the theoretical one. The mathematical models that predict and fit the experimental data considering the effects of reverse solute flux, polarization concentration (ICP and ECP) and membrane fouling are required.


  1. 1.
    El-Mokhtar I, Boubakri A, Bouguecha SAT, Hafiane A (2019) Modeling and experimental study of air gap membrane distillation unit: application for seawater desalination. Desalin Water Treat 154:72–81CrossRefGoogle Scholar
  2. 2.
    Karagiannis IC, Soldatos PG (2008) Water desalination cost literature: review and assessment. Desalination 223:448–456CrossRefGoogle Scholar
  3. 3.
    Wensong C, Jiahao M, Xing D, Zhien Z, Wenxiang Z (2019) Biomimetic dynamic membrane for aquatic dye removal. Water Res 15:243–251Google Scholar
  4. 4.
    Jiahao M, Qi Y, Na Z, Wenxiang Z, Yi Z, Zhien Z (2018) A review on agro-industrial waste (AIW) derived adsorbents for water and wastewater treatment. J Environ Manag 227:395–405CrossRefGoogle Scholar
  5. 5.
    Valladares LR, Li Z, Yangali-Quintanilla V, Li Q, Amy G (2013) Cleaning protocol for a FO membrane fouled in wastewater reuse. Desalin Water Treat 51:4821–4824CrossRefGoogle Scholar
  6. 6.
    Wang Y, Wicaksana F, Tang CY, Fane AG (2010) Direct microscopic observation of forward osmosis membrane fouling. Environ Sci Technol 44:7102–7109CrossRefGoogle Scholar
  7. 7.
    Cartinella JL, Cath TY, Flynn MT, Miller GC, Hunter KW, Childress AE (2006) Removal of natural steroid hormones from wastewater using membrane contactor processes. Environ Sci Technol 40:7381–7386CrossRefGoogle Scholar
  8. 8.
    Zhao S, Zou L, Mulcahy D (2012) Brackish water desalination by a hybrid forward osmosis–nanofiltration system using divalent draw solute. Desalination 284:175–181CrossRefGoogle Scholar
  9. 9.
    Nguyen HT, Nguyen NC, Chen SS, Ngo HH, Guo W, Li CW (2015) A new class of draw solutions for minimizing reverse salt flux to improve forward osmosis desalination. Sci Total Environ 538:129–136CrossRefGoogle Scholar
  10. 10.
    Chekli L, Phuntsho S, Shon HK, Vigneswaran S, Kandasamy J, Chanan A (2012) A review of draw solutes in forward osmosis process and their use in modern applications. Desalin Water Treat 43:167–184CrossRefGoogle Scholar
  11. 11.
    Ge Q, Ling M, Chung TS (2013) Draw solutions for forward osmosis processes: developments, challenges, and prospects for the future. J Membr Sci 442:225–237CrossRefGoogle Scholar
  12. 12.
    Cath TY, Childress AE, Elimelech M (2006) Forward osmosis: principles, applications, and recent developments. J Membr Sci 281:70–87CrossRefGoogle Scholar
  13. 13.
    Zhao S, Zou L, Tang CY, Mulcahy D (2012) Recent developments in forward osmosis: opportunities and challenges. J Membr Sci 396:1–21CrossRefGoogle Scholar
  14. 14.
    Batchelder GW (1965) Process for the demineralization of water. US Patent 3:171–799Google Scholar
  15. 15.
    McCutcheon JR, Elimelech M (2006) Influence of concentrative and dilutive internal concentration polarization on flux behavior in forward osmosis. J Membr Sci 284:237–247CrossRefGoogle Scholar
  16. 16.
    Adham S, Oppenheimer J, Liu L, Kumar M (2007) Dewatering reverse osmosis concentrate from water reuse using forward osmosis. Water Reuse Found Res Rep 5:1–52Google Scholar
  17. 17.
    Ge Q, Su J, Chung TS (2011) Amy G Hydrophilic superparamagnetic nanoparticles: synthesis, characterization, and performance in forward osmosis processes. Ind Eng Chem Res 50:382–388CrossRefGoogle Scholar
  18. 18.
    Ling MM, Chung TS (2011) Desalination process using super hydrophilic nanoparticles via forward osmosis integrated with ultrafiltration regeneration. Desalination 278:194–202CrossRefGoogle Scholar
  19. 19.
    Ge Q, Su J, Amy GL, Chung TS (2012) Exploration of polyelectrolytes as draw solutes in forward osmosis processes. Water Res 46:1318–1326CrossRefGoogle Scholar
  20. 20.
    Stone ML, Rae C, Stewart FF, Wilson AD (2013) Switchable polarity solvents as draw solutes for forward osmosis. Desalination 312:124–129CrossRefGoogle Scholar
  21. 21.
    Qasim M, Mohammed F, Aidan A, Darwish NA (2017) Forward osmosis desalination using ferric sulfate draw solute. Desalination 423:12–20CrossRefGoogle Scholar
  22. 22.
    Bouchrit R, Boubakri A, Mosbahi T, Hafiane A, Bouguecha SAT (2017) Membrane crystallization for mineral recovery from saline solution: study case Na2SO4 crystals. Desalination 412:1–12CrossRefGoogle Scholar
  23. 23.
    Ben Maouia D, Boubakri A, Bouchrit R, Hafiane A, Bouguecha SAT (2019) Chromium removal by forward osmosis: a flux modeling and experimental validation. Desalin Water Treat 154:30–38CrossRefGoogle Scholar
  24. 24.
    Smith JM, Van Ness HC, Abbott MM (2005) Introduction to chemical engineering thermodynamics. J Chem Educ 584:10–27Google Scholar
  25. 25.
    Pitzer KS, Mayorga G (1973) Thermodynamics of electrolytes. II. Activity and osmotic coefficients for strong electrolytes with one or both ions univalent. J Phys Chem 77:2300–2308CrossRefGoogle Scholar
