The effects of heat treatment on membrane fouling resistance and the rejection of small and neutral solutes by reverse osmosis (RO) membranes were elucidated. RO membrane modification by heat treatment reduced fouling and improved boron rejection. However, heat treatment also caused a decrease in the water permeability of RO membranes. Significant improvement on fouling resistance by heat treatment was observed when RO concentrate was used to simulate a feed solution with high fouling propensity. The improved fouling resistance is likely to be due to changes in the hydrophobic interaction between the membrane surface and foulants. Boron rejection by the ESPA2 membrane was enhanced by heat treatment from 26 to 68% (when evaluated at the permeate flux of 20 L/m2 h). Positron annihilation lifetime spectroscopy revealed that heat treatment did not significantly influence the free-volume hole-radius of the membrane active skin layer. The results reported in this study suggested that changes in the other membrane properties such as free-volume fraction and thickness may be the main cause improving boron rejection.

INTRODUCTION

Water recycling is an important strategy in arid regions to ensure a secured supply of water for municipal use (Shannon et al. 2008). In a typical water recycling scheme, wastewater is first treated to the secondary or tertiary effluent standard. The effluent is reclaimed and further purified by a series of advanced treatment processes and then used to replenish drinking water reservoirs or underground aquifers. Where necessary, this recycled water can be extracted and distributed to the population for potable use. Advanced treatment processes used for water recycling include microfiltration or ultrafiltration, reverse osmosis (RO) filtration, and advanced oxidation. Among these treatment processes, RO filtration is a key barrier against pathogenic agents, dissolved salts and trace organic chemicals. Given the high organic content of reclaimed effluent, the RO process is prone to membrane fouling, which leads to a decrease in membrane permeability and changes in separation performance (Xu et al. 2010). Membrane fouling in RO filtration systems involves a higher energy consumption and more frequent chemical cleaning, thus ultimately reducing the sustainable value of water recycling.

Membrane fouling depends on the interaction between the physicochemical characteristics of the RO membrane surface and foulants. A recent study by Fujioka & Nghiem (2013) demonstrated that the fouling resistance of commercially available polyamide low pressure reverse osmosis (LPRO) membranes can be improved by heat treatment. This modification was performed by soaking commercial RO membranes into hot water (e.g. 70 °C) for a certain period of time. As a result of heat treatment, the progress of membrane fouling was retarded by over 50% when sand-filtered wastewater was used for filtration (Fujioka & Nghiem 2013). Nevertheless, no previous studies have investigated the effects of heat treatment using actual treated wastewater with high fouling propensity – RO feed solutions and RO concentrates.

Heat treatment performed during or immediately after the interfacial polymerisation process can also improve salt rejection (Shintani et al. 2009). In addition to salts, the removal of specific contaminants such as boron and trace organic chemicals is also important when the recycled water is used for irrigation, industry or potable purposes. Boron is toxic to many plant species (e.g. citrus, onion, wheat, and barley) at concentrations as low as 0.3 mg/L (Tu et al. 2010). In some cases, when recycled water is used in thermal power plants and subsequently irrigation, due to evaporation in the cooling towers, the concentration of boron in recycled water must be reduced to 0.15 mg/L or lower. Owing to its neutral and low molecular weight (LMV) properties, the rejection of boron by LPRO membranes that are generally used for water recycling applications at the environmental pH (e.g. pH 6–8) has been reported to be typically less than 60% (Tu et al. 2010). Elevated boron concentration in the RO feed occurs in places where wastewater is affected by seawater intrusion; thus, the removal of boron in these locations is of great importance. Similarly, the occurrence of trace organic chemicals such as N-nitrosodimethylamine (NDMA) in reclaimed effluent is of great concern to human health. In Australia, the USA, and several other countries, the NDMA concentration limit in recycled water intended for potable water reuse has been regulated at 10 ng/L. Although the analysis of NDMA at a concentration of several ng/L is challenging, recent research has shown a strong correlation between boron and NDMA rejections by RO membranes (Tu et al. 2013). Thus, in this study, boron is selected as a model small and neutral solute of concern in water recycling applications.

