Can UVC-LEDs mitigate biofouling in community-scale photovoltaic-powered reverse osmosis systems? Open Access

The lack of safe drinking water infrastructure in off-grid communities is a significant risk to sustainable development, especially for low- and middle-income countries (LMICs). Photovoltaic-powered reverse osmosis (PVRO) has emerged as a promising method due to its performance, scalability, consistency, and the robust global supply chain of its components. However, reliability issues like biofouling can quickly reduce its performance, shorten membrane lifetime, and hinder adoption in off-grid settings. Ultraviolet light emitting diodes (UV-LEDs) ranging from 200–285 nm wavelengths reduce the number of microorganisms in water but there are mixed results about their use on mitigating biofouling in reverse osmosis systems. Herein, we aim to provide a preliminary assessment on whether UV-LED pre-treatment can mitigate biofouling in PVRO systems. Analysis of the E. coli concentration pre- and post-UV treatment for batch and flow cell experiments demonstrated reduced bacterial concentration after treatment but suggest that remaining bacteria after UV treatment can grow back over time, i.e., UV-LEDs do delay, but not completely eliminate, biofouling. These results encourage further investigation into how UV-LEDs can be optimally integrated in PVRO systems and also set the premise for controlled biofouling mitigation studies for intermittently operated, small-scale PVRO.

The United Nations' Sustainable Development Goal (SDG) 6.1 focuses on improving water access for the 2.2 billion people (27% of the global population) who do not have access to safely managed drinking water (United Nations 2022). This water stress is more acute in low- and middle-income countries (LMICs) in Africa and South Asia, especially in regions with zero or limited connections to water distribution networks and the electricity grid (Oxfam 2018). It is imperative to develop solutions that can be tailored to their unique settings.

Currently, solutions such as rainwater harvesting and groundwater wells provide water relief to some of these communities. However, sustainable use, adoption, and maintenance of these solutions are often hindered due to irregular and unreliable rainfall patterns, pump failures, groundwater contamination through arsenic and other heavy metals, seawater intrusion (in coastal areas), and flooding. Furthermore, with increasing knowledge of global technologies due to cell phones and connection to the internet, there is a growing call from LMICs for equitable and global standard solutions (Lum 2011). These challenges motivate exploring mature, robust technologies established in the industry and adjusting them for use in remote, off-grid communities.

Membrane and thermal processes are standard in industry for desalinating sea and brackish water, which accounts for the majority of the global water supply. Membrane processes operate by pressuring water through a semi-permeable membrane, which captures dissolved salts, bacteria, and other contaminants (Pooi & Ng 2018). Thermal processes operate by evaporating water through a semi-permeable membrane and then condensing it afterwards for collection (Subramani & Jacangelo 2015). The most robust desalination method is reverse osmosis (RO), occupying 65% of the global installed desalination capacity (Mito et al. 2019). RO's widespread adoption comes from factors such as outperforming other filtration methods in their ability to capture contaminants (Pooi & Ng 2018), having the lowest unit price of water compared to other desalination techniques (Qasim et al. 2019), and being able to adapt them in various configurations or sizes for different settings (Alghoul et al. 2009). However, a major concern with RO systems is that they are energy intensive (Abdallah et al. 2005), which limits their sustainable application in LMICs.

Recently, the growth of photovoltaic-powered RO (PVRO) has increased the prospect of implementing small-scale desalination (1–5 m³/day) for remote, off-grid, and climate-vulnerable communities in LMICs, especially those that have year-round sunshine. For example, in the Yucatan Peninsula region in Mexico, a 1,000 L/day PVRO system has been purifying brackish well water and rainwater for a village with 450 people (Elasaad et al. 2015). In the village of Bercy, Haiti, a PVRO system with a capacity of 900 gallons/day was constructed to filter water from a locally built well, eliminating the need for the community to walk 2 h to obtain drinking water (Bian et al. 2016). Islam et al. (2018) highlighted the importance of the maintenance of small-scale desalination plants that were implemented in coastal regions in Bangladesh, where rising sea levels and saltwater intrusion exacerbate freshwater scarcity. Other implementations are being evaluated around the world (Almaktoof et al. 2015; Education Post 2020; Hekmatmehr et al. 2024). Bilton et al. (2011) evaluated community-scale PVRO systems and concluded that by properly configuring a PVRO system based on the location, it can be more economically feasible than diesel-powered RO or transported water. With respect to SDG 6.1, PVRO appears to be a valuable way to help close the gap in sustainably obtaining safe drinking water. Concerns such as intermittency, local user understanding, and capital and resource constraints still present challenges and must be addressed, as they can accelerate fouling-related reliability issues.

