Autopsy and chemical cleaning of a reverse osmosis membrane from a pilot-scale brackish water desalination system Open Access

Ongoing research and development in brackish water and seawater desalination technologies aim to tackle water scarcity in various regions worldwide. Among various desalination technologies, reverse osmosis (RO) processes currently dominate the others, occupying over 60% of the global desalination installations because of their benefits concerning simple design and operation and low energy demand (Aryanti et al. 2023; Zolghadr-Asli et al. 2023). However, RO membranes are highly susceptible to impurities (i.e., inorganic, organic, and biological compounds) in water and can be easily fouled during operation (Ahmed et al. 2023). Inorganic fouling occurred when dissolved salt ions (e.g., Ca²⁺, Mg²⁺, ⁠, ⁠, and ⁠) precipitated and accumulated on the membrane surface once their concentration reached their saturated solubility, resulting in the phenomenon of inorganic fouling (Matin et al. 2021). Organic fouling resulted from the interaction between macromolecular matters (e.g., humic and fulvic acids, proteins, and polysaccharides) and the RO polyamide membrane through Van de Waals and electrostatic forces (Goh et al. 2018). The attachment of various microorganisms in the influent on the RO membrane and their growth by means of the uptake of organic nutrients caused the formation of biofouling (Suresh et al. 2023). Membrane fouling gradually increased the hydraulic resistance for water passage, requiring more energy to maintain the constant permeate flux. In addition, the occurrence of foulants on the RO membranes decreased their longevity and performance, consequently accelerating the membrane replacement (Nthunya et al. 2022). Therefore, fouling processes impaired the productive efficiency of RO systems, resulting in an increase in the cost of water treatment.

Pretreatment processes were frequently installed before the RO system to upgrade feed water quality and alleviate membrane fouling. In addition, cleaning processes, especially chemical methods, were employed when the RO elements were polluted to remove foulants on the RO membrane and recover its performance (Abdel-Karim et al. 2021; Abushawish et al. 2023). Understanding the nature and characteristics of impurities plays a decisive role in designing the pretreatment configuration as well as selecting appropriate chemicals for cleaning. Autopsies of used membranes can disclose reliable information on the composition of fouling layers, which is beneficial to suggest strategies for preventing and removing the buildup on the membrane (Adel et al. 2022; Wang et al. 2023). For example, inorganic minerals, including halloysite, silica, and lithium chloride, were determined as the main foulants of brackish water reverse osmosis (BWRO) membrane operated for roughly a year (Balcik 2021). In addition, interactions between inorganic compounds of Fe and Al and fulvic acid were responsible for membrane fouling of a BWRO desalination plant producing drinking water for approximately five years (Kim et al. 2015). Furthermore, an autopsy of RO membranes after 11 years of operation in a BWRO desalination plant revealed that inorganic compounds (i.e., calcium carbonate and aluminosilicates) and biofilm of diatoms primarily contributed to membrane fouling (Ruiz-García et al. 2018). It was obvious that the reasons for membrane fouling varied depending on the sources of feed water. Besides, there have been very few studies on the situation of the RO module after a period of operation without pretreatment installation. Generally, the absence of pretreatment in RO systems led to a decline in permeate flow caused by fouling, which in turn reduced the operational lifespan of the RO membranes (Shaheen & Cséfalvay 2024). Nevertheless, these studies were primarily conducted under laboratory conditions with simulated feedwater (Yang et al. 2019; Sun et al. 2024), leading to discrepancies when compared to field studies due to the lack of real-world environmental variables. Additionally, the failure to perform a membrane autopsy has restricted the acquisition of detailed physicochemical data regarding the fouling constituents (Cornelissen et al. 2021; Costa et al. 2024).

In this regard, we ran the BWRO system at a pilot scale of 20 m³/day in the absence of pretreatment processes in a batch mode, with an influent from the Luong Quoi River. The feed pressure and permeate flux values were monitored to assess the fouling time of the RO membrane. The fouled RO membrane was disconnected from the system, and the fouling samples were obtained for further analysis. The morphologies and composition of foulants were characterized using various analytical techniques, including scanning electron microscopy (SEM), X-ray diffraction (XRD), X-ray fluorescence (XRF), energy-dispersive X-ray (EDX), and Fourier-transform infrared (FT-IR). In addition, the efficiency of membrane cleaning using commercial chemicals of Hydrex 4506 and 4703 was investigated and discussed.

The BWRO system at pilot scale with a capacity of 20 m³/day was located at Luong Quoi water plant, Giong Trom District, Ben Tre Province (i.e., latitude: 10.2020951, longitude: 106.4776562). The feed water was drawn from a container (i.e., containing water from Luong Quoi River) by a centrifugal pump (CM 3-5 A-R-A-E-AVBE F-A-A-N, Grundfos), and delivered to an RO module via a high-pressure pump (CR 3-29 A-FGJ-A-E-HQQE, Grundfos). The characteristics of the original RO membrane are supplied in Table 1. The RO system was operated in the absence of pretreatment processes to gather data on membrane fouling composition and fouling time, facilitating the selection of suitable pretreatment technologies. The pressure and permeate flow rate of the RO system were monitored using a pressure transmitter (A-10, Wika) and an electromagnetic water meter (CZ-EFM, Sence).