The aim of this study was to elucidate the effects of heat treatment on fouling resistance and the rejection of small and neutral solutes. Untreated and heat-treated RO membranes were characterised by examining surface properties, permeability and boron separation performance. The impact of the heat treatment on membrane fouling was examined using actual RO feed and RO concentrates.

MATERIALS AND METHODS

Membranes and membrane filtration system

Two RO membranes, commercially known as ESPA2 and TFC-HR, were supplied by Hydranautics (Oceanside, CA, USA) and Koch Membrane Systems (Wilmington, MA, USA), respectively. They are composite polyamide RO membranes that comprise a thin polyamide active skin layer on the top of porous supporting layers. A laboratory-scale cross-flow RO filtration system used in this investigation is depicted in Figure S1 of the Supplementary Material (available online at http://www.iwaponline.com/ws/015/135.pdf).

Chemicals

Analytical grade NaCl, CaCl2, NaHCO3 and boric acid were supplied by Ajax Finechem (Taren Point, NSW, Australia). RO feed and concentrates were collected from an RO system which is from a water recycling plant in Australia. RO feed and concentrate samples from the first, second and third stage (Table S1 of the Supplementary Material, online at http://www.iwaponline.com/ws/015/135.pdf) were used to evaluate the fouling propensity.

Experimental protocols

The membrane samples were rinsed with Milli-Q water to remove preservative materials from the membrane surface. They were then immersed in Milli-Q water at 80 ± 0.1 °C for 4 h. The water temperature was regulated using a temperature-controlled water bath (TWB-12D, Thermoline Scientific, Wetherill Park, NSW, Australia).

Each rejection experiment started with a compaction step where the permeate flux of membrane sample was stabilised at 1,800 kPa using Milli-Q feed water. During the experiments the cross-flow velocity and feed temperature were maintained at 0.42 m/s and 20.0 ± 0.1 °C, respectively. Following the compaction step, the feed solution was conditioned at 20 mM NaCl, 1 mM CaCl2, 1 mM NaHCO3 and 5.0 mg/L boron. The permeate flux was then adjusted to 20 L/m2h.

The membrane fouling experiments also started with the compaction step described above. Feed solution was then replaced with the RO feed, or the first, second or third stage RO concentrate. Thereafter, the filtration system was operated for 15 h or until flux decline reached 50% at a feed pressure of 700 kPa (untreated membrane) and 1,000 kPa (heat-treated membrane) both of which corresponded to approximately 30 L/m2h permeate flux. The permeate flux was about 1.5 times higher than typically used for water recycling applications and thus membrane fouling could be accelerated.

Membrane characterisation and analytical techniques

Atomic force microscopy analysis

Membrane surface roughness and surface area were determined using a Bio-XE atomic force microscope (AFM) instrument (Park Systems, Suwon, Korea). The imaging was performed in air using tapping mode with Nanosensor PPP-FMR silicon cantilevers (spring constant of ∼2.8 N/m). The scanning area was 10 × 10 μm and three samples were analysed to obtain the average value.

Contact angle

The hydrophobicity of membranes was evaluated using contact angle measurements. The contact angle of the membrane surface was analysed using a Rame-Hart Goniometer (Model 250, Rame-Hart, Netcong, NJ, USA). Contact angles at 10 different locations were used to obtain the average value.

