2.4 Fundamentals in Reverse Osmosis

2.4 Fundamentals in Reverse Osmosis Francesca Macedonio, Institute on Membrane Technology of the National Research Council of Italy, ITM-CNR, Rende, Italy and University of Calabria, Rende, Italy Enrico Drioli, Institute on Membrane Technology of the National Research Council of Italy, ITM-CNR, Rende, Italy; University of Calabria, Rende, Italy; Hanyang University, Seoul, South Korea; and King Abdulaziz University, Jeddah, Saudi Arabia r 2017 Elsevier B.V. All rights reserved.

2.4.1 2.4.2 2.4.2.1 2.4.2.1.1 2.4.2.1.2 2.4.2.2 2.4.2.2.1 2.4.2.2.2 2.4.2.2.3 2.4.2.3 2.4.2.3.1 2.4.2.3.2 2.4.3 2.4.4 2.4.4.1 2.4.4.2 2.4.4.2.1 2.4.4.2.2 2.4.4.2.3 2.4.4.3 2.4.5 2.4.6 2.4.7 2.4.7.1 2.4.7.2 2.4.7.3 2.4.7.4 2.4.7.5 2.4.8 References Further Reading

2.4.1

Introduction Models for Description of Solvent and Solute Fluxes Phenomenological Transport Models Irreversible thermodynamics—Phenomenological transport model Irreversible thermodynamics—Kedem-Spiegler model Nonporous Transport Models Solution-diffusion model Extended solution-diffusion model Solution-diffusion-imperfection model Porous Transport Models Friction model Finely-porous model Membrane Charge Limiting Factors: Concentration Polarization, Fouling, Scaling, Biofouling, and Membrane Deterioration Concentration Polarization Membrane Fouling/Scaling/Biofouling Chemical foulants Physical foulants or particulate matter Biological foulants Membrane Deterioration Membrane Materials for RO Membrane Modules for RO New Materials for RO Membranes Ceramic/Inorganic Membranes Mixed Matrix Membranes Carbon Nanotubes Graphene and Graphene Oxide Biomimetic RO Membranes Conclusions

79 81 82 82 82 83 83 83 84 84 84 85 86 86 86 87 87 88 89 89 89 90 90 91 91 91 92 92 92 93 94

Introduction

Reverse osmosis (RO) is a membrane process used for a wide range of applications, most of which are in the purification of water to produce potable water, mainly desalination of sea (TDSE35,000 ppm) and brackish water (TDS in the range of 1000–5000 ppm). Another important application is in the production of ultra-pure water for the semiconductor industry. The principle of the process is as follows: forcing a solvent through the molecular structure of a membrane, while trapping impurities and salts. In nature, when a semipermeable membrane (i.e., a membrane permeable for the solvent but impermeable for the solute) separates two compartments at different concentration, the water tends to flow from the lower to the higher concentrated compartment according to the natural osmosis phenomenon. Thus, the concentrated solution will be diluted until when the equilibrium between the compartments is reached and the trans-membrane flux becomes zero. Reverse osmosis is when water flows through the membrane from the concentrated to the diluted solution. To obtain this an external pressure higher than the osmotic pressure difference has to be applied to the concentrated solution (Fig. 1). Therefore, when a semipermeable membrane separates two solutions (the first one indicated by (1) whereas the second one by (2)), three different situations can be distinguished depending on the concentrations and hydrostatic pressures in the two phases: (a) solution (1) and solution (2) have the same hydrostatic pressure but the solute concentration in solution (1) is higher than the one in solution (2). This situation is referred to as osmosis because there will be a flow of solvent from the more diluted solution (2) into the more concentrated solution (1) due to the higher osmotic pressure of solution (1).

Comprehensive Membrane Science and Engineering II, Volume 2

doi:10.1016/B978-0-12-409547-2.12264-4

79

80

Fundamentals in Reverse Osmosis

Fig. 1 Reverse osmosis phenomenon.

Fig. 2 Solvent flux between two solutions of different concentrations through a strictly semipermeable membrane as function of the hydrostatic pressure applied to the more concentrated solution.

(b) The two solutions have different hydrostatic pressures, but the difference in hydrostatic pressure is equal to the difference in the osmotic pressures between the two solutions acting in opposite direction. This situation is referred to as osmotic equilibrium and there will be no flow of solvent through the membrane although the concentrations in the two solutions are different. (c) The two solutions have different hydrostatic pressures, but the difference in hydrostatic pressure across the membrane is larger than that in the osmotic pressure and is acting in opposite direction. Thus, solvent will flow from the solution (1) with the higher solute concentration into the solution (2) with the lower solute concentration. This phenomenon is referred to as reverse osmosis. Fig. 2 illustrates the flux of solvent through a strictly semipermeable membrane separating two solutions of different concentrations as a function of the hydrostatic pressure applied to the more concentrated solution. In order to allow the solvent (i.e., water) to pass through the membrane, the applied pressure DP (between the concentrated side and the dilute side) must be higher than the osmotic pressure Dp. As it can be seen from Fig. 2, the solvent flows from the dilute solution to the concentrated solution if the applied pressure is smaller than the osmotic pressure. When the applied pressure is higher than the osmotic pressure, the solvent flows from the concentrated solution to the dilute solution.Thermodynamically, the osmotic pressure p is defined as1: p¼ 

RT lnðxw Þ Vb

where Vb is the molar volume of water, xw the mole fraction of water, and R the ideal gas constant. In dilute solutions, the osmotic pressure can be estimated using Van t’Hoff's law, which is of the same form as the ideal gas law: p¼ 

ns RT V

or p ¼ CRT

Fundamentals in Reverse Osmosis

81

where ns is the total amount of moles of solutes in solution, C the total concentration of solutes [moles/L], and V the volume of solvent. Taking into account nonideality and dissociation of the ions in solution, Van t’Hoff's law can be rewritten as p ¼ ifCRT with i representing the dissociation parameter, which is equal to the number of ions and molecules per mole of solute produced by dissolution of the solute and f representing a correction factor that takes into account nonidealities.

2.4.2

Models for Description of Solvent and Solute Fluxes

Reverse osmosis membranes have in general an asymmetric or a thin-film composite structure where a porous and thin top layer acts as selective layer and determines the resistance to transport. Macroscopically these membranes are homogeneous. However, at the microscopic level, they are systems with two phases in which the transport of water and solutes takes place. As reported by Jonsson and Macedonio,2 several models on RO transport mechanisms have been developed to describe solute and solvent fluxes through RO membranes. The general purpose of a membrane mass transfer model is to relate the fluxes to the operating conditions. The power of a transfer model is its ability to predict the performance of the membrane over a wide range of operating conditions. To reach the aim the model has to be integrated with some transport coefficients often determined based on some experimental results. When theories are proposed to describe membrane transport, either the membrane can be treated as a “black box” in which a purely thermodynamic description is used, or a physical model of the membrane can be introduced. The general description obtained in the first case gives no information on flow- and separation-mechanisms. On the other hand, the correctness of data on flow and separation mechanisms obtained in the second case depends on the chosen model. The transport models can be divided into three categories: 1. phenomenological transport models which are independent of the mechanism of transport and are based on the theory of irreversible thermodynamics (irreversible thermodynamics-phenomenological transport and irreversible thermodynamicsKedem-Spiegler models), 2. nonporous transport models, in which the membrane is supposed to be nonporous or homogeneous (solution-diffusion, extended solution-diffusion, and solution-diffusion-imperfection models), 3. porous transport models, in which the membrane is supposed to be porous (preferential sorption-capillary flow, KimuraSourirajan analysis, finely porous and surface force-pore flow, and friction models). Most models for RO membranes assume diffusion or pore flow through the membrane while charged membrane theories include electrostatic effects. For example, Donnan exclusion models can be used to determine solute fluxes in the often negatively charged nanofiltration membranes. Fig. 3 provides a schematic presentation of a thin-film composite membrane structure with (i) the highly selective skin layer that acts as a barrier, (ii) the intermediate porous layer where the selectivity decreases to zero, and (iii) the nonselective porous sublayer. The porous sublayer influences the total hydraulic permeability Lp3: 1 1 1 1 ¼
 þ
 þ
 Lp Lp sl Lp il Lp pl

ð1Þ

but it has almost no influence on the solute rejection properties of the membranes. Therefore, most transport and rejection models of RO membranes have been derived for single-layer membranes focusing almost only on the surface thin layer. Transport models can help in the identification of the most important membrane structural parameters and in showing how membrane performance can be improved by varying some specific parameters. One of the main membrane intrinsic parameters is

Fig. 3 Schematic presentation of thin-film composite membrane structure. From Jonsson, G.; Macedonio, F. Fundamentals in reverse osmosis, chapter 2.01. In E. Drioli and L. Giorno (Eds.). Comprehensive membrane science and engineering, Vol. 2. Elsevier B.V. 2010.