  26. 26.
    Hu ZA, Wu HY, Gao JZ (1999) Calculation of osmotic pressure difference across membranes in hyperfiltration. Desalination 121:131–137CrossRefGoogle Scholar
  27. 27.
    Laliberte M, Cooper WE (2004) Model for calculating the density of aqueous electrolyte solutions. J Chem Eng Data 49:1141–1151CrossRefGoogle Scholar
  28. 28.
    Christov C (2001) Thermodynamic study of the K–Mg–AI–CI–SO4–H2O system at the temperature 298.15 K. Calphad 25:445–454CrossRefGoogle Scholar
  29. 29.
    Phuntsho S, Shon HK, Vigneswaran S, Kandasamy J, Hong SK, Lee SY (2012) Influence of temperature and temperature difference in the performance of forward osmosis desalination process. J Membr Sci 416:734–744CrossRefGoogle Scholar
  30. 30.
    Yang Q, Wang KY, Chung TS (2009) Dual-layer hollow fibers with enhanced flux as novel forward osmosis membranes for water production. Environ Sci Technol 43:2800–2805CrossRefGoogle Scholar
  31. 31.
    Rannany M, Sousa S, Lora-Garcia J, López-Pérez MF (2018) Experimental study and modeling of forward osmosis process for activated sludge concentration by using residual brine from a stuffed olive factory as draw solution. J Water Process Eng 21:143–153CrossRefGoogle Scholar
  32. 32.
    Park M, Kim JH (2013) Numerical analysis of spacer impacts on forward osmosis membrane process using concentration polarization index. J Membr Sci 427:10–20CrossRefGoogle Scholar
  33. 33.
    Wang KY, Teoh MM, Nugroho A, Chung TS (2011) Integrated forward osmosis–membrane distillation (FO–MD) hybrid system for the concentration of protein solutions. Chem Eng Sci 66:2421–2430CrossRefGoogle Scholar
  34. 34.
    McCutcheon JR, McGinnis RL, Elimelech M (2005) A novel ammonia–carbon dioxide forward (direct) osmosis desalination process. Desalination 174:1–11CrossRefGoogle Scholar
  35. 35.
    Elimelech M, Bhattacharjee S (1998) A novel approach for modeling concentration polarization in crossflow membrane filtration based on the equivalence of osmotic pressure model and filtration theory. J Membr Sci 145:223–241CrossRefGoogle Scholar
  36. 36.
    Gao Y, Haavisto S, Li W, Tang CY, Salmela J, Fane AG (2014) Novel approach to characterizing the growth of a fouling layer during membrane filtration via optical coherence tomography. Environ Sci Technol 48:14273–14281CrossRefGoogle Scholar
  37. 37.
    She Q, Hou D, Liu J, Tan KH, Tang CY (2013) Effect of feed spacer induced membrane deformation on the performance of pressure retarded osmosis (PRO): implications for PRO process operation. J Membr Sci 445:170–182CrossRefGoogle Scholar
  38. 38.
    Mi B, Elimelech M (2010) Gypsum scaling and cleaning in forward osmosis: measurements and mechanisms. Environ Sci Technol 44:2022–2028CrossRefGoogle Scholar
  39. 39.
    Zou S, Gu Y, Xiao D, Tang CY (2011) The role of physical and chemical parameters on forward osmosis membrane fouling during algae separation. J Membr Sci 366:356–362CrossRefGoogle Scholar
  40. 40.
    Mi B, Elimelech M (2010) Organic fouling of forward osmosis membranes: fouling reversibility and cleaning without chemical reagents. J Membr Sci 348:337–345CrossRefGoogle Scholar
  41. 41.
    She Q, Wang R, Fane AG, Tang CY (2015) Membrane fouling in osmotically driven membrane processes: a review. J Membr Sci 30270:2376–7388Google Scholar
  42. 42.
    Xie M, Price WE, Nghiem LD, Elimelech M (2013) Effects of feed and draw solution temperature and transmembrane temperature difference on the rejection of trace organic contaminants by forward osmosis. J Membr Sci 438:57–64CrossRefGoogle Scholar
  43. 43.
    Alnaizy R, Aidan A, Qasim M (2013) Copper sulfate as draw solute in forward osmosis desalination. J Environ Chem Eng 1:424–430CrossRefGoogle Scholar
  44. 44.
    Bhattacharya M, Dutta SK, Sikder J, Mandal MK (2014) Computational and experimental study of chromium (VI) removal in direct contact membrane distillation. J Membr Sci 450:447–456CrossRefGoogle Scholar
  45. 45.
    Wang W, Zhang Y, Esparra-Alvarado M, Wang X, Yang H, Xie Y (2014) Effects of pH and temperature on forward osmosis membrane flux using rainwater as the makeup for cooling water dilution. Desalination 351:70–76CrossRefGoogle Scholar
  46. 46.
    Bahar R (2010) Conversion of saline water to fresh water using air gap membrane distillation AGMD (thèse)Google Scholar
  47. 47.
    Touati K, Tadeo F (2016) Study of the reverse salt diffusion in pressure retarded osmosis: influence on concentration polarization and effect of the operating conditions. Desalination 389:171–186CrossRefGoogle Scholar

Copyright information

© The Tunisian Chemical Society and Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.Laboratory of Water, Membranes and Environmental BiotechnologiesCenter for Water Researches and TechnologiesSolimanTunisia
  2. 2.Department of Mechanical Engineering, Faculty of EngineeringKing Abdul-Aziz UniversityJeddahSaudi Arabia

Personalised recommendations