Positron annihilation lifetime spectroscopy

The free-volume hole-radius of the RO membranes were determined using positron annihilation lifetime spectroscopy (PALS) with a slow positron beam as previously described by Fujioka et al. (2013). The analysis was conducted at the National Institute of Advanced Industrial Science and Technology in Tsukuba, Japan. When positrons are injected into a solid sample, the positrons annihilate with electrons of the solid sample and emit gamma-rays. Before the annihilation, some of the positrons may form ortho-positronium (o-Ps), the spin parallel positron–electron bound state, or para-positronium (p-Ps), the spin antiparallel positron–electron bound state. The lifetime of o-Ps (τo-Ps) can be used for the determination of the free-volume hole-size (r) using the Tao-Eldrup model (Equation (1)) 
formula
1
where r (≤1 nm) is the radius of the free-volume hole approximated as a spherical shape. The analysis was carried out under vacuum at 10−5 Pa. The positron incident energy was set at 1.0 keV to analyse free-volume hole-radii at a mean depth of around 40 nm of the sample. The positron lifetime spectrum of each sample was obtained from the collection of about 2 × 106 positron annihilation events and analysed to deduce τo-Ps by using a nonlinear least-squares fitting programme. The relative measurement uncertainty of τo-Ps was less than 5%.

Size exclusion chromatography analysis

The dissolved organic carbon compositions of the RO permeate were investigated. These samples were characterised with a liquid chromatography-organic carbon detection (LC-OCD) Model 8 system (DOC-LABOR, Karlsruhe, Germany) described previously (Henderson et al. 2010).

RESULTS AND DISCUSSION

Membrane surface properties

Heat treatment did not substantially change the morphology of the ESPA2 membrane. Scanning electron microscope (SEM) images of the untreated and heat-treated membranes revealed no discernible difference in visual appearance (Figure S2 of the Supplementary Material, available online at http://www.iwaponline.com/ws/015/135.pdf). Membrane surface morphology evaluated by AFM analysis, which is a complementary technique to SEM for surface investigations, also revealed no significant changes after heat treatment (Figure S3 of the Supplementary Material, available online at http://www.iwaponline.com/ws/015/135.pdf). Nevertheless, heat treatment caused a slight increase in membrane surface roughness and effective surface area of the ESPA2 membrane from 79 to 101 nm and from 220 to 233 μm2, respectively (Table 1). These values are representative of images taken from three different membrane samples. Thus, the changes in surface roughness reported here can be within the natural variation among different membrane samples. Heat treatment also rendered the surface of the ESPA2 membrane more hydrophilic (contact angle decreased from 48 to 35 °) (Table 1). The mean free-volume hole-radius of the untreated and heat-treated ESPA2 membrane analysed by the PALS were determined to be nearly identical (0.267 and 0.266 nm, respectively), indicating that heat treatment did not have any significant impact on free-volume hole-radii (Table 1). Heat treatment performed on the TFC-HR membrane caused a discernible increase in free-volume hole-radius (from 0.267 and 0.287 nm).

Table 1

Membrane properties of untreated and heat-treated ESPA2 and TFC-HR membranes

CategoryMethodParameterESPA2TFC-HR
UntreatedHeat-treatedUntreatedHeat-treated
Permeability Filtration system Pure water permeabilitya [L/m2hbar] 5.0 ± 0.1 3.2 ± 0.1 3.1 ± 0.1 1.9 ± 0.1 
Topography AFM Surface roughnessa [nm] 79 ± 6 101 ± 11 n.a. n.a. 
 AFM Surface areab [μm2220 ± 3 233 ± 7 n.a. n.a. 
Hydrophobicity Goniometer Contact angle [°] 48 ± 3 35 ± 3 n.a. n.a. 
Free-volume hole-size PALS τo-Ps [ns] 1.83 1.82 1.83 2.04 
 Mean free-volume hole-radius, rp [nm] 0.267 0.266 0.267 0.287 
CategoryMethodParameterESPA2TFC-HR
UntreatedHeat-treatedUntreatedHeat-treated
Permeability Filtration system Pure water permeabilitya [L/m2hbar] 5.0 ± 0.1 3.2 ± 0.1 3.1 ± 0.1 1.9 ± 0.1 
Topography AFM Surface roughnessa [nm] 79 ± 6 101 ± 11 n.a. n.a. 
 AFM Surface areab [μm2220 ± 3 233 ± 7 n.a. n.a. 
Hydrophobicity Goniometer Contact angle [°] 48 ± 3 35 ± 3 n.a. n.a. 
Free-volume hole-size PALS τo-Ps [ns] 1.83 1.82 1.83 2.04 
 Mean free-volume hole-radius, rp [nm] 0.267 0.266 0.267 0.287 

n.a.: data not available.

aDetermined with Milli-Q water at 1,000 kPa and 20 °C feed temperature.

bSurface area was determined by the scanning area of 100 μm2.