82

Fundamentals in Reverse Osmosis

the reflection coefficient s, introduced by Staverman4 and defined as:   lpp DP ¼ s lp Dp Jv ¼ 0

ð2Þ

s acts to describe the effect of the pressure driving force on the flux of solute and it represents the relative permeability of the membrane to the solute: – s¼1 for a high-separation membrane – s¼ 0 for a low-separation membrane in which the solute is significantly carried through the membrane by solvent flux. In RO the intrinsic retention Rmax is related to s and normally srRmax (as reported in Ref.5). Push6 derived the following relationship between Rmax and s: Rmax ¼ 1  ð1  sÞ 

csmax c0s

where csmax is the mean salt concentration at infinite Jv.

2.4.2.1 Phenomenological Transport Models 2.4.2.1.1

Irreversible thermodynamics—Phenomenological transport model

The membrane is treated as a “black box” when nothing on the transport mechanism and membrane structure is known. In this case, the thermodynamics of irreversible processes (IT) can be applied to membrane systems. According to IT theory, the flow of each component in a solution is related to the flows of other components. Then, different relationships between the flux through the membrane and the forces acting on the system can be formulated. Onsager7 suggests that fluxes Ji are related to the forces Fj through the phenomenological coefficient Lij: X Ji ¼ Lii Fi þ Lij Fj f or i ¼ 1; …; n ð3Þ ia j

For systems close to equilibrium, the cross-coefficients are equal: Lij ¼ Lji

f or ia j

ð4Þ

8

Kedem and Katchalsky used the linear phenomenological relationships (3) and (4) to derive the phenomenological transport Eqs. (5) and (6): Jv ¼ lp ðDP  sDpÞ

ð5Þ

Js ¼ oDp þ ð1  sÞJv ðcs Þln

ð6Þ

where parameters lp, o, and s are simple functions of the original phenomenological coefficient Lij. Usually RO systems are far from equilibrium; therefore, Eq. (4) may not be correct. Moreover, phenomenological transport Eqs. (5) and (6) have been rarely applied for describing RO membrane transport both because the often large concentration difference across the membranes invalidates the linear laws and because this analysis doesn’t give much information regarding the transport mechanism.

2.4.2.1.2

Irreversible thermodynamics—Kedem-Spiegler model

Spiegler and Kedem9 bypassed the problem of “linearity” by rewriting the original IT equations for solvent and solute flux in differential form (Eqs. (7) and (8)):   dP dp s Jv ¼ pv ð7Þ dx dx dcS þ ð1  sÞcS Jv ð8Þ dx where pv is the water permeability, x the coordinate direction perpendicular to the membrane, and ps the solute permeability. Integrating Eqs. (7) and (8) over the thickness of the membrane by assuming pv, ps, and s constant, the following equations for the solvent flux Jv and the retention R are achieved: Js ¼ pS

Jv ¼

pv ðDP  sDpÞ Dx

ð9Þ

sf1  exp½Jv ð1  sÞDx=ps g ð10Þ 1  sexp½Jv ð1  sÞDx=ps  where Dx is the membrane thickness. Eq. (10) can be rearranged as follows:   0 cs 1 s Dx  exp J ¼ ð 1  s Þ ð11Þ v c00s 1s 1s ps However, similar to phenomenological transport equations, Spiegler and Kedem relationships also do not give information on the membrane transport mechanism. R¼

Fundamentals in Reverse Osmosis

83

2.4.2.2 Nonporous Transport Models 2.4.2.2.1

Solution-diffusion model

The solution-diffusion model assumes that (i) membrane surface layer is homogenous and nonporous and (ii) both solute and solvent dissolve in the surface layer and then diffuse across it independently. Water and solute fluxes are proportional to their chemical potential gradient. The latter is expressed as the pressure and concentration difference across the membrane for the solvent, whereas it is assumed to be equal to the solute concentration difference across the membrane for the solute: Jv ¼ AðDP  DpÞ

ð12Þ

Dv cv Vv ℜTDx  00 Js ¼ B c‴s  cs

ð13Þ

A¼

ð14Þ

Ds k ð15Þ Dx 00 ‴ where A is the hydraulic permeability constant lp, B is the salt permeability constant, and cs and cs are, respectively, the salt concentrations on the feed and permeate sides of the membrane. Dv and D s are the diffusivities of the solvent and the solute in the membrane, respectively; cv is the concentration of water in the membrane; Vv is the partial molar volume of water; ℜ is the universal gas constant; T is the temperature; and k is the partition or distribution coefficient of solute defined as follows: B¼

K¼

kg solute=m3 membrane kg solute=m3 solution

ð16Þ

k measures the solute affinity to (k41) or repulsion from (ko1) the membrane material. Following Eqs. (12)–(15), differences in solubilities and diffusivities of the solute and solvent in the membrane phase are important in this model since these differences strongly influence the fluxes through the membrane. Moreover, these equations prove that the solute flux through the membrane is independent of water flux. 00 Because the concentration of salt in the permeate solution cs is usually much smaller than c‴s , Eq. (14) can be simplified as follows: Js ¼ B c‴s

ð17Þ

Eqs. (12) and (17) show that the water flux is proportional to the applied pressure, whereas the solute flux is independent of pressure. This means that the membrane selectivity increases with increasing pressure. The membrane selectivity can be measured as solute rejection R given by  00  c R ¼ 1  ‴s  100% ð18Þ cs 00

By combining Eqs. (12)–(18) with the relationship (19) between cs , Jv, and Js, the membrane rejection can be expressed as follows: 00

cs ¼  R¼ 1

Js r Jv v

 rv  B  100% AðDP  DpÞ

ð19Þ ð20Þ

where rv is the density of water. The main advantage of the solution-diffusion model is its simplicity. One of its restrictions is that it foresees rejection equal to 1 at infinite flux (DP-1), a limit not reachable for many solutes. Therefore, this model is appropriate for solvent-solute-membrane systems where the separation is close to 1. Moreover, it can be noted that Eq. (5) is reduced to the solution-diffusion model when s¼ 1.

2.4.2.2.2

Extended solution-diffusion model

In the solution diffusion model the effect of pressure on solute transport is neglected.10,11 In order to include the pressure term, the salt chemical potential gradient has to be written as:  ‴ c Dms ¼ ℜT ln s00 þ Vs DP ð21Þ cs where Dms is the solute chemical potential difference across the membrane and Vs is the
solute volume. For  partialmolar c‴ s DP when ln c00s ≫8:0  106 DP. sodium–chloride–water separation, Burghoff et al.10 suggest to ignore the pressure term VℜT s Including pressure, particularly for organic-water systems, the solute flux is given by Ds k  ‴ 00 Js ¼ cs  cs þ lsp DP ð22Þ Dx where lsp is the pressure-induced transport parameter. Eq. (22) has been proved to be accurate for different organic solutes with cellulose acetate membranes.10

