Chemistry of alkaline and non-alkaline scaling in community RO for brackish water treatment operated with and without antiscalant doses

RO

Reverse osmosis

FTIR

Fourier transform infrared spectroscopy

SEM

Scanning electron microscope

EDS

Energy dispersive x-ray spectroscopy

TFC

Thin-film composite

BIS

Bureau of Indian Standards

BWRO

Brackish water reverse osmosis

TDS

Total dissolved solids

meq

Miliequivalent

Water scarcity due to ever declining groundwater levels has asked for immediate attention to establish new and innovative technologies that can be utilized for sustainable solutions to combat the drinking water crisis in Rajasthan. The state of Rajasthan, which predominantly relies on groundwater sources for its drinking supplies, faces a serious water shortage problem. It is further aggravated by the quality of its groundwater that contains high levels of fluoride, nitrate, and salinity. Dhindsa (2006) found that of a total of 75,266 water samples from various sources including open wells, tube wells, and hand pumps, 42,352 had total dissolved solids, fluoride, and nitrate over the prescribed limit 1,500, 1.5 and 45 mg·L⁻¹, respectively, emphasizing on the importance of the grave situation in Rajasthan in relation to the chemical quality of the water. As per a recent Central Ground Water Board (CGWB) report on groundwater quality (Groundwater yearbook 2016–2017), the analysis of 646 water samples drawn from different parts of the state showed that the presence of chlorides, nitrates, fluorides and sulphates exceeded the acceptable limits as per of 250, 45, 1 and 200 mg·L⁻¹ (BIS 10500 2012) in 39.46, 34.47, 41.51 and 28.45% of the samples, respectively. Since the sources do not meet the desired standards, it has opened up new challenges for the water supply engineers to come out with safe treatment technologies at an affordable cost. The Government of India (GoI) has recently issued guidelines under the ‘National Rural Drinking water program’ (NRDWP), that a minimum of 8 L per day per capita (LPCD) potable water conforming to BIS 10500 has to be supplied to meet the requirements of drinking and cooking. To comply with these, the Public Health Engineering Department (PHED), Government of Rajasthan (GoR) has installed about 3,500 community-based reverse osmosis (RO) plants with the choice of technology being governed by their capacity to tackle multiple contaminants simultaneously, compactness and modular design and low land requirements.

Reverse osmosis (RO) membrane desalination has emerged as the primary choice for removal of multiple contaminations from the brackish groundwater of Rajasthan. RO is currently being used as an energy-efficient separation technology for salt rejection of relatively small installations, with high water permeability and desalination capacity, allowing it to comply with the stringent public health standards (López-Ramírez et al. 2006; Xu et al. 2013). Polyamide thin-film composite (TFC) RO membranes are predominantly used for supplying freshwater from brackish groundwater resources in the state for their high fluxes at low applied pressures, but they face acute problems of chemical scaling. Consequently, there is a decrease in membrane permeability and lowering of permeate quality, ultimately resulting in increased energy consumption and shorter membrane life. The fouling of RO membrane can be categorised into scaling/inorganic fouling, organic fouling, particulate/colloidal fouling, and biological fouling, depending on the characterisation of foulant (Koo et al. 2013). Membrane scaling is mainly due to the crystallization of ions in the supersaturated solution onto the surface of membrane through two main pathways: surface crystallization and bulk crystallization. Both pathways are equally responsible for flux decline concurrently (Lee & Lee 2005). Scaling can also be further classified as alkaline (calcium carbonate), non-alkaline (calcium sulphate), and silica, which is based on the properties of deposits (Wang 2005). Alkaline-based scaling (calcium carbonate) can mostly exist in the form of calcium and bicarbonate ions in brackish water resources of Rajasthan; moreover, the degree of scaling depends largely on the calcium hardness and bicarbonate alkalinity, as well as the other characteristics of feed water such as pH, temperature and total dissolved solid (TDS), etc. Furthermore, the presence of biofilm onto the membrane surface can aggravate the scaling of calcium sulphate, as both processes involve the adsorption and precipitation of mineralising ions (Dahdal et al. 2016). Besides carbonates, sulphates constitute another major foulant, which is even more challenging to clean compared to their alkaline counterparts. Therefore, the best practice to inhibit the sulphate scaling is by operating the desalination system below the saturation level (Landaburu-Aguirre et al. 2016). Biofouling is the phenomenon of microorganism adhesion and proliferation on the surface of the membrane (Maddah & Chogle 2017). The presence of bio-fouling is not of much concern because of the very deep groundwater table in the state of Rajasthan. The unification of the process such as deposition, reactions, and interactions of high molecular-weight organic molecules tend to result in organic fouling, e.g. natural organic matters (NOMs) and/or TEPs (transparentexopolymer particles) with the membrane surface (Goh et al. 2018). Organic fouling is usually found in membranes treating seawater having high ionic strength and natural organic matters, but a deep groundwater table of Rajasthan does not permit organic contamination and hence the organic fouling is almost non-existent. Silica scaling is a ubiquitous problem in the brackish water RO plants of Rajasthan due to non-cohesive sandy strata. Consequently, feed water carries a lot of sand with it resulting in frequent choking of the membrane. Deposition of silica due to silica polymerization on the surface of the membrane can cause severe concentration polarization. This can lead to an increase in hydraulic resistance and, as a result, it decreases the permeate flux or requires higher operating pressure to maintain the same flux (Ang & Elimelech 2007). Therefore, the main challenges are to understand the reasons behind the observed accelerated fouling of membrane that causes substantial flux reduction and bring out the significant differences in the mechanism of fouling between seawater and brackish water desalination.