Heat treatment caused a substantial decrease in the permeability of the ESPA2 and TFC-HR membranes from 5.0 to 3.2 L/m2hbar and 3.1 to 1.9 L/m2hbar, respectively. In the pore-flow model (Kiso et al. 2011), the permeability of RO membranes can be governed by the physicochemical characteristics of their active skin layer, as described in the following equation: 
formula
2
where Lp is water permeability, rp is effective free-volume hole-radius of the active skin layer, Ak is the free-volume fraction, η is the viscosity of solution in a free-volume hole, and Δx is the length of the free-volume hole through the membrane which is correlated with the active skin layer thickness. Among the physicochemical characteristics of their active skin layer, the mean free-volume hole-radius of the ESPA2 membrane analysed by PALS revealed no discernible changes after heat treatment. Thus, the other properties (i.e., free-volume fraction and thickness) can be the important factors influencing the changes in permeability. However, it was not possible in this study to accurately measure the free-volume fraction and thickness of the membrane active skin layer. There are currently no analytical techniques available to accurately determine the free-volume fraction. In addition, it is not generally possible to measure the effective thickness of the active layer of these membranes due to artefacts being introduced during sample preparation. Very recently, the overall thickness of the active skin layer of RO membranes including cavities and valleys that are present in the layers has been measured by high resolution transmission electron microscopy (Kurihara & Hanakawa 2013). This required special sample preparation that was not readily available and was therefore beyond the scope of the present study. It will be the subject of further investigations.

Resistance to membrane fouling

The impact of heat treatment on membrane fouling was evaluated using the RO feed and three different RO concentrates. A negligible difference in fouling development between the untreated and heat-treated was observed when the RO feed was used (Figure 1(a)). The fouling behaviour differed remarkably when RO concentrates were used as the feed solution (Figures 1(b)1(d)). For example, the permeate flux decline of the untreated membrane using the first stage RO concentrate reached 17% with 15 h filtration, while the heat-treated membrane exhibited only a marginal decrease (6%) (Figure 1(b)). The extended period of filtration (50 h) using the first stage RO concentrate caused a larger difference in permeate flux decline between the untreated membrane (39%) and heat-treated membrane (21%) (data not shown). The heat-treated membrane exhibited a slower flux decline than untreated membranes when the second and third stage RO concentrates were used (Figures 1(c) and 1(d)). Nevertheless, both heat-treated and untreated membranes exhibited a significant drop in permeate flux (over 50%). The heavy fouling observed using the second and third stage RO concentrates was due probably to the high foulant concentration. In fact, total organic carbon (TOC) of the third stage RO concentrate was as high as 50 mg/L (Table S1 of the Supplementary Material, online at http://www.iwaponline.com/ws/015/135.pdf). Overall, the results obtained here indicate that heat treatment could improve fouling resistance for RO concentrates. An increase in hydrophilicity of the RO membrane by heat treatment (Table 1) is one potential cause for the enhanced fouling resistance. Hydrophilic modification of RO membrane surfaces has been reported to reduce the organic fouling of membranes (Zou et al. 2011). A hydrophilised membrane surface can lead to a decrease in the hydrophobic interaction between membrane surface and foulants.

Figure 1

Fouling development on untreated and heat-treated ESPA2 membranes using the (a) RO feed, and (b) first stage, (c) second stage, and (d) third stage RO concentrates (cross-flow velocity 40.2 cm/s, feed temperature 20.0 °C, feed pressure for heat-treated membranes 1,000 kPa and untreated membranes 700 kPa). Each filtration experiment started with approximately 30 L/m2h permeate flux and operated under the constant pressure.