The phase structure of impurities on the RO membrane surface was determined using an X-ray diffractometer (Empyrean, PANalytical). XRD spectra of samples were recorded in the range of 2θ = 5 − 80° with scanning speed = 1.6 °/min at an accelerating voltage of 45 kV and an applied current of 40 mA using a Cu anode material as an X-ray source with a Kα wavelength of 1.54 Å. The morphology and the elemental analysis of the fouled RO membrane were performed using a scanning electron microscope (SEM) (S4800, Hitachi) coupled with an EDX (H-7593, HORIBA). The analysis was carried out under an accelerating voltage of 10 kV, an emission current of 9,000 nA, and a working distance of 8 mm. The membrane samples were analyzed without any pretreatment or surface coating and were affixed to the specimen holder using conductive double-sided carbon adhesive tape. XRF (ZETIUM, PANalytical) analysis was employed to investigate the chemical composition of the dried foulant. Organic functional groups of fouling samples and RO membranes were detected using FT-IR spectroscopy (NICOLET 6700, Thermo). Contact angles of RO membranes were measured using optical contact angle goniometers (OCA50, Dataphysics). In addition, the presence of several microorganisms, including Escherichia coli and coliform, in the liquid sludge samples was examined by the liquid-medium culture technique based on the guidelines of International Organization for Standardization ISO 7251:2005 and ISO 4832:2007, respectively (ISO 2005, 2006).

Chemical cleaning processes were carried out to remove various foulants from the contaminated membrane surface using Hydrex 4703 (Veolia) and Hydrex 4506 (Veolia) as acidic and alkaline cleaning agents, respectively. First, 50 L of cleaning solutions (i.e., Hydrex 4703 3% or Hydrex 4506 3% or clean water produced from the RO system) was prepared in a 100-L polyethylene tank. The solution pH was adjusted to the desired values using a 20% solution of citric acid (Xilong). The solution was circulated in the system between the tank and the RO unit for a predetermined time by a centrifugal pump (CM 10-1 A-R-A-E-AVBE F-A-A-N, Grundfos). Membrane cleaning procedures included three steps: first, rinse the RO membrane at acidic conditions for 60 min using Hydrex 4703 3%; second, rinse the RO membrane at basic conditions for 60 min using Hydrex 4506 3%; final, rinse the RO membrane using clean water until the pH of the solution reaches around 7. The efficiency of the cleaning process was evaluated based on a recovery in permeate flux and feed pressure.

In addition to the EDX results, an XRF analysis was also carried out to provide a more comprehensive assessment of the elemental composition in the dried fouling sample (Table 2). It should be noted that light elements such as C, N, and O cannot be found using the XRF technique. The other elements present in the EDX spectra were also detected by the XRF and substantially contributed to the composition of the fouling sample. The detection of P, S, Cl, Ca, and Na elements indicates that inorganic ions frequently found in water, such as ⁠, ⁠, Cl⁻, Ca²⁺, and Na⁺, were precipitated on the membrane surface. In addition, a small amount of transition metals, such as Ti, Pb, Co, Ni, Zr, Cr, and Zn, were also present in the dried fouling samples.

A comparative analysis was conducted to examine the primary fouling constituents observed on RO membranes used in the brackish water treatment system at the Luong Quoi Water Plant, located in Giong Trom, against those identified in other RO systems treating both brackish and seawater sources (Table 3). The findings indicate that the predominant foulants are generally consistent across systems and include primary inorganic deposits based on Si, Al, and Fe elements, as well as organic and biological substances such as humic acids, proteins, and polysaccharides. Despite this apparent similarity, it is essential to note that such comparisons serve only as general references due to substantial differences among the systems. Variations in feed water characteristics, operational scale, membrane types, and pretreatment processes can significantly influence the composition and severity of membrane fouling. Therefore, site-specific evaluations remain critical to accurately identifying fouling mechanisms and implementing effective mitigation strategies.

Table 4 presents a comparative summary of the effectiveness of various chemical cleaning protocols for restoring membrane permeate flux, as reported in previous studies, in relation to the findings of the present investigation. Typically, membrane cleaning procedures involve the sequential or combined application of acidic and alkaline agents. Moreover, the inclusion of chelating agents, such as EDTA and citric acid or surfactants, such as sodium dodecyl sulfate (SDS) and SLS, can enhance cleaning efficacy. In scenarios involving severe biofouling or organic contamination, oxidizing agents like chlorine or hydrogen peroxide may be employed to degrade fouling compounds. However, due to their potentially deleterious effects on membrane integrity, the use of such oxidants should be carefully controlled. The combination of the SLS surfactant, EDTA as a chelating agent, and an alkaline cleaning solution (Hydrex 4506), along with citric acid (Hydrex 4703), had demonstrated a high degree of cleaning efficiency across a broad range of foulants. Consequently, this integrated cleaning approach achieved near-complete restoration of membrane permeate flux and is particularly suitable for treating membranes fouled by complex mixtures of inorganic, organic, and biological substances.

