Application of the multi-wavelength UV-LED/chlorine process to improve reverse osmosis membrane performance for reused water treatment in the steel industry Open Access

China has a large steel production, which accounts for nearly 50% of the total global yield of steel in the world (Chu et al. 2021). During steel-making, a great amount of water is required to cool the system and wash the steel. Thus, the steel industry is one of the biggest water-intensive sectors and also wastewater dischargers. Treating and reusing industrial steel wastewater is an important way to enhance water resource utilization and achieve water conservation (Ma et al. 2020). With the widespread application of advanced wastewater treatment methods such as membrane separation, the efficiency of water reuse has been greatly improved (Chen et al. 2023). The reverse osmosis (RO) technology, with its good effluent quality, small footprint, and ease of operation, has become a core process for industrial reused water treatment and resource utilization (Chen et al. 2024). However, the prominent issue of membrane fouling is still the main obstacle to limiting its further application (Pourbozorg et al. 2017; Li et al. 2020). Membrane fouling is a complex process among the influent, the membrane, and the associated dynamic environment. Membrane fouling is the adsorption, deposition, or blockage of solute molecules, colloidal particles, or biological substances from the feedwater onto the membrane surface or into the membrane pores through physical or chemical interactions (Park et al. 2020; Liu et al. 2021). This generally leads to an increase in transmembrane pressure or a decrease in permeate flux during the filtration process, reducing the membrane's lifespan and increasing the operation costs (Pourbozorg et al. 2016; Wang et al. 2022a). Therefore, various strategies have been implemented to reduce the RO membrane fouling for industrial reused water treatment (Richardson & Kimura 2020).

Recently, advanced oxidation pretreatment technologies have been adopted greatly for membrane fouling control (Zhang et al. 2020). Among all these oxidation methods, the ultraviolet (UV)-based advanced oxidation processes were dominant since the UV mainly generates reactive free radical species through the UV-excited oxidants, including UV/persulfate (PS) (Fu et al. 2019), UV/H₂O₂ (Nihemaiti et al. 2018), UV/chlorine (Xu et al. 2022), and other processes (Li et al. 2023). As chlorine requires a lower cost compared to PS and H₂O₂ and can produce more free radicals per unit concentration, the combined UV/chlorine process has been widely used for the degradation of various organic micropollutants in water bodies (Xu et al. 2021). In addition, the UV/chlorine could generate a higher free radical production rate than the UV/H₂O₂ due to the higher molar absorptivity (absorbance of a solution per unit amount of substance) and the quantum yield of chlorine compared to H₂O₂. Thus, the combined process has been considered an efficient method for membrane fouling control (Zhou et al. 2021). While with pre-chlorination alone, the biofouling can be effectively reduced (Zheng et al. 2015). However, with the combined UV/chlorine process, various reactive free radicals generated during UV irradiation can degrade organic pollutants in the feedwater (Zhang et al. 2019), thereby greatly alleviating membrane fouling. Additionally, chlorine is further consumed during the radiation process (Miklos et al. 2019), and its oxidative damage to membrane materials in subsequent filtration processes thus can be significantly reduced. Therefore, UV/chlorine oxidation pretreatment for RO membrane fouling control is feasible and worthy of further investigation. However, the existing conventional mercury lamp has a low matching degree with the absorption peak of free chlorine (Ji et al. 2018), and its wavelength is fixed and cannot be adjusted, leading to a significant deviation from the process parameters and actual production conditions. UV light-emitting diodes (UV-LEDs), on the other hand, have advantages such as adjustable wavelengths (Lu et al. 2021; Silva et al. 2021), safety, and environmental friendliness (Bhat et al. 2023), they are used more and more commonly and are replacing the traditional combined UV/chlorine process. Meanwhile, the UV-LED/chlorine advanced oxidation process can not only efficiently remove various types of organic pollutants (Cha et al. 2022) but also effectively sterilize microorganisms in water bodies (Wan et al. 2020), providing a good potential approach to solving the problems of organic fouling and microbial fouling in RO processes. However, currently, no such related studies have been reported as far as we are aware, so it is necessary to study the combined UV-LED/chlorine process for the fouling control of RO membrane filtration.

In this study, the combined UV-LED/chlorine process was adopted to reduce the membrane fouling of a RO system for a domestic steel wastewater treatment plant, focusing on the alteration of organic foulant properties (molecular weight distribution, three-dimensional fluorescence characteristics, etc.) and the mineralization rate. In addition, the impact of UV-LED/chlorine pretreatment on the RO membrane filtration performance was also thoroughly examined, including the membrane surface morphology, permeate flux, and recovery rate. With these measurements, the mechanism of organic foulant removal and membrane fouling control by the combined UV-LED/chlorine process thus can be revealed. The research findings from this study can contribute to the promotion of water resource recycling and utilization in the steel industry, providing the theoretical basis and technical support for the efficient and green development of membrane-based reused water treatment.

