An autopsy of spiral wound reverse osmosis (RO) membrane operated in brackish water treatment was conducted to understand the origin and extent of foulants and fouling mechanisms. Structural and chemical characterization was determined by visual inspection and instrumental analysis such as scanning electron microscopy/energy-dispersive X-ray spectroscopy (SEM-EDS) and X-ray diffraction (XRD). It was observed that the membrane surfaces were completely covered with a gray/brown pollutant layer in all membrane sheets. SEM images proved accumulation of mineral pollutants on membrane surface. Also, high levels of Al and Si, which were attributed to aluminum silicates originating from feed water, were determined on membrane surfaces. Additionally, the XRD analysis results showed that the foulant sample collected from membrane surfaces included halloysite, SiO2 and LiCl components. Fujiwara's result proved that no damage occurred on the membrane surface due to oxidation. Consequently, a fouling control strategy for RO-based brackish water treatment plants was also recommended to increase the membrane life.
An autopsy of spiral wound RO membrane operated in brackish water treatment was conducted to understand the fouling behaviors.
Various instrumental and analytical analyses were performed to characterize the organic and inorganic foulant on membrane surface.
A fouling control strategy for RO-based brackish water treatment plans was also recommended to increase the membrane life.
The major processes applied for desalination and water reuse are thermal and membrane-based technologies. Although the reverse osmosis (RO) is known as an energy-and cost-effective process, it is faced with major obstacles. The most important obstacle for membrane processes is fouling, which reduces the membrane performance and product quality. Membrane fouling also causes an increase in operating costs due to the more frequent cleaning, shorter membrane lifespan and loss in permeate flux (Sweity et al. 2014).
Commonly, membrane fouling occurs through two mechanisms. The first one takes place in membrane pores, while the second occurs via accumulation of fouling, such as various impurities, i.e., suspended inorganic or organic materials, on the membrane surface. Forms of membrane fouling are generally classified according to the characteristics of foulants present in the feed stream, which may consist of biological fouling, colloidal fouling, scaling fouling, and organic fouling. Therefore, fouling must be well managed to prevent deterioration in membrane performance and prolong the membrane lifespan. Selection of configuration and membrane material, operation conditions, pretreatment, regularity of membrane cleaning and maintenance are some important parameters to achieve successful membrane operation (Liu et al. 2019).
Despite the progress in membrane processes, the previous experiences of plants operating membrane technologies for water and wastewater treatment have shown that membrane fouling remains the main problem for the effective operation of membrane processes (Jacquemet et al. 2006; Van Agtmaal et al. 2007; Hoek et al. 2008; Karime et al. 2008; Yang et al. 2008). Despite various research efforts, to date the characterization of fouling of RO membranes has not progressed significantly (Tran et al. 2007). Many studies based on laboratory experiments were conducted to understand the contribution of some important factors to membrane fouling, including the feed water characteristics, concentration of some important components, temperature, membrane properties and hydrodynamic conditions (Lee & Elimelech 2006; Lee et al. 2006; Ang & Elimelech 2007; Boussu et al. 2007, 2008; Fonseca et al. 2007; Tang et al. 2007). Although these laboratory-based studies have contributed to our knowledge about membrane fouling, they are limited in fully representing the hydrodynamic conditions of real-scale applications. It is well known that in real-scale membrane applications, membrane performance and fouling potential are strongly affected by variations in water quality and temperature, pre-treatment efficiency and operating conditions (Xu et al. 2010).
The most effective application for understanding the membrane fouling and operational problems is to conduct a membrane autopsy. Various analyses help to explain the morphology of a fouled membrane surface and the identification and the distribution of organic/inorganic and biological components onto the membrane surface. While the membrane autopsies provide accurate information about the origin and the extent of foulants and fouling mechanisms, more efficient membrane operation could be achieved by considering the membrane autopsy results (Kim et al. 2015).
Membrane autopsy has gained importance to be the most efficient method in understanding the fouling characterization of membranes used in treatment systems. Studies on fouled membrane autopsies have been performed to evaluate the type and level of fouling (Vrouwenvelder et al. 2003; Darton et al. 2004; Al-Amoudi & Lovitt 2007; De Roever & Huisman 2007). Since real-scale membrane applications generally consist of sea water (Pontié et al. 2005; Van Agtmaal et al. 2007), brackish water (Boubakri & Bouguecha 2008; Karime et al. 2008; Yang et al. 2008) and surface water treatment (Speth et al. 1998), membrane autopsy studies have focused on these issues.