Figure 1

Fouling development on untreated and heat-treated ESPA2 membranes using the (a) RO feed, and (b) first stage, (c) second stage, and (d) third stage RO concentrates (cross-flow velocity 40.2 cm/s, feed temperature 20.0 °C, feed pressure for heat-treated membranes 1,000 kPa and untreated membranes 700 kPa). Each filtration experiment started with approximately 30 L/m2h permeate flux and operated under the constant pressure.

Solute rejection

Boric acid is hydrophilic and most boron exists in the uncharged form of boric acid (B(OH)3) at the tested pH of 7.9 (Tu et al. 2013). Heat treatment led to a considerable increase in the rejection of boric acid under a range of permeate flux (Figure 2). For example, heat treatment applied to the ESPA2 membrane increased boron rejection from 26 to 68% under 20 L/m2h permeate flux (Figure 2(a)). By increasing permeate flux to 40 L/m2h permeate flux, boron rejection by the heat-treated ESPA2 membrane reached as much as 78%. A similar improvement on boron rejection was observed for the TFC-HR membrane (Figure 2(b)). The results reported here indicate that heat treatment is an effective method to improve the rejection of small and neutral solute (i.e. boric acid).

Figure 2

Effect of heat treatment on boron rejection by the: (a) EPSA2, and (b) TFC-HR membranes as a function of permeate flux (20 mM NaCl, 1 mM NaHCO3, 1 mM CaCl2, cross-flow velocity 40.2 cm/s, feed pH 7.9, feed temperature 20.0 °C).

Figure 2

Effect of heat treatment on boron rejection by the: (a) EPSA2, and (b) TFC-HR membranes as a function of permeate flux (20 mM NaCl, 1 mM NaHCO3, 1 mM CaCl2, cross-flow velocity 40.2 cm/s, feed pH 7.9, feed temperature 20.0 °C).

The molecular volume of boric acid (0.071 nm3) is smaller than the free-volume hole-space of the virgin RO membrane (0.080 nm3); thus, the degree of boron rejection can be potentially affected by free-volume hole-radii within the active skin layer of RO membranes. A previous study (Henmi et al. 2010) reported that the rejection of boron increased with decreasing free-volume hole-radius. Nevertheless, no apparent difference in free-volume hole-radius before and after heat treatment was observed in this study (Table 1). On the other hand, a strong correlation between permeability and boron rejection was observed (Figure 3). As described earlier, the free-volume fraction and thickness of the active skin layer are also important factors affecting permeability (Equation (2)). Thus, the improvement of solute rejection by heat treatment may be attributed to these two active skin layer properties.

Figure 3

Boron rejection as a function of pure water permeability by the untreated and heat-treated membranes (permeate flux 20 L/m2h). Results were obtained from Table 1 and Figure 2.

Figure 3

Boron rejection as a function of pure water permeability by the untreated and heat-treated membranes (permeate flux 20 L/m2h). Results were obtained from Table 1 and Figure 2.

The rejection of organic matter by the untreated and heat-treated membranes was also evaluated using the first stage RO concentrate. Permeate samples were collected shortly after the start of the filtration tests. As a result, TOC rejection by the heat-treated membrane (99.64%) was equivalent to that by the untreated membrane (99.60%). Further investigation was carried out by analysing organic size fractions in the permeate solutions using LC-OCD analysis. Overall, there was negligible difference between organic size fractions detected in the permeate of untreated and heat-treated membranes (Figure 4) and therefore ratios of each organic fraction (e.g. building blocks, low-molecular-weight (LMW) neutrals, and LMW acids) present in the respective permeates were almost equivalent (Table S2 of the Supplementary Material, online at http://www.iwaponline.com/ws/015/135.pdf). Nevertheless, a peak of LMW neutrals observed in the permeate of the untreated membrane (between 85 and 95 min retention time) was not detected in the permeate of the heat-treated membrane (Figure 4). The results indicate that the heat-treated membrane can enhance the rejection of small organic matter in comparison to the untreated membrane.

Figure 4

LC-OCD chromatograms of the RO permeate treated by the: (a) untreated, and (b) heat-treated ESPA2 membranes. OCD and UVD represent organic carbon detection and UV detection at 254 nm, respectively.