84

Fundamentals in Reverse Osmosis

2.4.2.2.3

Solution-diffusion-imperfection model

The solution-diffusion model is one of the most referred membrane models. It presupposes that membrane surface is homogenous/nonporous and has the limitation that the intrinsic value of retention is always unity. The solution-diffusion-imperfection model (SDIM) developed by Sherwood et al.12 considers that small imperfections exist on the membrane surface due to the membrane-making process, and solvent and solute can flow through them without any change in concentration. Therefore, SDIM includes pore flow as well as diffusion of solute and solvent through the membrane and it can be considered a compromise between solution-diffusion and porous models. Moreover, Jonsson and Boesen13 proved that SDIM can be used to determine a parameter identified with the reflection coefficient. According to the model, water and solute fluxes can be written as:   k3 K3 DP ¼ ðk1 þ k3 Þ DP  Dp ð23Þ Jv ¼ k1 ðDP  DpÞ þ |fflffl{zfflffl} |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl} k1 þ k3 diffusion

pore flow contribution to water flux 0

Js ¼ k2 Dp þ

K3 DPcs |fflfflffl{zfflfflffl}

ð24Þ

pore flow of solute through the membrane

where K3 DP is the term proportional to the pressure driving force; k1 and k2 are the transport parameters for diffusive water and solute flux, respectively; and k3 is the transport parameter for the pore flow. Eqs. (23) and (24) can be rearranged to give the reduction factor5: 0 0 ðDP  DpÞ þ kk31 DP cs cs Jv ¼ 00 ¼ k3 k2 Dp cs Js k 0 þ k DP 1

cs

ð25Þ

1

and comparing Eq. (23) with Eq. (5), s can be obtained s¼

1 1 þ kk31

ð26Þ

where the ratio kk31 is a measure of the relative contribution of pore flow compared to diffusive flow. This model has been successfully applied for the performance description of several solutes and membranes,13 particularly it is proper for those membranes exhibiting lower separation than that calculated from solubility and diffusivity measurements.

2.4.2.3 Porous Transport Models Among the transport models in which the membrane is supposed to be porous, friction and finely porous models are described in this section.

2.4.2.3.1

Friction model

Friction model considers that the transport through porous membrane occurs both by viscous and diffusion flow. Therefore, the pore sizes are considered so small that the solutes cannot pass freely through the pores but friction between solute-pore wall, solvent-pore wall, and solvent-solute occurs. The frictional force F is linearly proportional to the velocity difference through a proportionality factor X called “friction coefficient” indicating the interaction between solute and pore wall: F23 ¼  X23 ðu2  u3 Þ ¼  X23  u2

ð27Þ

F13 ¼  X13 ðu1  u3 Þ ¼  X13  u1

ð28Þ

F21 ¼  X21 ðu2  u1 Þ

ð29Þ

F12 ¼  X12 ðu1  u2 Þ ð30Þ Eqs. (27)–(30) are derived considering the membrane as reference (u3 ¼ 0). Considering the fact that the frictional force per mole of solute F23 is given by F23 ¼  X23  u2 ¼  X23

J2p c2p

ð31Þ

Eq. (27) can be written as J2p ð32Þ c2p Jonsson and Boesen13 have presented a detailed description of this model and have shown that, as F21 is the effective force for diffusion of solute in the center of mass system, the solute flux per unit pore area J2p is given by: F23 ¼  X23

J2p ¼

1 c2p ðF21 Þ þ c2p  u X21

ð33Þ

Fundamentals in Reverse Osmosis

85

A balance of applied and frictional forces is equal to: F2 ¼  ðF21 þ F23 Þ

ð34Þ

Neglecting the pressure term and in the case of dilute solution behavior, F2 is equal to F2 ¼ 

RT dc2p c2p dx

ð35Þ

Defining b as the term that relates the frictional coefficient X23 (between solute and membrane) and X21 (between solute and water) b¼

X21 þ X23 X21

ð36Þ

and inserting in Eq. (34) the Eqs. (29), (32), (35), and (36), J2p will be equal to: J2p ¼ 

RT dc2p c2p  u þ b X21  b dx

ð37Þ

The coefficient of distribution K of the solute between bulk solution and pore fluid is given by K ¼ c2p =c2

ð38Þ

with Jv ¼ e  u, Ji ¼ J2  e, and ξ ¼ t  x, using the product condition 00

c2 ¼

J2p u

ð39Þ

and integrating Eq. (37) with the boundary conditions: 0

c2p ¼ Kc2

x¼0 : x¼tl : the following equation for the ratio

0 c2 00 c2

00

c2p ¼ Kc2

is obtained: 
 0 X21 1 þ Kb exp ue tl 1 c2
tle XRT 00 ¼ c2 exp ue e RT21

ð40Þ

In this derivation K, b, and X21 are assumed to be independent of the solute concentration.

2.4.2.3.2

Finely-porous model

Finely-porous model was developed by Merten14 using a balance of applied and frictional forces proposed by Spiegler.15 It is a combination between viscous flow and frictional model presented in detail by Jonsson and Boesen.13 The premise of the model is to describe reasonably, the transport of water and solutes in the intermediate region between solution-diffusion model and Poiseuille flow:

• •

the first is reasonable when applied to very dense membranes and solutes which are almost totally rejected, whereas Poiseuille flow can be used to describe the transport through porous membranes consisting of parallel pores. Jonsson and Boesen13 showed that the following equation can be used to determine Rmax from RO experiments:     0 c2 b b t  l Jv þ 1 exp   00 ¼ K K e D2 c2

ð41Þ

where D2 is the solute diffusion coefficient. From Eq. (41), the maximum rejection Rmax (at Jv -1) is given by Rmax ¼ s ¼ 1 

K 1 ¼1K b 1 þ XX23

ð42Þ

21

Eq. (42) shows how rejection is related to a kinetic term (the friction factor b) and to a thermodynamic equilibrium term (K). Spiegler and Kedem9 derived the following corresponding expression:   1 X13 u2 1 þ s¼1K ð43Þ X21 u1 1 þ XX23 21

Eqs. (42) and (43) are identical apart from the correction term X13 u2 =X21 u1 which is much smaller than 1 for highly selective membranes because the solubility of the solute in the membrane must be as low as possible. This can be achieved by a proper choice of the polymer.

86

2.4.3

Fundamentals in Reverse Osmosis

Membrane Charge

The models illustrated in Section 2.4.2 can be applied for a wide range of solutes in neutral membranes. However, a different behavior can be observed with charged solutes in membranes containing fixed charged groups. In fact, membrane composition combined with solvent and solute characteristics can influence rejection via electrostatic double layer interactions or other hindrances: if a solution containing ions is brought in contact with membranes possessing a fixed surface charge, the passage of ions possessing the same charge as the membrane (co-ion) can be inhibited. This condition is termed Donnan Exclusion. Moreover, the membrane can exchange ions between the feed solution and the ion exchange groups on the membrane. This can lead to swelling of the membrane structure and, as a consequence, to variations in the transport properties of the membrane. As a first approximation, the discussed models can be used also in the case of charged membranes, but the transport parameters are a strong function of the operating conditions. For the salt MzyYzm, which ionizes to Mzm þ and Yzy, a dynamic equilibrium occurs when a charged membrane is placed in the salt solution. At equilibrium and in the case of the negatively charged membranes usually utilized in nanofiltration, the following 0 equations for the salt distribution coefficient K* and the rejection R can be used16: K ¼



   zy  zy þzm 1=zm cyðmÞ cy g z ¼ zyy  cy cm gm 0

R ¼ 1  K

ð44Þ ð45Þ

where zi ¼charge of species i, cy and cy(m) ¼ concentrations of co-ion Y in the bulk solution and in the membrane phase respectively, g and gm ¼activity coefficients, c*m ¼ charge capacity of the membrane. Eq. (44) and (45) furnish a qualitative description of the solute rejection, which is function of both the membrane charge capacity, and the solute concentration in the feed and the charge of the ions. However, they do not take into account the diffusive and convective fluxes that are also important in the charged membrane processes.

2.4.4

Limiting Factors: Concentration Polarization, Fouling, Scaling, Biofouling, and Membrane Deterioration

Real RO process is limited by concentration polarization phenomena, membrane fouling/scaling/biofouling/deterioration. These phenomena strongly reduce the performance of membrane operations because they decrease mass flux and/or separation performance, i.e., salt rejection. Consequently, these phenomena negatively affect the economy of the membrane separation processes and their control is one of the major problems in the design of membrane systems.