Ruiz-García et al. (2018) carried out an autopsy of fouled brackish water reverse osmosis (BWRO) membrane elements which were taken over 11 years of operation from a full-scale desalination plant. The plant had a capacity of 15 m3/day with a recovery of 60% and a pretreatment step consisting of cartridge filters and antiscalant dosing prior to membrane process. The results indicated that inorganic fouling was mainly formed from calcium carbonate and aluminosilicates, and operation under appropriate conditions and correct pretreatment increased the membrane lifespan. In another study, Kim et al. (2015) performed a membrane autopsy of three fouled spiral wound RO membranes from different pressure vessels to identify the origin and extent of membrane fouling. The results showed that fulvic acid components from feed water were the main contributor to fouling. Additionally, a small amount of inorganic fouling due to Fe and Al was observed, especially in the first-stage RO vessel. Also, the accumulation of organic and inorganic foulants was greater on the back side of the membrane sheet than on the front side. Karime et al. (2008) conducted a spiral wound BWRO membrane autopsy with 6 years' operation in a desalination plant. Feed water had a total dissolved solids (TDS) value of 6,000 mg/L and water recovery ratio was 75%. The desalination plant consisted of sand and cartridge filter, acid and antiscalant dosing as pretreatment prior to the two-stage RO membranes. Autopsy results stated that main foulants were polysaccharide, clay, SiO2, CaSO4, CaSiO3, AlPO4 and Fe3O4. Although considerable efforts have been made to understand the basics of membrane fouling, the knowledge about membrane fouling for the efficient operation of real-scale membrane plants is still limited. Another full-scale BWRO membrane autopsy with 1 year's operation was performed by Tran et al. (2007). Before the membrane process, a pretreatment procedure consisting of coagulation-flocculation, dissolved air flotation, filtration and cartridge filter was applied. Polysaccharides, organic-Al-P complexes, and aluminum silicates were found as a main foulant contributor.
In this study, an autopsy of spiral wound RO membrane operated in brackish water treatment was conducted to understand the fouling behaviors. Various analyses were performed to characterize the organic and inorganic foulants on membrane surface. In addition, a fouling control strategy for RO-based brackish water treatment plans was also recommended to increase the membrane life.
MATERIALS AND METHODS
The full-scale brackish water treatment plant in Turkey had been in operation to produce process water for a company using raw groundwater. The wastewater had a pH of 7.6, a TDS of 506 mg/L, a conductivity of 607 μS/cm, a bicarbonate of 295 mg/L, a calcium of 63 mg/L, a magnesium of 27 mg/L, a sodium of 60 mg/L, a chloride of 77 mg/L. This full-scale plant was composed of cartridge filter and RO membranes. The feed water taken from the well nearby the plant was pretreated with cartridge filter. After cartridge filter, the water was treated with spiral wound RO membrane for water recovery. The spiral wound RO membrane used in this study was 7.9 inches BW30-400 (DOW, FilmTec, USA). It was characterized with an active area of 37 m2, spacer thickness of 28 mil, permeate flow rate of 40 m3/d, and minimum salt rejection of 99% (conditions: pH 8, 15.5 bar, 2,000 ppm NaCl, 25 °C and 15% recovery) by the manufacturer. The fouled RO membrane with an approximately 1-year operation was collected from the RO module on full-scale plant.
Preparation and analysis of foulant sample on membrane surface
The fiberglass protective cover of the membrane module was cut to take out the membrane sheets. After the protective cover separated from the membrane, the membrane was opened and examined sheet by sheet. The gray/brown colored fouling layers on the membrane surface were collected by a spatula spoon for the XRD analysis. The moisture content of the sample was removed by drying in an oven at 105 °C to prepare for XRD analysis. Additionally, the wastewater sample was supplied and prepared for XRD analysis. Also, the collected foulant samples from membrane surface were extracted with hexane to measure the oil content. XRD measurements were performed using the A Bruker D8 (Cu Kα at 40 kV) X-ray diffractometer, in the 2θ range of 10–100° with the scanning rate of 1°/min and 0.02° step size, using Cu Kα radiation to determine the crystallographic structure of foulant material.
Membrane surface morphology and elemental composition analysis
The inorganic elemental composition of the foulants on the membrane surfaces was determined by scanning electron microscopy (SEM, Philips XL 30S FEG) coupled with energy dispersion spectrometry (SEM-EDS). Secondary electron (SE) images were also taken at different magnification from the SEM images. Two membrane samples were cut from different highly contaminated locations on the membrane sheet and EDS spectrum was taken. Prior to SEM/EDS analysis, membrane samples were dried in an oven to remove moisture content.