Sodium hypochlorite (NaClO, with effective chlorine 4.00–4.99%), metronidazole, sodium hydroxide (NaOH), sodium chloride (NaCl), sodium bicarbonate (NaHCO₃), potassium iodide (KI), potassium iodate (KIO₃), potassium dihydrogen phosphate (KH₂PO₄), and sodium thiosulfate (Na₂S₂O₃) were purchased from Sigma-Aldrich company in the USA, while anhydrous sodium tetraborate (Na₂B₄O₇) and sodium persulfate (Na₂S₂O₈) were commercially available from Shanghai Aladdin Biochemical Technology Co., Ltd. All chemicals used were of pharmaceutical grade purity in the experiments, and they were dissolved in Milli-Q ultrapure water (18.2 MΩ·cm) to prepare the solutions required for the measurement of the ultraviolet irradiance.

The main process of water production in the steel plant included the ultrafiltration (UF) – weak acid bed – RO – disk-tube reverse osmosis (DTRO) – mechanical vapor recompression (MVR) system. The RO system was arranged in a two-stage configuration for this study. The feed water had a total dissolved solid (TDS) centration (TDS) of ∼3,000 mg/L, and was operated at a pressure of ∼10 bar, inducing a maximum water production of ∼170 m³/h, recovery rate of ∼80%, and salt rejection above 98%.

The UF permeate (i.e., RO feed) was collected from the steel plant before entering the lab-scale RO system. The collected water samples were filtered with a 0.45 μm filter membrane to remove the suspended particles, stored at a 4 °C fridge, and taken as soon as possible for the determination of various water quality indicators and subsequent UV-LED/chlorine pretreatment tests. During the UV-LED/chlorine treatment, the pretreated water was mixed with a certain dosage of chlorine and then irradiated with UV-LED channels at three different wavelengths, i.e., 255, 275, and 295 nm. After that, it was fed to the RO system.

Before each RO filtration experiment, a new RO membrane sheet was taken and soaked in Milli-Q water for at least 24 h. Before the RO filtration, 5 L of the water sample was treated with UV-LED and chlorine and then filtered through the RO membrane device. The RO filtration process was conducted in a constant pressure mode at 30 ± 0.3 bar with cross-flow filtration, with the concentrate and product water being circulated back to the feed tank to maintain a generally constant volume and water quality. Each RO filtration cycle lasted around 4 h.

In this study, various standard instrumentation methods were adopted to characterize the water quality and the RO membrane before and after the filtration. The main water quality parameters measured in this study included pH value, dissolved organic carbon (DOC) concentration, total nitrogen (TN), organic molecule weight distribution, etc. The total chlorine and free chlorine were examined using the N, N-diethyl-p-phenylenediamine (DPD) spectrophotometric method with a UNICO SQ-4802 UV-visible spectrophotometer (Shanghai China). The pH of the water sample was measured with a Mettler-Toledo FE20-FiveEasy precision pH meter (Switzerland). The DOC and TN were determined with a Shimadzu TOC-VCSH analyzer from Japan. The composition of organic foulants was qualitatively analyzed using three-dimensional fluorescence spectroscopy (EEM) (Hitachi F7100, Japan), and the molecular weight distribution of organic matter was analyzed with the LC–organic carbon detection (OCD)–OND instrument. The anion concentrations were obtained using an ICS-5000 ion chromatography instrument (Dionex Company, USA), equipped with a Dionex IonPac AS19 column (2 × 250 mm), and an IonPoc AG19 guard column (2 × 250 mm). In addition, the surface physicochemical properties of the RO membrane were also examined to investigate the anti-fouling mechanism, with the morphology of the clean and the filtered RO membrane measured at the Institute of Seawater Desalination and Multipurpose Utilization, Ministry of Natural Resources, Tianjin, China. This included scanning electron microscopy (SEM) (FEI Quanta 200, Netherlands), Fourier transform infrared spectroscopy (Shimadzu, Japan), Raman spectroscopy (Thermo Fisher DXR, USA), and zeta potential (Anton Paar SurPass, Austria).

Supplementary Figure S2(a) and S2(b) shows the variation in flux and the recovery rate of the RO membrane when the UF permeate water was treated with 295 nm UV-LED with different concentrations of free chlorine. It can be seen that after UV-LED pretreatment, increasing the concentration of chlorine does not make a significant improvement in the filtration flux or the recovery rate of the RO membrane. When the chlorine dosage increased from 0.2 to 1.0 mg/L, both membrane flux and recovery rate maintained nearly stable and similar in all the 4 h RO filtration, which suggested that the combined UV-LED/chlorine process played a significant role in the RO fouling control of the steel reused water, while the chlorine dose had a minor effect.

The specific components of all collected fluorescence excitation emission matrix spectra (EEM) data were then quantitatively analyzed for fluorescence component variation using parallel factor analysis (PARAFAC) (Supplementary Text S3) to successfully decompose into three fluorescence components (C1, C2, and C3) (Gao et al. 2021). C1 and C2 are components of soluble microbiological byproducts, while C3 is a component of humic acid (Gao et al. 2020b). Components with wider emission and excitation wavelengths generally have characteristics of large molecules, complex structures, and more aromatics. Therefore, the fluorescence group of component C2 may contain more condensed or polymerized macromolecular structures, with a molecular size larger than C3 and C1 (Guo et al. 2017). The three fluorescence components in this study were sorted in the order of increasing molecular weight as C2 > C1 > C3.