2.4.4.1 Concentration Polarization Rejection of dissolved matter by the membrane leads to accumulation of these substances in front of the membrane, with highest concentrations directly at the membrane surface. This phenomenon is called concentration polarization. Thus, a concentration gradient between the solution at the membrane surface and the bulk is established which leads to a back transport of the material accumulated at the membrane surface by diffusion. Although concentration polarization can also be found on the permeate side, it is usually neglected in RO since it is much less pronounced than feed side polarization. A typical concentration profile is shown in Fig. 4. Concentration polarization has several negative effects on RO performance: – concentration polarization leads to an increase in the osmotic pressure which is directly proportional to the solute concentration at the membrane surface, and thus a decrease in the trans-membrane flux at constant applied hydrostatic pressure; – the quality of the filtrate is impaired since the solute leakage through the membrane is also directly proportional to the solute concentration at the membrane feed side surface; – particles are accumulated at the membrane which can lead to cake formation on the surface; – especially for divalent ions solubility limits can be exceeded, leading to a precipitation layer on the membrane surface, which negatively influences mass transfer. Concentration polarization complicates the modeling of membrane systems because experimental calculation of the wall concentration is difficult. For high feed flow rates, it has often been assumed that the wall concentration is equal to the bulk concentration due to the high mixing which is however seldom the case. At low flow rates, this assumption is certainly no more applicable and can cause substantial errors. In order to estimate the extent of concentration polarization, the film theory is the most well-used technique,1718:   00 c‴s  cs Jv ¼ exp c0s  c00s k

ð46Þ

Fundamentals in Reverse Osmosis

87

Fig. 4 Concentration profile. From Jonsson, G.; Macedonio, F. Fundamentals in reverse osmosis, chapter 2.01. In E. Drioli and L. Giorno (Eds.). Comprehensive membrane science and engineering, Vol. 2. Elsevier B.V. 2010.

In Eq. (46) k denotes the mass transfer coefficient which can be estimated using a Sherwood correlation such as the following derived by Gekas and Hallstrom.19 Sh ¼ 0:023Re0:8 Sc0:33 f or turbulent flow

ð47Þ


Sh ¼ 1:86  Re  Sc  dh =LÞ0:33 f or laminar flow

ð48Þ

Concentration polarization phenomena can be reduced by promoting a good mixing of the bulk feed solution with the solution near the membrane surface. This goal can be achieved by modifying the membrane module in order to encourage mixing, for example including turbulence promoters in the feed channel, or increasing feed flow rate (thus increasing the axial velocity and promoting turbulent flow).

2.4.4.2 Membrane Fouling/Scaling/Biofouling Membrane fouling is caused by the deposition and/or adsorption of certain feed constituents at the membrane surface, causing a decline in flux over time when all operating parameters, such as pressure, flow rate, temperature, and feed concentration are kept constant. Membrane fouling may be the result of concentration polarization but it may also be just the consequence of adsorption of feed solution constituents at the membrane surface and, especially in microfiltration, also within the membrane structure. Membrane foulants can be classified into four categories depending on the material deposited on membrane surface18: (a) (b) (c) (d)

Chemical foulants, which cause scaling. Physical foulants or particulate matter, which are related to deposition of particles and colloidal matters on the membrane surface. Organic foulants, which can interact with the membrane. Biological foulants, which can either deteriorate the membrane or form a biofilm layer, which inhibits flux across the membrane due to growth of bacteria on the membrane surface.

Although for the first three foulants there are well-established, chemically-based and membrane-based pretreatments, biofouling remains one of the most tenacious and least understood forms of membrane fouling.

2.4.4.2.1

Chemical foulants

Scaling of a reverse osmosis membrane occurs if concentrations of sparingly soluble salts, i.e., divalent and multivalent ions exceed their solubility level. Concentrations in the feed channel inside a module increase, and with increasing recovery, the risk of scaling grows. However, solubility levels only define the minimum concentration level at which scaling might occur. In practical operation, even at higher concentrations, scaling may not occur due to the long induction times of crystallization. However, it is common practice not to exceed solubility limits. Dissolved inorganic chemicals most likely to cause scaling are Ca2 þ , Mg2 þ , CO3 2þ , SO4 2þ , silica and iron. If solubility limits are exceeded; CaCO3; sulfates of calcium, strontium, and barium; CaF2; and various silica compounds are the most likely compounds found as scaling on the membrane surface. Hydroxides of Al, Fe, and Mn are normally precipitated before contact with the membrane. Most natural surface and groundwater display high CaCO3 concentrations close to saturation. Therefore, the scaling tendency of a given feed water is often evaluated using the Langelier Saturation Index (LSI) for brackish waters and the Stiff and Davis Stability Index (S&DSI) for seawaters.18 Carbonate, sulfate and calcium fluoride scaling can be avoided by addition of antiscalants such as organic polymers, surface active agents, organic phosphonates and phosphates, e.g., polyhexametaphosphate (Calgon), which interfere with the kinetics of

88

Fundamentals in Reverse Osmosis

crystal nucleation, formation and/or growth. The presence of silica greatly complicates an RO desalting process. Threshold limits of silica scale precipitation are difficult to predict as they are influenced by a large number of parameters. Another difficulty is the lack of a silica anti-scalant that can be confidently used to extend water recovery limits. Moreover, silica scales deposited on a membrane are difficult and costly to remove. In the presence of silica it is customary to restrict the recovery limits below the silica saturation limit of about 120 mg/L. Antiscalants may allow operation to a silica concentration of at most 220 mg/L.18

2.4.4.2.2

Physical foulants or particulate matter

Particulate fouling is the deposition of suspended solids, colloidal, and microorganism matters on the membrane surface originating from the raw water. The suspended solids and colloidal matter are clay minerals, organic materials, coagulants such as Fe (OH)3 and Al(OH)3, algae, extra cellular polymer substance (EPS), and transparent exopolymer particles. Particulate matter in natural waters can be classified into four different categories depending on particle size:

• • • •

settable solids4100 mm, supra-colloidal solids 1–100 mm, colloidal solids 0.001–1 mm, dissolved solidso10 A1.

The most problematic feeds are those containing colloidal particles not easily removed by granular beds either because of their minute size or because of electrostatic repulsion effects of the media. In such cases it is necessary to add a coagulant or flocculating agent (such as ferric chloride, alum and cationic polymers, the latter can cause membrane fouling difficulties). Particles larger than 425 mm can be easily removed by various pretreatments prior to RO unit such as screens, cartridge filters, dual-media filters etc., whereby the presence of suspended solids can be monitored by the silt density index (SDI) test, turbidity analysis, zeta potential measurement, and particle counting. Membrane manufacturers require a turbidity NTU (Nefelometric Turbidity Units) o0.2, zeta potential 430 mV, and SDIo3–5 to prevent membranes from particulate fouling. Indeed, beach well raw waters are much less loaded with colloidal material and often no further reduction of colloid content is needed.18 Additional source of colloidal matter in systems may arise from corrosion products as with carbon steel pumps, piping, and filters prior to the membrane filtration system. Analysis of the color of the filter after filtration is also interesting for the identification of sticky or particular deposit. Table 1 gives some examples of the filter appearance and the indications about the possible corresponding fouling origin. This is essential to determine whether only suspended solids were in the water or whether it was adsorbed organic matter. Organic foulants can be defined as interaction between organic compound present in the feed water with the membrane surface. Organic matter components consist of proteins, carbohydrates, fats, oil and greases, and aromatic acids such as humic acids. In reality, the humic substances represent the organics in natural waters, whose concentrations range from 0.5 and 20 mg/L in brackish water and up to 100 mg/L in surface seawater.20 Dissolved organics, e.g., humic acids, proteins, carbohydrates, and tannins are the most serious foulants and they are difficult to remove via conventional treatment. Organic matter present in natural waters is undesirable because it is responsible for color in the water, formation of carcinogenic disinfection by-products (DPB's) during water disinfection, complexation with heavy metals and calcium, etc. Moreover, the adsorption of organics on the membrane surface results in permeability decline, which even can be an irreversible process. It was found that mainly the hydrophobic humic substances are deposited on the membrane surface and that the adsorption process is favored with positively charged, high molecular mass compounds. Similarly, the most hydrophilic membranes have been found less prone to fouling by organic colloids, i.e. humic acids. In recent years membrane processes have been advanced for the removal of natural organic matter (NOM) for potable and other water uses. Several aspects of such processes have been the subject of intense research efforts with emphasis on NOM removal efficiency and on the inevitable fouling of the membranes, which limits their performance and lifetime. Important NOM properties relating to membrane performance are the nature of organic compounds, their hydrophilicity and charge, and the molecular weight distribution. Equally, important membrane properties are their pore size or MWCO, surface charge, and hydrophilicity. In addition, water properties such as pH and ionic strength, as well as the presence of specific ions such as calcium, have been recognized to play a prominent role in NOM adsorption and fouling of membranes. Natural organic matter compounds are divided into humic substances or poly-hydroxy aromatics, and nonhumic substances such as proteins, polysaccharides, and Table 1 Origin of the fouling compounds according to SDI membrane appearance Color