Based on the comprehensive analysis of fouled RO membranes and associated solid and liquid sludge samples using a variety of analytical techniques including SEM-EDX, XRD, XRF, FT-IR, as well as assessments of organic and microbiological indicators, the principal foulants were identified. These included inorganic mineral scales such as Quartz, Muscovite-2M1, Kaolinite, Dickite-2M1, among others; organic constituents such as polysaccharides and proteins; and biological contaminants including microorganisms like coliforms. It is evident that in the absence of a pretreatment system, the fouling of RO membranes occurs more rapidly and severely. Although chemical cleaning methods were effective in restoring membrane performance, they might compromise the physical and chemical integrity of the membrane surface, accelerate subsequent fouling, and increase operational costs due to chemical and energy consumption.

Accordingly, the implementation of an appropriate pretreatment system is essential for enhancing feed water quality prior to RO membrane filtration. In alignment with recommended design principles for RO desalination systems (Henthorne & Boysen 2015; Kavitha et al. 2019; Abushawish et al. 2023; Pesarakloo et al. 2024), a pretreatment process tailored to the pilot-scale RO unit utilizing Luong Quoi River water is proposed. The feed water will first undergo coagulation and flocculation, followed by storage in a 10 m³ tank. Subsequently, the water will pass through a dual-media filter (comprising quartz sand and gravel), a triple-media filter (consisting of greensand, quartz sand, and gravel), and finally a cartridge filtration unit equipped with a filter element having a pore size of 5 μm.

The coagulation–flocculation stage is particularly effective in removing a wide range of suspended solids, organic matter, and microbial contaminants. Greensand media exhibit strong adsorption capacity for heavy metals, while the presence of oxidizing agents (e.g., dissolved oxygen) facilitates the degradation of organic and microbial foulants. The integrated use of flocculation and multi-layer filtration has demonstrated high efficiency in mitigating common fouling agents. Furthermore, the final-stage cartridge filter acts as a physical barrier to particles larger than 5 μm, reducing the particulate load on the RO membrane.

To further improve the removal of microbial and inorganic foulants, a chlorine dosing system can be integrated for disinfection purposes. However, dechlorination using sodium bisulfite is required prior to membrane contact to prevent oxidative damage. Reducing the pore size of the cartridge filter to values such as 0.2 μm could enhance fine particle removal, albeit at the expense of increased system pressure. This would necessitate higher-capacity feed pumps and reinforced piping to accommodate the pressure demands. Therefore, a careful evaluation of the trade-offs between treatment efficiency and economic feasibility is crucial in optimizing the pretreatment strategy.

The BWRO system at Luong Quoi water plan was performed in a batch mode without pretreatment processes. The feed pressure increased to 10% of the initial value, and the permeate flux decreased to 45% after 142 h of operation. The visual and scanning electron microscopic observation on the RO module revealed serious foulants on the RO membrane, but no physical damage on the other parts. The analytical results from XRD, XRF, EDX, and FT-IR indicated the contribution of inorganic minerals, organic substances, and microorganisms to the fouling formation on the RO membrane, which can provide valuable data for pretreatment configuration design. A combination of Hydrex 4703 and 4506 chemicals in the chemical cleaning method efficiently removed the foulants on the RO membrane, which can be applied in practice. However, the repeated use of membrane cleaning chemicals can gradually degrade the hydrophilic functional groups on the membrane surface, leading to an increased rate of re-fouling. Consequently, it is essential to carefully assess the frequency of membrane cleaning and determine the optimal membrane replacement schedule to ensure economic efficiency. Furthermore, exploring the integration of physical cleaning techniques into membrane cleaning protocols may help mitigate the adverse effects of chemical agents on the properties of RO membranes. The scalability of the cleaning approach for industrial RO systems requires adjustments in chemical dosing strategies and hydraulic configurations. Additionally, incorporating the cleaning protocol into automated frameworks may enhance operational reliability and reduce labor-intensive maintenance.

We would like to express our sincere gratitude to the Ministry of Science and Technology, the Department of Science and Technology of Ben Tre, and Nguyen Tat Thanh University for their support in terms of time, material resources, and experimental laboratory facilities for this research.

This research is funded by the project ‘Study on the construction of a saline water treatment system with IoT application to supply domestic water for people in saltwater intrusion areas of Ben Tre province and some neighboring provinces,’ with project code ĐTĐL.CN–123/21.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.