Membrane fouling remains a persistent challenge for membrane technologies, including PVRO systems. Inorganic fouling, such as scaling, occurs from salts precipitating on the membrane's surface (Sim et al. 2018). Biofouling refers to the bacteria in the feedwater producing a gel-like biofilm on the membrane (Sim et al. 2018). All these mechanisms reduce the membrane's effective area over time, affecting the permeability. Along with the lower volume and quality of the permeate, fouling ultimately leads to higher costs due to maintenance, repairs, and membrane replacement.

There are various methods to manage fouling to prolong the membrane's lifetime, such as modifying membranes to have an antifouling layer, pre-treating the feedwater to reduce contaminants, and cleaning the membrane (e.g., backwashing and chemical agents) (Jiang et al. 2017). Our previous work investigated the effect of scaling on small PVRO systems and found that intermittent operation did not negatively affect the membrane's permeability over time compared to continuous operation (Sarker & Bilton 2021), and using an anti-scalant and rinsing with clean water before shutting down the system prolonged the membrane's permeability (Freire-Gormaly & Bilton 2019). However, managing biofouling in community-scale PVRO systems in LMICs still needs to be investigated, as the intermittent operation may increase biogrowth, and the complex nature of living organisms can make standard cleaning methods inadequate for their circumstances. For instance, chlorinating the water to kill bacteria will damage the membrane (Matin et al. 2011), leading to an early replacement, and some LMICs do not prefer chlorinated water. Sarker et al. (2025) discovered that although 15–30 min of rinsing following a 1-min osmotic backwash could significantly slow down scaling in intermittently operated RO systems, it may not work effectively for biofouling, which still requires physical and chemical interventions. Tow et al. (2016) showed that in alginate scaling, rinsing breaks away chunks of foulant, which is not ideal for cleaning biofilm as it can disperse and reattach downstream. Additionally, backwashing is not ideal as a sole cleaning mechanism for RO systems, especially in the context of marginalized communities, since it reduces the total permeate quantity and requires trained personnel. An emerging, low-cost solution to mitigate biofouling may be ultraviolet light-emitting diodes (UV-LEDs).

Disinfecting feed water using UV, particularly UV lamps, is a common practice in water treatment (González et al. 2023). The lamps emit wavelengths in the ultraviolet light C (UVC) range (200–280 nm), which inactivates bacterial cells, making the water safe for consumption. Recently, UVC-LEDs gained popularity as an alternative to UV lamps because of their smaller form factor, low energy requirements (see the Supplementary Material for details), non-toxicity, low cost, and longer lifetime (Song et al. 2016). These factors also make it a favorable integration into community-scale PVRO systems for remote, off-grid communities to control biofouling on membranes and prolong their operational life. As biofilm consists of living and dead cells, UV lamps and LEDs are being studied for biofouling control based on the presumption that inactivated bacteria should produce less biofilm on the membrane. Additionally, UV light can impact the extracellular polymeric substance (EPS), which is composed of proteins, polysaccharides, lipids, nucleic acids, and biomolecules (Subhadra 2022), in a biofilm. (Wang et al. 2023a) found a reduction of β-sheet in proteins in the EPS, which was attributed to UV light-inhibiting genes that produce the protein. However, as outlined in Table 1, there are mixed results in the literature on whether treating feedwater with UV actually mitigates biofilm formation. Some studies found the expected decrease in membrane biofouling, while others found it actually resulted in an increase.

Given the mixed results in the literature, our study aims to provide a clearer understanding of the uncertainties associated with UVC-LED pre-treatment for RO systems and to evaluate whether UVC-LED pre-treatment-induced disintegration of the bacteria (in the feedwater stream) could minimize biofouling on small-scale RO systems. These objectives were addressed through a series of batch scale tests to analyze the impact of UVC-LED on bacterial growth, followed by evaluating the same on RO membrane biofouling. Additionally, we explored the challenges and uncertainties of implementing UVC-LED-based pre-treatment for RO feedwater, aiming to facilitate the design of optimal UVC-LED units. The method section details the experimental setup, feedwater preparation, and analysis techniques, while the results and discussion section provides an in-depth analysis of our findings. This work offers a preliminary proof-of-concept for UVC-LED's scope as an RO biofilm mitigation tool and serves as an initial outlook for further investigation into how UVC-LEDs can be optimized for RO pre-treatment and how factors such as intermittency and variabilities in feedwater compositions in PVRO systems could affect their performances.