Identification

Yellow/brown Red/brown Dark/grey Particles

Organics Iron Activated carbon Suspended solids

From Mosset, A.; et al., The sensitivity of SDI analysis: from RO feed water to raw water, Desalination 2008, 222, 17–23.

Fundamentals in Reverse Osmosis

89

amino-sugars. Humic substances are more hydrophobic than nonhumic substances and constitute a significant fraction of surface water NOM. NOM is regarded as a carbon skeleton to which various functional groups are attached. The main components of this skeleton are Aliphatic units “straight chained or branched carbon units” and Aromatic units “based on benzene ring”.

2.4.4.2.3

Biological foulants

The presence of microorganisms is ubiquitous. All raw waters contain microorganisms such as bacteria, algae, fungi, viruses, and higher organisms such as protozoa, living or dead, or biotic debris such as bacterial cell wall fragments. The difference between microorganisms and nonliving particles is the ability of microorganisms to reproduce and form a biofilm under favorable conditions. Consequently, biofouling is due to the growth of biofilm (bacterial) on the membrane’s surfaces. Microorganisms entering a RO/NF system, find a large membrane surface where dissolved nutrients from the water are enriched due to concentration polarization, thus creating an ideal environment for the formation of a biofilm. Biological fouling of the membranes may seriously affect the performance of the RO system. The symptoms are an increase in the differential pressure from feed to concentrate, finally leading to telescoping and mechanical damage of the membrane elements and a decline in membrane flux. Sometimes, biofouling develops even on the permeate side, thus contaminating the product water. A biofilm is difficult to remove because it protects its microorganisms against the action of shear forces and biocide chemicals. In addition, if not completely removed, remaining parts of a biofilm lead to a rapid regrowth. Therefore, enhanced pretreatment process and microbiological activity control lead to biological fouling prevention. New techniques for biofouling control are also developing. One example is quorum quenching that is based on interrupting the quorum sensing between the microorganisms responsible for cell to cell communication, formation of biofilm and excretion of extracellular polymeric substances21. The immobilized quorum quenching enzymes have proven very useful in controlling the biofilm formation by interrupting the quorum sensing. Due to the problems associated with enzyme based quorum quenching (difficult extraction and purification and instability), the new efforts focus on the utilization of bacteria that produce quorum quenching enzyme.21,22 In conclusion, fouling adversely affects membrane systems for the following reasons: – membrane flux decline resulting from the formation of a permeability-reducing film on the membrane surface; – membrane biodegradation due to the production of acidic by-products by microorganisms, which are concentrated at the membrane surface where they can cause the most damage; – increased salt passage thereby reducing the quality of the product water; – increase in energy consumption. To maintain the same production rate differential pressure and feed pressure must be increased to counteract the reduction in permeability brought on by the increase in resistance that the fouling causes. But, damage to the membrane elements may occur if the operating pressure exceeds the manufacturer's recommendations. While concentration polarization can be minimized by hydrodynamic means, the control of membrane fouling is more difficult. Fouling can be prevented through the following means: pretreatment of the feed solution; modification of membrane surface; hydrodynamic optimization of the membrane module; recourse to proper chemical agents for cleaning; and back-flushing. Particulate fouling in current practice is inhibited by mechanical pretreatment of the RO feed water by use of screens, sand filtration, and cartridge filters or membrane pretreatment. Biological fouling, caused by microorganisms sticking to the membrane producing a gel like layer, is a serious problem to the operation of a RO plant and has to be prevented by chlorination in pretreatment prior to the actual RO stage. Fouling can never fully be prevented even with optimized pretreatment. Therefore, periodical membrane cleaning has to be performed. Good operating practice calls for chemical cleaning of the membranes if either normalized permeate flow decreases by 10%, feed channel pressure loss increases by 15%, or normalized salt rejection decreases by 10% from initial conditions during the first 48 h of plant operation.23 However, complete fouling removal is not possible and it has to be tolerated up to a decrease of mass flux down to 75% of original flux.24

2.4.4.3 Membrane Deterioration Various chemicals can harm the active layer of the membrane, leading to irreversible damage associated with reduced rejection capability and even destruction of the membrane.18 Oxidants used in pretreatment of the reverse osmosis feed water or as cleaning chemicals are the most important group of chemicals responsible for membrane deterioration. Presence of even trace amounts of these compounds may oxidize the membrane surface and damage the active membrane layer. Membrane suppliers therefore give restrictions on exposure to oxidants. In addition, polymeric membranes are more or less susceptible to very low or high pH values. Therefore, pH adjustment and control is necessary to ensure stable operation.

2.4.5

Membrane Materials for RO

For efficient processes, membranes should display high flux and high rejection. Not all the materials are suitable for every NF/RO operation because the constant A and B (in Eqs. (13) and (15)) must have optimal values for a given application. Moreover, solvent flux through the membrane is approximately inversely proportional to the membrane thickness. Therefore, RO membranes have an asymmetric structure, with a thin dense top layer (thicknessr1 mm) supported by a porous sublayer (thickness in the

90

Fundamentals in Reverse Osmosis

range of 50–150 mm). The selectively permeable layer is reduced to a very fine skin in order to limit the resistance to transfer related to the layer thickness. This layer rests upon another thicker substrate that has much larger pores which intends to provide the membrane with satisfactory mechanical properties without significantly impeding the flow of water. In the early 1960s the first asymmetric reverse osmosis membranes were produced by Loeb and Sourirajan.25 These membranes showed up to 100 times higher flux than any symmetric membranes known. This development paved the way for the commercial success of reverse osmosis. On the basis of the internal structure, there are two main types of asymmetric membranes for NF/RO: asymmetric homogeneous membranes and composite membranes.

•

•

In asymmetric homogeneous membranes both top-layer and sublayer consist of the same material. Cellulose esters (especially cellulose diacetate and triacetate) were the first commercially used materials, in particular for water desalination due to their high permeability towards water and low solubility towards salts. Unfortunately, these materials have poor chemical stability and tend to hydrolyze over time depending on temperature and pH operating conditions (typical operation conditions of cellulose ester membranes are over the pH range 5–7 and at a temperature below 301C). They are also subjected to biological degradation. Other materials frequently used for RO membrane are aromatic polyamides, polybenzimidazoles, polybenzimidazolones, polymidehydrazide, and polyimides.26 Polyimides can be used over a wider pH range, approximately from 5 to 9. The main drawback of polyamides (or polymers with an amide group –NH-CO in general) is their susceptibility against free chlorine Cl2 which causes degradation of the amide group. Composite membranes are made by assembling two distinct parts composed of different polymeric materials: a very fine layer (0.05–0.5 mm), representing the salt barrier of perm-selective material (i.e., polyamide) obtained through interfacial polymerization of the microporous layer (30–50 mm) made, for example, in polysulphone, which is itself often asymmetrical and all of which is attached to a support medium (100–150 mm). Composite membranes can combine various materials and provide optimum properties depending on their use.