Oxidation is the one of the major problems that causes an irreversible degradation of the RO active layer, which leads to membrane replacement. Fujiwara test is performed to check whether the membranes are affected by chlorine exposure. Fujiwara test is based on a spectrophotometric measurement which detects the halogenated organic materials with absorbance at 530 nm. In this study, membrane samples were cut at different locations and placed in the bottles. Sodium hydroxide and pyridine were added to membrane samples and there was a one-minute wait for the reaction at 90 °C. Pink color was accepted as a positive result while no color change was considered a negative result.
RESULTS AND DISCUSSION
Evaluation of visual inspection of RO module and membrane
Firstly, the physical examination of the spiral wound RO module brought to the laboratory was performed, and no physical damage was observed in the feed and permeate channels. However, when the outer protective shell of the module consisting of fiberglass was examined, it was determined that some part of the protective shell was cracked. The fiberglass membrane cracks in the outer shell have several causes. When the membrane system is stopped, the air remaining in the membrane or pressure cover causes cracking of the fiberglass material due to the sudden rise in feed water pressure while the system is restarted. Therefore, the feed water pressure should be increased gradually to prevent the formation of cracks. Another reason for damage to the protective fiberglass cover is the high pressure drop caused by membrane contamination. Thus, the pressure difference between feed and concentrate stream above the limit values should be avoided. The images of RO module, feed and permeate stream channel, and the crack detected on the fiberglass cover of the module, are given in Figure 1.
The fiberglass protective cover of the membrane module was cut with a spiral saw to take out the membrane sheets. After the protective cover separated from the membrane, the membrane was opened on a flat clean surface and examined sheet by sheet.
An examination was carried out on the membrane surfaces and places of bonding to each other, spacers, and sealing gasket, and no physical damage was detected in these parts. The images of membrane sheets and membrane surfaces are given in Figure 2. It was observed that the membrane surface was completely covered with a gray/brown pollutant layer in all membrane sheets. The fouling may be attributed to clay and silica derivatives, which cause membrane contamination even at low concentrations.
Preparation and XRD analysis of foulant sample on membrane surface
The gray/brown colored fouling layer on the membrane surface was collected by a spatula spoon for the analysis. The moisture content of the sample was removed by drying in an oven at 105 °C to prepare for XRD analysis. Additionally, the wastewater sample was supplied and prepared for XRD analysis. The images of the pollutant samples collected on the membrane surface and wastewater samples before and after drying in the oven are shown in Figure 3. Additionally, the collected foulant samples from membrane surface were extracted with hexane to measure the oil content. No oil content was detected in samples according to the oil analysis results.
The XRD analysis result of foulant collected from membrane surface is given in Figure 4(a) and XRD analysis of the wastewater sample is given in Figure 4(b). When the XRD profile of the wastewater sample was examined, it was observed that it contained especially CaCO3 and CaCl2 compounds. Since the autopsy membrane was used for brackish water treatment, the XRD results of the wastewater have been attributed to brackish water properties. Karmal et al. (2020) carried out an XRD analysis with the aim of determining the crystal structure of membrane foulants. They reported the diffraction peaks of CaCO3 very similarly to our study. It is well known that brackish water is characterized with Ca(HCO3)2, CaCl2 and NaCl types according to its hydrogeochemical properties. Water moving through the ground reacts to varying degrees with the surrounding minerals, and it is these rock-water interactions that give the water its characteristic chemistry. Groundwater can dissolve the calcite rocks, and the calcite solubility is controlled by the amount of carbon dioxide available. Its solubility increases due to the formation of more soluble calcium bicarbonate in brackish water saturated with carbon dioxide. Besides, the XRD analysis results showed that the foulant sample collected from the membrane surface includes halloysite, SiO2 and LiCl components. Halloysite is an aluminosilicate clay mineral with the empirical formula Al2Si2O5(OH)4. Its main constituents are oxygen (55.78%), silicon (21.76%), aluminum (20.90%), and hydrogen (1.56%). Halloysite typically forms by hydrothermal alteration of alumina-silicate minerals, and it is often found near carbonate rocks. The reason why carbonate derivatives are not seen in the foulant sample is that these derivatives were removed from the membrane surface with the acidic chemical cleaning of the membrane during the operation. Similarly, Ruiz-García et al. (2018) carried out a RO membrane autopsy to determine the fouling behaviors related to feed water inorganic composition and operating conditions. They reported that the main inorganic foulants were calcium carbonate and aluminosilicates, which were attributed to brackish water properties. In another BWRO membrane autopsy study, higher amounts of Al, Ca, P and lesser amounts of Fe, S, Mg, K and Na were reported as inorganic fouling composition. They also stated that their usage of aluminum sulfate as a coagulant and phosphonate-based antiscalant could contribute to the high levels of Al and P (Tran et al. 2007).