The concentrate was circulated back into the feed tank. After 5 days, the flow cell was opened, and the membrane was extracted for biofilm analysis. Feedwater samples throughout the 5 days of experiments were also collected for measuring bacterial concentration by optical density measurement (OD).

For the batch experiment, Petri dishes were filled with bacterial media in 10% LB: (a) the control was not exposed to any UV, (b) and (c) were exposed to UV light using the 275 nm UV-LED chip for 30 s and 5 m, respectively. The equivalent UV doses are presented in the Results and Discussion section. Then, the bacteria in the Petri dishes were allowed to grow for 5 days, with OD measurements and cell counting done right after UV exposure and on the second and fifth days. On day 5, after removing the bacterial media, 6 mL of 0.05% crystal violet was used to stain the biofilm that was formed in the Petri dish. After flushing with tap water and thoroughly dissolving the stain in 4 mL of acetic acid, 1 mL of this solution was measured with the spectrophotometer to get an OD measurement, which was indicative of the amount of biofilm present in the plate.

During UV-enabled flow experiments, first, the feedwater was pumped through the Pearl Aqua Micro UV unit to recreate a UV-dosage equivalent of at least 40 mJ/cm², 280 nm, and stored in the feed tank (Figure 3(a)). For the control experiment (no UV), the bacterial feedwater was still passed through the Pearl Aqua Micro UV unit, but the unit was switched off. Then, the feedwater was pumped through the system continuously for 5 days at 20 psi and 5.5 mL/s (Figure 3(b)). It is to be noted that during this preliminary study, the pressure was kept minimal to ensure the 3D-printed flow cell did not leak or fail, and a small volume (∼60 mL/day) of permeate was produced. The OD of the feedwater was measured throughout the experiment to monitor the bacterial concentration. After 5 days, the flow cell was opened, and the membrane was extracted for biofilm analysis via imaging.

During batch and flow experiments, the OD of the bacterial media and feedwater was also measured using a spectrophotometer (Agilent Technologies, USA) to obtain an understanding of how the concentration of bacteria in the feedwater changed with time. For the feedwater OD measurement, 1 mL samples were collected every few hours, and cell counting was done to determine the number of live bacteria in the feedwater (i.e., planktonic bacteria). Serial dilution was used, i.e., 100 μL of solution was diluted in 900 μL phosphate buffer (PBS) and repeated 8 times. Afterward, three 20 μL drops from each dilution microtube were placed in agar plates and incubated at 35 °C for 16–18 h before it was counted for viable colonies.

A smartphone camera (iPhone XR, USA) was used to photograph the membrane right after the experiment was completed and the flow cell was opened. These images provided a complete, undisturbed view of the amount of biofilm formed. Areas of the membrane with biofilms were also studied using the EVOS FL Auto Imaging Fluorescence Microscope (ThermoFischer Scientific, Canada). First, the membrane was cut into ∼0.75 cm × 0.75 cm pieces using sterilized tweezers and scissors. Then, after washing the membrane with 1 × PBS, 200 μL of SYTO 9 and propidium iodide mixture (ThermoFischer Scientific, USA) was placed on the membrane in a Petri dish wrapped in aluminum foil. After 30 min, the dye was removed with a pipette, and the samples were placed in the fluorescent microscope for imaging.

Biofilm formed on the Petri dishes was quantified using optical density measurements, which showed that UV exposure did reduce biofilm growth in comparison to non-UV-treated samples throughout days 2 and 5 (Figure 5(b)). This reduction may be attributed to UV-induced inhibition of genes related to EPS production, a critical element in biofilm growth (Wang et al. 2023a). It is interesting to note that higher UV exposure (180 mJ/cm²) resulted in more biofilm than 18 mJ/cm², hinting that excessive UV dosing could aggravate biofouling, as previously observed by Wu et al. (2021). The aggravated stress of the bacteria cells after 180 mJ/cm² exposure may have also contributed to the negligible decline in biofilm OD from day 2 to day 5, unlike the control and the 18 mJ/cm² UV-exposed samples (Dawan & Ahn 2022). These findings hint at optimizing the UV exposure level, as simply increasing the dosage may not yield the best biofilm control outcomes for RO operations. Overall, more comprehensive experimental analyses are needed to detail the underlying mechanistic interaction between UV exposure, bacterial physiology, and biofilm formation dynamics to refine UV-LED dosing strategies for PVRO systems.