2.4.6

Membrane Modules for RO

The application, efficiency, and economics of an RO process also depend on the packaging of the membranes. There are four possible membrane geometries: 1. Spiral wound membrane, which consists of consecutive layers of large membrane and support material in an envelope type design rolled up around a perforated steel tube. This design tries to maximize surface area in a minimum amount of space. It is less expensive but more sensitive to pollution due to its manufacturing process. Spiral membranes are only used for nanofiltration (NF) and RO applications. 2. Plate and frame modules make use of flat-sheet membranes (in sandwich configuration) separated by support plates. These modules have low packing densities and are correspondingly expensive; they are primarily used to produce potable water in small-scale applications 3. Tubular membrane. Generally used for viscous or bad quality fluids, tubular membranes are not self-supporting membranes. They are located on the inside of a tube, made of a special kind of microporous material. This material is the supporting layer for the membrane. Because the feed solution flows through the membrane core, the permeate passes through the membrane and is collected in the tubular housing. The main cause for this is that the attachment of the membrane to the supporting layer is very weak. Tubular membranes have a diameter of about 5–15 mm. Because of the size of the membrane surface, plugging of tubular membranes is not likely to occur. Therefore, these modules do not need a preliminary pretreatment of the water. The main drawback is that tubular membrane is not very compact and has a high cost per m2 installed. 4. Hollow fiber membrane. The modules contain several small tubes or fibers (diameter of below 0.1 mm), consequentially the chances of plugging of a hollow fiber membrane are very high. The membranes can only be used for the treatment of water with a low suspended solids content. The packing density of a hollow fiber membrane is very high. Hollow fiber membranes are nearly always used merely for NF and RO. Table 2 presents some general characteristics of the four basic membrane-module types.

2.4.7

New Materials for RO Membranes

To date, all commercial RO membrane comprises polar or hydrophilic pores and only polymeric membranes have been employed for industrial use. However, the advances in conventional polymeric RO membrane have been rather limited since the late 1990s. Recently, advances in nanotechnology have led to the development of nanostructured materials which may form the basis for new RO membranes. Among the technological developments, carbon nanotubes (CNTs) and other carbon-based materials, like graphene and graphene oxide (GO), as well as inorganic membrane, mixed matrix membranes (MMMs), and biomimetic membranes are emerging as developed membranes with superior permeability, durability, and selectivity in particular for water purification.

Fundamentals in Reverse Osmosis

Table 2

91

General characteristics of RO membrane modules Module type

Typical packing density (m2/m3) Required feed flow rate (m3/m2 s) Feed side pressure drop (kg/cm2) Membrane fouling propensity Ease to cleaning Typical feed stream filtration requirements Relative expense

Spiral-wound 800 0.25–0.5 3–6 High Poor to good 10–25 mm filtration Low

Hollow-fiber 6000 B0.005 0.1–0.3 High Poor 5–10 mm Low

Tubular 70 1–5 2–3 Low Excellent Not required High

Palate-and frame 500 0.25–0.5 3–6 Moderate Good 10–25 mm filtration High

2.4.7.1 Ceramic/Inorganic Membranes Ceramic membranes are mostly made from alumina, silica, titania, zirconia, or any mixture of these materials. Due to the high manufacturing cost, use is currently limited to applications where polymeric membranes cannot be used (i.e., high operating temperatures, radioactive/heavily contaminated feeds, and highly reactive environments27). The interest in ceramic membranes is due, in particular, to their robustness. Moreover, molecular dynamic simulation results showed 100% of ion rejection by perfect all-Si ZK-4 zeolite membranes.28 Though the improvement of zeolite membranes has been tremendous in the past 10 years, their performance and economics are still no match for polymeric membranes.29 The zeolite membrane thickness is still at least three times higher than the current state of the art polymeric RO membranes, causing higher resistance to water flux. Consequently, ceramic membranes require at least 50 times higher membrane area than polymeric ones to achieve an equivalent production capacity. This value can be even higher when the higher density and lower packing effectiveness are considered. Moreover, while zeolite membranes are claimed to have high organic rejection, organic fouling has caused almost 25% loss in flux after only 2 h of operation, though full recovery of flux was achieved after chemical washing.30

2.4.7.2 Mixed Matrix Membranes The concept of MMM, the combination of organic and inorganic material, is not new but started in 1980 in the field of gas separation. The incorporation of inorganic materials into organic RO thin-film composite membranes only started in the early 2000s.31 The main objective of MMM is to combine the benefits offered by each material, i.e., the high packing density, good permselectivity, and long operational experience of polymeric membranes, coupled with the superior chemical, biological, and thermal stability of inorganic membranes.32 An example can be found in the TiO2 nanoparticles self assembled aromatic polyamide thin-film-composite (TFC) membrane.33 Titanium oxide (TiO2) is a well known photocatalytic material, widely used for disinfection and decomposition of organic compounds,34 and these properties make it interesting as an antifouling coating. Testing with E. coli-containing feed water showed that TiO2 nanoparticles self assembled aromatic polyamide TFC membrane has superior antibiofouling properties, especially with the aid of UV excitation, without compromising the flux and salt rejection performance of the original membrane. Zeolite nanoparticles have also been used to prepare MMMs. Different RO membranes with various zeolite loadings were prepared and consequent changes in membrane characteristics were observed, i.e., the membranes were smoother, more hydrophilic, and more negatively charged with increasing nanoparticle loading.35 The MMM membrane exhibited 90% of flux and a slight improvement in salt rejection relative to the hand cast TFC membrane without zeolite nanoparticles.

2.4.7.3 Carbon Nanotubes CNTs have been studied extensively, especially in the past 20 years, owing to their broad range of applications.36 A series of simulations and recent measurements of water flow in CNT sand between flakes of GO have predicted and demonstrated that, without hydrogen bonding to the walls of pores, i.e., sans the no-slip condition, water should and does pass through such hydrophobic channels orders of magnitude faster than in hydrophilic pores.37–42 Proper alignment of CNTs in a membrane skin layer is very difficult, so there are far fewer experimental data available than from simulations. CNTs must be aligned and at high density to achieve good permeability. To reduce the energy barrier that must be overcome for water to enter a hydrophobic pore, the ends of the CNTs could be functionalized with hydrophilic groups.35 This can drastically improve performance by increasing flux, mechanical and thermal stability, and fouling resistance.43 It is also possible to attach different metal nanoparticles (such as are copper (Cu), silver (Ag), platinum (Pt), and titanium dioxide (TiO2)) to the CNTs to enhance antibacterial and antibiofouling effects.44 Experimental results by Holt et al.38 showed that the flow rate of water through CNTs is three orders of magnitude higher than that predicted from no-slip hydrodynamic flow by the Hagen-Poiseuille equation. When the pore size is less than 20 Å , the permeability is higher than that of conventional polycarbonate membranes.35 Two more additional benefits of CNTs are: (i) their antibacterial property (CNTs are able to rupture bacterial cells, disrupt metabolic pathways, and cause oxidative stress (148)), and (ii) their energy consumption (pore size of CNTs ranges between RO and nanofiltration membranes, but CNTs do not require a high pressure owing to nearly frictionless flow through them, unless incorporated into a membrane). However, synthesis of CNTs

92

Fundamentals in Reverse Osmosis

and implementation into membrane skin layers are very difficult experimentally: more work is needed in the development of efficient synthesis methods to align arrays of single-walled CNTs, with subnanometre diameters, and also the development of tipfunctionalization for more efficient salt rejection.45