Figure 4

LC-OCD chromatograms of the RO permeate treated by the: (a) untreated, and (b) heat-treated ESPA2 membranes. OCD and UVD represent organic carbon detection and UV detection at 254 nm, respectively.

CONCLUSIONS

Modification of the RO membranes through heat treatment resulted in an improvement on fouling resistance and boron separation performance. However, heat treatment reduced the pure water permeability of the RO membranes. The improved fouling resistance by heat treatment was observed for different RO concentrate solutions (i.e. first, second and third stage RO concentrates). The improved fouling resistance of the heat-treated membrane was due possibly to the decreased hydrophobic interaction between the modified RO membrane surface and foulants. Heat treatment improved boron rejection significantly. Heat treatment resulted in an increase in boron rejection by the ESPA2 membrane from 26% (without heat treatment) to 68% (after heat treatment). PALS revealed that heat treatment did not significantly influence the free-volume hole-radius of the membrane active skin layer. The study suggested that other active skin layer properties such as the free-volume fraction and thickness may be the critical factors in improving the rejection of boron. Further research focusing on membrane characterisation of the untreated and heat-treated membranes is necessary to clarify the mechanism of the improved effects of heat treatment.

ACKNOWLEDGEMENTS

Hydranautics is thanked for the provision of membrane samples. The authors acknowledge Mr Xiang Li for his assistance with LC-OCD analysis.

REFERENCES

REFERENCES
Fujioka
T.
Nghiem
L. D.
2013
Modification of a polyamide reverse osmosis membrane by heat treatment for an enhanced fouling resistance
.
Water Science & Technology: Water Supply
13
(
6
),
1553
1559
.
Fujioka
T.
Oshima
N.
Suzuki
R.
Khan
S. J.
Roux
A.
Poussade
Y.
Drewes
J. E.
Nghiem
L. D.
2013
Rejection of small and uncharged chemicals of emerging concern by reverse osmosis membranes: the role of free volume space within the active skin layer
.
Separation and Purification Technology
116
,
426
432
.
Henderson
R. K.
Stuetz
R. M.
Khan
S. J.
2010
Demonstrating ultra-filtration and reverse osmosis performance using size exclusion chromatography
.
Water Science & Technology
62
(
12
),
2747
2753
.
Henmi
M.
Fusaoka
Y.
Tomioka
H.
Kurihara
M.
2010
High performance RO membranes for desalination and wastewater reclamation and their operation results
.
Water Science & Technology
62
(
9
),
2134
2140
.
Kiso
Y.
Muroshige
K.
Oguchi
T.
Hirose
M.
Ohara
T.
Shintani
T.
2011
Pore radius estimation based on organic solute molecular shape and effects of pressure on pore radius for a reverse osmosis membrane
.
Journal of Membrane Science
369
(
1–2
),
290
298
.
Shannon
M. A.
Bohn
P. W.
Elimelech
M.
Georgiadis
J. G.
Marinas
B. J.
Mayes
A. M.
2008
Science and technology for water purification in the coming decades
.
Nature
452
(
7185
),
301
310
.
Tu
K. L.
Nghiem
L. D.
Chivas
A. R.
2010
Boron removal by reverse osmosis membranes in seawater desalination applications
.
Separation and Purification Technology
75
(
2
),
87
101
.
Tu
K. L.
Fujioka
T.
Khan
S. J.
Poussade
Y.
Roux
A.
Drewes
J. E.
Chivas
A. R.
Nghiem
L. D.
2013
Boron as a surrogate for N-nitrosodimethylamine rejection by reverse osmosis membranes in potable water reuse applications
.
Environmental Science & Technology
47
(
12
),
6425
6430
.
Zou
L.
Vidalis
I.
Steele
D.
Michelmore
A.
Low
S. P.
Verberk
J. Q. J. C.
2011
Surface hydrophilic modification of RO membranes by plasma polymerization for low organic fouling
.
Journal of Membrane Science
369
(
1–2
),
420
428
.

Supplementary data