These results indicate there could be potential advantages in using a UV-LED pre-treatment system to mitigate biofouling in membranes, though further analysis needs to be done to account for uncertainties due to concentration differences and limited samples. Our previous works (Sarker & Bilton 2021; Sarker et al. 2025) demonstrated that inorganic scaling in community-scale PVROs can be effectively addressed through optimized rinsing and backwashing protocols. However, these approaches alone may not be sufficient for addressing biofouling, which remains a persistent challenge in such systems, particularly under intermittent operation and low-maintenance conditions (Tow et al. 2022; Sarker et al. 2025). Our current results highlight the potential of UV-LED's ability to disrupt microbial growth in the feedwater, which could be integrated with existing rinsing and backwashing strategies to tackle biofilm formation on RO membranes more comprehensively, contributing to longer operational lifetimes for PVRO membranes in last-mile applications. As an immediate next step, advanced characterization techniques such as scanning electron microscopy, transmission electron microscopy (Sarker et al. 2022), confocal microscopy (Tow et al. 2022), and Fourier-transform infrared spectroscopy (Martino 2018) are being employed to further investigate bacterial growth and biofilm development dynamics on membranes following UV-LED pre-treatment. These analyses will be included in future work to provide a more comprehensive understanding of bacterial growth and biofilm development dynamics on membranes following UV-LED pre-treatment. Additionally, more macroscale images of the undisturbed biofilms from those studies are being collected into a database so that they could serve as inputs in future image-based biofouling analytical models for further analysis.

Biofouling mitigation in community-scale PVRO remains a critical hurdle for sustainable drinking water access for remote, off-grid, and climate-vulnerable communities in LMICs. This study provides experimental evidence that integrating UVC-LEDs as a pre-treatment technology in PVRO operation can significantly reduce biofilm formation on membranes during stagnant and operational periods. A 275 nm UV-LED chip, with an average irradiance of 0.602 mW/cm², achieved a bacterial log reduction of 0.48 and 5.26 after 30 s and 5 min of batch-scale UV exposure, respectively. With bacterial reduction being confirmed, bacterial feedwater exposed to at least 40 mJ/cm² UV-LED was used in a bench-scale RO crossflow system to evaluate biofilm growth. The results demonstrated 22.7% reduced biofilm coverage on the membrane compared to the control over 5 days of operation, highlighting that UV-LED pre-treatment of bacterial feedwater can impair bacterial adhesion and biofilm formation on RO membranes, rather than requiring complete bacterial inactivation. These findings establish UV-LEDs as a viable biofouling control strategy for community-scale PVROs used in LMICs, demonstrating their potential to extend membrane lifespan and reduce dependance on periodic chemical-based cleaning. The results also underscored that integrating UV-LEDs with existing rinsing and backwashing protocols could provide a more holistic, low-energy approach to fouling mitigation in PVRO and encourage further experimental investigation into validating these results under dynamic field conditions, considering factors such as intermittency, nutrient availability, and variabilities in operating conditions in PVRO systems. Future work should also focus on optimizing flow dynamics, LED arrangement, and reactor geometry to maximize uniform UV exposure and dosage to feedwater in PVRO systems while minimizing energy consumption.

The authors would like to thank Mina Mahdian and summer students Dina Bernstein and Joaquin Arcilla for their help in progressing this research. The author would also like to thank Dr Ron Hofmann and Dr Kevin Golovin for providing guidance and support in reviewing results and planning out future steps.

The authors are thankful to the Natural Sciences and Engineering Research Council (NSERC) of Canada for the NSERC Discovery Grant, the Department of Mechanical and Industrial Engineering at the University of Toronto for their research fellowship, the University of Toronto Water Seed Grant, and the Mitacs Globalink Research Award.

N.S.K., N.R.S., and A.M.B. conceptualized the study, wrote, reviewed, and edited the article. N.S.K. and D.A. conducted the experiments. D.A. and B.H. reviewed and edited the article.

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

The authors declare there is no conflict.