2.4.7.4 Graphene and Graphene Oxide Graphene-based materials are considered superhydrophobic owing to their extremely high water contact angle (4150 degree). Pure graphene can be produced using methods such as chemical vapor deposition, lithography, templating, electrospinning, electrodeposition, sol–gel processes, and layer-by-layer deposition to form different surface and roughness properties.46,47 Membranes can also be produced from diamond-like carbon (DLC),48,49 and can be synthesized using chemical vapor deposition of various organic compounds. The structure has a 12% porosity and pore sizes r1 nm. DLC membranes are very stiff, with elastic moduli B50 times greater than engineering thermoplastics and only 10 times less than carbon. These membranes were tested in laboratory for filtering organic solutes greater than 1 nm from organic solvents.35 They have solvent fluxes up to three orders of magnitude higher than NF membranes typically used for organic solvents. Graphene that is more porous can be obtained either by inducing defects into the layered structure (via chemical etching50 or irradiation with an electron beam,51,52 or ion bombardment,53) or by chemical modification of graphene to make it more hydrophilic. The resultant oxidized material is known as GO that has gained a lot of attention in recent years owing to its unusually high water-transport properties. In 2012, Nair et al.40 demonstrated that water could pass through layered GO sheets (flakes) at extremely high rates. The GO sheets can be stacked on top of each other, forming layers between 0.1 mm and 10 mm thick. The average distance between each sheet is B10 Å .35,54,55 When the GO sheets were chemically reduced, the membranes became 100 times less permeable to water because pore size decreased from 10 to 4 Å . Molecular dynamics simulations showed that water was unable to fill the capillaries for interlayer spacing lower than 6 Å , but when the spacing was 410 Å , two or more layers of water were able to form between the sheets. Belfort et al.56 indicated that optimal interlayer spacing is between 6 and 10 Å to form a monolayer of water between sheets. Nair et al.40 declared that at this scale, water is able to permeate at a velocity on the order of 1 m s1. The major drawbacks are that (i) it is difficult to control the extent of oxidation or reduction of graphene to form GO, (ii) GO is expensive to synthesize. Moreover, because GO membranes are made up of stacked GO flakes, the species must travel a torturous path; therefore, single-layer graphene is predicted to have better performance than GO owing to its shorter path length.49 However, single-layer graphene is far too brittle whereas multiple stacked layers of graphene increase resistance and path length.35 GO can also be chemically modified to form GO frameworks (GOFs). GOFs are a class of nanoporous materials consisting of layers of GO sheets covalently interconnected by linear boronic acid (or with other similar chemistries) pillaring units, also called linkers.35,57 Imbrogno et al.35 indicate that high water permeability and salts rejection 499.9% can be achieved with GOF membranes. Moreover, Mi58 reports that GOF membranes selectively retain different species, such as ions (desalination), polyelectrolytes (fuel or chemical purification), or nanoparticles (biomedical filtration).

2.4.7.5 Biomimetic RO Membranes The excellent water transport properties of biological membranes has led to the study of membranes incorporating aquaporins (AQPs), which are proteins functioning as water-selective channels in biological cell membranes.59 AQPs are water conducting channels found in biological membranes and have a unique hourglass architecture with a “pore opening” of 2.8 Å ; the narrow pore prevents the passage of large molecules.60 Membranes incorporating bacterial Aquaporin Z proteins have been reported to show at least an order of magnitude improvement in permeability compared to commercially available TFC RO membranes.61 Many practical issues, such as identification of appropriate support materials, understanding of the resistance to membrane fouling, and even identification of an appropriate range of operating conditions must be carried out to develop this membrane for practical use.27

2.4.8

Conclusions

Today reverse osmosis represents the clearest example of membrane technology success, in particular due to its increasingly important role in water supply, both via desalination (seawater and brackish water) and via water reclamation. Commercial interest in RO technology is increasing globally due to continuous process improvements. These advances include developments in RO membrane material, structure, and morphology (in order to improve permeability, selectivity, and mechanical/chemical/ biological stability). Alongside the advancements in RO membranes, the development of other aspects of RO technology has undeniably made RO desalination more efficient and economic, such as advancements in module and process design, feed pretreatment and reduction in energy consumption, concentration polarization, and fouling phenomena. Despite major earlier breakthroughs such as the Loeb-Sourirajan asymmetric membrane (1960s), fully cross linked aromatic TFC membrane (1970s to 1980s), and controlling morphological changes by monitoring polymerization reactions (1990s), the evolutionary improvement of a commercial RO membrane has been rather slow during the first decade of this century.29 One reason for this is that the development of thin-film composite membranes with selectivity higher than the existing RO commercial membrane modules (between 99.40% and 99.80%29) is difficult. This is a direct consequence of the separation mechanism of thin-film composite

Fundamentals in Reverse Osmosis

93

membranes, where increasing selectivity to allow higher removal of ions will substantially reduce the membrane permeability and will increase energy consumption. Developing RO membranes with higher selectivity without sacrificing water permeability will necessitate a major paradigm shift, as it will require membranes that do not follow the solution-diffusion mechanism.62 Various nanostructured RO membranes have been proposed that offer attractive characteristics and that could possibly bring revolutionary advancements (for example the mixed matrix membranes). However, the development of such membranes is only in the initial stages and many problems are yet to be overcome, such as the high cost of nanostructured materials and the difficulty in scaling up nanomembrane manufacturing processes.

See also: 4.6 Membrane Systems for Seawater and Brackish Water Desalination. 4.7 The Most Advanced Membrane Analysis and the SaveEnergy Type Membrane-Low-Pressure Seawater Reverse Osmosis Membrane Developed by “Mega-ton Water System” Project. 4.10 Basic Aspects and Applications of Membrane Processes in Agro-Food and Bulk Biotech Industries