The characteristic variation of C1, C2, and C3 during the dual-wavelength UV-LED/chlorine process is shown in Supplementary Fig. S3. After treatment with the 255 + 275, 255 + 295, and 275 + 295 nm UV-LED/chlorine processes, the proportion of C1 increased slightly from 50.2 to 50.4, 50.5, and 50.5%, respectively; the proportion of C2 increased from 22.7 to 23.4, 23.8, and 23.7%, respectively, while the proportion of C3 decreased from 27.0 to 26.2, 25.7, and 25.9%, respectively. In addition, it can be also found that the proportion of components of soluble microbiological byproducts, C1 and C2, increased, while the proportion of components of humic acid decreased with the order of C1 > C3 > C2.

The organic matters in filtered water were classified into five chromatographic segments, i.e., biopolymers (>2,000 g/mol), humic substances (∼1,000 g/mol), building blocks (300–500 g/mol), low-molecular-weight acids (<350 g/mol), and low-molecular-weight neutrals (<350 g/mol) (Gerchman et al. 2020). Here, the OCD detector was used to monitor the organic matter content, namely hydrophobic organic carbon (HOC) and hydrophilic organic carbon (CDOC), and examine the molecular weight distribution of the CDOC (Wu et al. 2022); the UVD detector was simultaneously used to measure the UV₂₅₄ response value of humic substances, with its value over the DOC value representing the aromaticity index, while the OND detector could simultaneously obtain the content of dissolved organic nitrogen in bio-polymer and humic substances and determine the conversion of the large molecules in the filtered water into small molecules after UV-LED chlorine treatment.

Figure 7(b) shows the Raman spectrum of RO membranes before and after the filtration of the UF permeate treated with dual-wavelength UV-LED and 1 mg/L chlorine. The characteristic peaks of the contaminants on the membrane after dual-wavelength UV-LED/chlorine were similar to those after direct RO filtration and the blank membrane but with different peak intensities. It can be seen that the peaks at 1,598 and 1,580 cm⁻¹ represented the characteristic absorption peaks of the benzene ring (C = C), the peak at 1,336 cm⁻¹ referred to the characteristic absorption peak of aldehydes (C–H), the peaks at 1,144 and 1,109 cm⁻¹ corresponded to the characteristic absorption peaks of alcohols (C–O), the peak at 1,069 cm⁻¹ represented aromatic thioether molecules, and the peak at 937 cm⁻¹ referred to the characteristic absorption peak of sulfur bonds (C–S), all of which had higher peak intensities than those of the blank membrane (Wang et al. 2022b), indicating the existence of these substances in the feed. However, with the increasing dual wavelengths of UV-LED, the peak intensity of these characteristic absorption peaks increased, indicating that under the action of dual-wavelength UV-LED, shorter wavelengths led to the better degradation efficiency of such organics.

This study addressed the critical bottleneck issue of RO membrane fouling in the zero-discharge system for steel reused water treatment. The combined UV-LED/chlorine process was used as a pretreatment to reduce RO membrane fouling during the filtration of the UF permeate from a steel plant. The organic removal efficiency and RO filtration performance with the combined UV-LED/chlorine pretreatment were investigated. The main conclusions are as follows:

• (1) The combined UV-LED/chlorine process could not only efficiently remove DOC and TN from the RO feed but also achieve the transformation of organic compounds from large molecules to small ones. Compared to the dual wavelengths of 255 + 275 and 255 + 295 nm, the 275 + 295 nm UV-LED/chlorine pretreatment exhibited a double removal efficiency of 14.3% of organic compounds.

• (2) The combined UV-LED/chlorine pretreatment can significantly alter the organics properties and thus enhance the RO membrane filtration performance during the steel reused water treatment. Compared to the shorter wavelengths of 255 + 275 and 255 + 295 nm, the longer wavelengths of the 275 + 295 nm UV-LED/chlorine process played a more noticeable effect in inducing a looser and thinner foulant structure on the RO membrane surface, thus to reduce the RO fouling, and enhance the RO filtration performance with flux and recovery increasing ∼15%.

• (3) With the increase of wavelengths of the UV-LED process, the flux and recovery rate of the RO membrane were both improved, while the combined dual wavelengths of the UV-LED/chlorine process were more favorable to improve the RO filtration performance as compared to the single wavelength, to effectively alleviate RO membrane fouling, and to significantly improve the permeate flux and the recovery rate.

This work was sponsored by the National Natural Science Foundation of China (Nos 51978483, 52070144, 52270010, and 52100012), Shanghai Rising-Star Program (No. 22QC1400500), and Science Fund for Creative Research Groups (No. 52221004).

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

The authors declare there is no conflict.