The SEM/EDS analysis is performed to confirm the difference in the extent of fouling across the membrane surface as observed by optical microscopy, and it gives detailed information on the nature of the fouling layer. SEM images obtained from membrane surface are given in Figure 5. A typical fouled membrane surface comprising scale layer formed in an apparently crystalline morphology can be seen in Figure 5. Closer SEM images proved accumulation of mineral pollutants on the membrane surface. Covering the membrane surface completely with a fouling layer results in decreasing permeate flux and causes the physical membrane damage with an increasing pressure difference in the system.
The EDS analyses (Figure 6) show that the fouled membrane surface had relatively high levels of Aluminum (Al), Silicon (Si), Oxygen (O), Sulfur (S) and Carbon (C) elements and quite low levels of Ca, Mg, Cl, Na, Fe and K. The S and C peaks are likely due to the membrane structure. The high levels of Al and Si on the membrane surface were mainly attributed to aluminum silicates, which are commonly observed as RO foulants, especially during filtration of brackish water. Researchers who studied the presence of silica in the membrane deposit reported that silica scaling has a relationship with the presence of trivalent cations such as aluminum and iron. Furthermore, the Si constitutes a major portion of clay, quartz, and sand (Farhat et al. 2018). It is known that the composition of clay minerals mainly consists of hydrous aluminum silicates. Here, SiO2 and Al2O3 tetrahedrons and octahedrons combine in various ways to form layers and form clay minerals with various properties. It is thought that Al and Si minerals originate from clay structures. The element Si can also originate from quartz minerals, which have SiO2 in their structure. Although a cartridge filter is used as pretreatment, only large-sized silt/clay particles are retained by the cartridge filter, while finer particles might remain in the feed and cause a fouling layer on the membrane surface (Tran et al. 2007). When the XRD and SEM-EDS results are compared with each other, it is observed that the Al and Si minerals detected on the membrane surface originate from clay and quartz. Consequently, possible quartz and clay minerals present in the feed water were considered as the major factor contributing to the high levels of Si and Al on the membrane surface.
The Fujiwara test is performed to determine whether the polymer network in the aromatic polyamide structure of reverse osmosis and nanofiltration membranes is damaged by oxidizing halogenic components such as chlorine, bromine, or iodine. The analysis is based on the reaction of halogens with pyridine in a strongly alkaline environment, and a pink color appears if the membrane is damaged by oxidative components. For this purpose, membrane samples were taken from many different locations on the membrane and a Fujiwara test was performed. As is shown in Figure 7, no color change was observed in the samples during the Fujiwara test. Thus, this result proved that no damage occurred on the membrane surface due to oxidation.
The autopsy of fouled spiral wound RO membrane was performed to determine the possible sources of reduction in flux and membrane fouling. Physical damage was observed on the fiberglass protective cover, and this was attributed to the sudden rise in feed water pressure while the system is restarted. Therefore, it was recommended to increase the pressure gradually while restarting the membrane plant. Additionally, a high amount of gray/brown colored foulant layer, which completely covered the membrane surface, was detected on the membrane surface. High pressure drops may occur due to the membrane fouling, and both membrane protective cover and membrane may be damaged. Thus, it was recommended not to exceed specified limit values for pressure drops at the feed and concentrate stream of the membrane process. As a result of detailed SEM/EDS and XRD analysis, high levels of Si and Al were detected on the membrane surface. Considering both membrane fouling layer and wastewater characteristics, it was concluded that an effective pretreatment procedure consisting of coagulation/flocculation and filtration must be applied before the RO membrane process. Additionally, the amount of silicon should be measured in the feed water before feeding the membrane process. Also, it was recommended to use the correct type of antiscalant in the membrane filtration system to prevent carbonate precipitation. Therefore, a jar test should be applied to determine the appropriate pretreatment process and coagulant/flocculant type and dosages to prolong the membrane performance and lifespan.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.