References 1. Fritzmann, C.; Löwenberg, J.; Wintgens, T.; Melin, T. State-of-the-art of reverse osmosis desalination. Desalination 2007, 216, 1–76. 2. Jonsson, G.; Macedonio, F. Chapter 2.01: Fundamentals in reverse osmosis. In Comprehensive membrane science and engineering; Drioli, E.; Giorno, L., Eds.; vol.2, Elsevier B.V, 2010. 3. Jonsson, G. Selectivity in membrane filtration. In Synthetic membranes: science, enginering and applications; Bungay, P. M., et al., Eds.; D. Reidel Publishing Company: Dordrecht, 1986. 4. Staverman, A. J Rec Trav Chim Pays-Bas 1951, 70, 344. 5. Jonsson, G. Methods for determining the selectivity of reverse osmosis membranes. Desalination 1978, 24, 19–37. 6. Pusch, W. Ber Bunsenges Phys Chem 1977, 81, 269. 7. Onsager, L. Phys Rev 1931, 37, 405–425. 8. Kedem, O.; Katchalsky, A. Biochem Biophys Acta 1958, 27, 229–246. 9. Spiegler, K. S.; Kedem, O. Desalination 1966, 1, 311. 10. Burghoff, H.-G.; Lee, K. L.; Push, W. J Appl Polym Sci 1980, 25, 323–347. 11. Sourirajan, S., Ed.; Reverse osmosis and synthetic membranes; National Research Council Canada: Ottawa, 1977. 12. Sherwood, T. K.; Brian, P. L. T.; Fisher, R. E. Ind Eng Chem Fundamentals 1967, 6, 2–12. 13. Jonsson, G.; Boesen, C. E. Desalination 1975, 17, 145–165. 14. Merten, U. In Desalination by reverse Osmosis; Merten, U., Ed.; M.I.T. Press: Cambridge, 1966; pp 15–54. 15. Spiegler, K. S. Trans Faraday Soc 1958, 54, 1408–1428. 16. Bhattacharyya, D.; Cheng, C. Separation of metals chelates by charged composite membranes. In Recent developments in separation science; Li, N. N., Ed.; vol. 9, CRC Press: Boca Raton, FL, 1986; p 707. 17. Bhattacharyya, D.; Williams, M. E. Chapter VI Reverse osmosis. In Membrane handbook; Winston Ho, W. S.; Sirkar, K. K., Eds.; Springer: Berlin, 1992. 18. Fritzmann, C.; Löwenberg, J.; Wintgens, T.; Melin, T. State-of-the-art of reverse osmosis desalination. Desalination 2007, 216, 1–76. 19. Gekas, V.; Hallstrom, B. Mass transfer in the membrane concentration polarization layer under turbulent cross flow. J Membr Sci 1987, 30, 153. 20. Redondo, J. A.; Lomax, I. Experiences with pretreatment of raw water with high fouling potential for reverse osmosis plant using FILMTEC™, membranes. Desalination 1997, 110, 167–182. 21. Kim, S.; Oh, H.; Jo, S.; Yeon, K.; Lee, C.; Lim, D.; Lee, C.; Lee, J. Biofouling control with bead-entrapped quorum quenching bacteria in membrane bioreactors: physical and biological effects. Environ Sci Technol 2013, 47, 836–842. 22. Oh, H.-S.; Yeon, K.-M.; Yang, C.-S.; Kim, S.-R.; Lee, C.-H.; Park, S. Y.; Han, J. Y.; Lee, J.-K. Control of membrane biofouling in MBR for wastewater treatment by quorum quenching bacteria encapsulated in microporous membrane. Environ Sci Technol 2012, 46 (9), 4877–4884. 23. Cleaning chemicals, DOW FILMTEC Membranes, Tech Manual Exerpt, Form No. 609-02091-704. 24. Rautenbach, R.; Melin, T. Membranverfahren. Grundlagen der Modul- und Anlagenauslegung, 2nd ed., 2003. 25. Loeb, S.; Sourirajan, S. High flow porous membranes for separation of water from saline solutions, U.S. Patent 3,133,132, 1964. 26. Mulder, M. Basic principles of membrane technology; Kluwer Academic Publishers: London, 1996. 27. Siskens, C. A. M. Chapter 13 Applications of ceramic membranes in liquid filtration. In Membrane science and technology; Burggraaf, A. J.; Cot, L., Eds.; Elsevier: Amsterdam, 1996; pp 619–639. 28. Lin, J.; Murad, S. A computer simulation study of the separation of aqueous solutions using thin zeolite membranes. Mol Phys 2001, 99, 1175–1181. 29. Lee, K. P.; Arnot, T. C.; Mattia, D. A review of reverse osmosis membrane materials for desalination-development to date and future potential. J Membr Sci 2011, 370, 1–22. 30. Lu, J.; Liu, N.; Li, L.; Lee, R. Organic fouling and regeneration of zeolite membrane in wastewater treatment. Sep Purif Technol 2012, 72, 203–207. 31. Okumus, E.; Gurkan, T.; Yilmaz, L. Development of a mixed-matrix membrane for pervaporation. Sep Sci Technol 1994, 29, 2451–2473. 32. Ismail, A. F.; Goh, P. S.; Sanip, S .M.; Aziz, M. Transport and separation properties of carbon nanotube-mixed matrix membrane. Sep Purif Technol 2012. 33. Kim, S. H.; Kwak, S.-Y.; Sohn, B.-H.; Park, T. H. Design of TiO2 nanoparticle selfassembled aromatic polyamide thin-film-composite (TFC) membrane as an approach to solve biofouling problem. J Membr Sci 2003, 211, 157–165. 34. Sunada, K.; Kikuchi, Y.; Hashimoto, K.; Fujishima, A. Bactericidal and detoxification effects of TiO2 thin film photocatalysts. Environ Sci Technol 1998, 32, 726–728. 35. Jeong, B.-H.; Hoek, E. M. V.; Yan, Y.; Subramani, A.; Huang, X.; Hurwitz, G.; Ghosh, A. K.; Jawor, A. Interfacial polymerization of thin film nanocomposites: a new concept for reverse osmosis membranes. J Membr Sci 2007, 294, 1–7. 36. LeDuc, Y.; Michau, M.; Gilles, A.; Gence, V.; Legrand, Y.-M.; et al. Imidazole-quartet water and proton dipolar channels. Angew Chem Int Ed 2011, 50, 11366–11372. 37. Imbrogno, J.; Belfort, G. Membrane desalination: where are we, and what can we learn from fundamentals? Annu Rev Chem Biomol Eng 2016, 7, 29–64. 38. Peter, C.; Hummer, G. Ion transport through membrane-spanning nanopores studied bymolecular dynamics simulations and continuum electrostatics calculations. Biophys J 2005, 89, 2222–2234. 39. Striolo, A. The mechanism of water diffusion in narrow carbon nanotubes. Nano Lett 2006, 6, 633–639. 40. Holt, J. K.; Park, H. G.; Wang, Y.; Stadermann, M.; Artyukhin, A. B.; et al. Fast mass transport through sub-2-nanometer carbon nanotubes. Science 2006, 312, 1034–1037. 41. Paul, D. R. Creating new types of carbon-based membranes. Science 2012, 335, 413–414.

94

Fundamentals in Reverse Osmosis

42. Nair, R.; Wu, H.; Jayaram, P.; Grigorieva, I.; Geim, A. Unimpeded permeation of water through helium-leak–tight graphene-based membranes. Science 2012, 335, 442–444. 43. Majumder, M.; Corry, B. Anomalous decline of water transport in covalently modified carbon nanotube membranes. Chem Commun 2011, 47, 7683–7685. 44. Van Hooijdonk, E.; Bittencourt, C.; Snyders, R.; Colomer, J.-F. Functionalization of vertically aligned carbon nanotubes. Beilstein J Nanotechnol 2013, 4, 129–152. 45. Mauter, M. S.; Elimelech, M. Environmental applications of carbon-based nanomaterials. Environ Sci Technol 2008, 42, 5843–5859. 46. Darmanin, T.; de Givenchy, E. T.; Amigoni, S.; Guittard, F. Superhydrophobic surfaces by electrochemical processes. Adv Mater 2013, 25, 1378–1394. 47. Darmanin, T.; Guittard, F. Wettability of conducting polymers: from superhydrophilicity to superoleophobicity. Prog Polym Sci 2014, 39, 656–682. 48. Karan, S.; Samitsu, S.; Peng, X.; Kurashima, K.; Ichinose, I. Ultrafast viscous permeation of organic solvents through diamond-like carbon nanosheets. Science 2012, 335, 444–447. 49. Robertson, J. Diamond-like amorphous carbon. Mater Sci Eng R Rep 2002, 37, 129–281. 50. Koenig, S. P.; Wang, L.; Pellegrino, J.; Bunch, J. S. Selective molecular sieving through porous graphene. Nat Nanotechnol 2012, 7, 728–732. 51. Garaj, S.; Liu, S.; Golovchenko, J. A.; Branton, D. Molecule-hugging graphene nanopores. Proc Natl Acad Sci U S A 2013, 110, 12192–12196. 52. Merchant, C. A.; Healy, K.; Wanunu, M.; Ray, V.; Peterman, N.; et al. DNA translocation through graphene nanopores. Nano Lett 2010, 10, 2915–2921. 53. O’Hern, S. C.; Boutilier, M. S. H.; Idrobo, J.-C.; Song, Y.; Kong, J.; et al. Selective ionic transport through tunable subnanometer pores in single-layer graphene membranes. Nano Lett 2014, 14, 1234–1241. 54. Dikin, D. A.; Stankovich, S.; Zimney, E. J.; Piner, R. D.; Dommett, G. H. B.; et al. Preparation and characterization of graphene oxide paper. Nature 2007, 448, 457–460. 55. Eda, G.; Chhowalla, M. Chemically derived graphene oxide: towards large-area thin-film electronics and optoelectronics. Adv Mater 2010, 22, 2392–2415. 56. Belfort, G.; Sinai, N. Relaxation studies of adsorbed water in porous glass. In Presented at Water in Polymers: 178th Meet. American Chemical Society, Washington, DC, 1980. 57. Burress, J. W.; Gadipelli, S.; Ford, J.; Simmons, J. M.; Zhou, W.; Yildirim, T. Graphene oxide framework materials: theoretical predictions and experimental results. Angew Chem Int Ed Engl 2010, 49, 8902–8904. 58. Mi, B. Graphene oxide membranes for ionic and molecular sieving. Science 2014, 343, 740–742. 59. Agre, P. Membrane water transport and aquaporins: looking back. Biol Cell 2005, 97, 355–356. 60. Macedonio, F.; et al. Chem Eng Process 2012, 51, 2–17. 61. Kumar, M.; Grzelakowski, M.; Zilles, J.; Clark, M.; Meier, W. Highly permeable polymeric membranes based on the incorporation of the functional water channel protein aquaporin Z. Proc Natl Acad Sci U S A 2007, 104, 20719–20724. 62. Elimelech, M.; Phillip, W. A. Science 2011, 333, 712–717.

Further Reading Mosset, A.; et al. The sensitivity of SDI analysis: from RO feed water to raw water. Desalination, 2008, 222, 17–23.