The main objectives of this study were: to perform detailed chemical analysis of water samples from feed, treated and reject streams; carrying out mass balance to identify the major inorganic scalants depositing on the membrane; decipher the scaling mechanism through the calculation of Langlier Saturation Index (LSI) and Ryznar stability index (RSI) for plants operating with or without antiscalants; corroborating these results with membrane autopsies and detailed examination of salt deposits through SEM-EDS and FTIR studies, and bring out differences in brackish and seawater desalination.

Comparatively old community RO plants installed at locations having relatively high TDS groundwater source in Bharatpur district of Rajasthan were selected for this study. Rejected membranes after severe fouling were obtained from these plants, along with water samples of feed, permeate and RO concentrate (ROC) streams. Villages PIDHYANI and MAHAGAYEN were selected based on antiscalant being used, while it was not used at GANGARSOLI and HANTRA plants. All the plants were of similar configuration and flow diagram of the RO plant is shown in Figure 1.

Water samples from the three streams described above were analysed for various chemical parameters to assess their efficiency of removal as well as to indicate major foulants depositing on membranes through a mass balance approach. pH of the samples were measured using portable pH meter (HANNA, model: HI98130) and TDS was monitored through a TDS tester (HANNA, model: HI98302). Calcium, potassium and sodium were assessed by microprocessor-based flame photometric method (ESICO, model-1385/1382), total hardness by EDTA titrimetric method, magnesium hardness as CaCO₃ was calculated indirectly by subtracting the calcium hardness as CaCO₃ from the total hardness, chloride by Argentometric method, fluoride and nitrate by an ion-selective electrode (HACH model: ISEF121 probe and ISENO3181 probe), sulphate by turbidimetric method (Shimadzu, model: UV1800 240 V), and alkalinity by titration method as described in 23rd edition of Standard Methods for the Examination of Water and Wastewater (Rice et al. 2013).

Characterisation of surface morphology of the fouled membrane was carried out by Nova Nano FE-SEM 450 Model Scanning Electron Microscope with a resolution of 1.4 cm⁻¹, and it was coupled with the energy-dispersive X-ray spectroscopy (EDX) to analyse the elemental composition. The infrared spectra of the membranes were observed using Perkin Elmer Spectrum Version 10.4.00 FT-IR spectrometer with a working range of 4,000–400 cm⁻¹ with a resolution of 4 cm⁻¹.

Acceptable percent difference criteria are:

pH is the measured water pH

Table 1 presents the salient features of four plants operating with or without the use of antiscalants, visited during the study. These showed relatively low recovery of 25–35% being obtained due to excessive fouling observed in the membranes despite excessively high operating pressures being employed. The important thing to note was that the first two plants, installed during the initial phase of the project, did not use antiscalants and highly accelerated fouling was observed in these resulting in a very short active life of the membranes of 6–10 months.

The unaccounted-for ion percentage represents its tendency to deposit on the membrane, which varied widely among different anions and cations. It was observed that removal of chemical parameters depended strongly upon the quality of feed water with carbonates and sulfates of calcium and magnesium (as represented by hardness and alkalinity) showing very high tendency to deposit on the membranes. A large fraction of these ions remained unaccounted for in the first two plants at GANGARSOLI and HANTRA, which were not using any antiscalants as shown in Table 2.

The ratio of chlorides to hardness was much lower than that found in sea water, indicating a wide difference in the fouling tendency (Sarath Prasanth et al. 2012; Mugisidi & Heriyani 2018; Boyd 2020). While the sea water desalination suffers from biological fouling, it was almost non-existent in brackish water desalination, which was dominated by inorganic scaling as described later. In the present system, both alkaline and non-alkaline scaling were equally dominant as alkalinity, hardness, calcium and sulphate indicated high tendency to deposit.

Interestingly, when the results of aforementioned plants were compared with those of MAHAGAYEN and PIDHYANI plants, which were using antiscalant, it was observed that besides monovalent ions, even alkalinity, hardness and calcium indicated minimal tendency to deposit on the membrane. The use of antiscalant changed the scenario entirely and only sulphates were observed to exhibit the tendency to deposit as the percentage of unexplained ions dropped to a low value for all other parameters at these plants. Thus, only sulphates remained as the major culprit for chemical scaling and need to be tackled through specially designed antiscalants. The plants using commonly used antiscalant showed enhanced life span of membrane taking it to the range of 2 to 2.5 years. This represented a good response to antiscalant in the system to alkaline scaling of calcium carbonates. However, there was a high percentage of unaccounted-for sulphate in the system in all the four plants indicating that the antiscalant was not effective against sulphate and hence non-alkaline scaling remained the major concern for the membrane life.

Addition of antiscalants increases the effective solubility limits of scaling salts and hence the economic benefit of achieving higher product recovery (Asadollahi et al. 2017). Commercially available antiscalants can be classified into three major categories: phosphates, phosphonates, and polycarboxylates. Polyphosphates, especially sodium hexametaphosphate (SHMP) ((NaPO₃)₆), was the first antiscalant commercially available to the membrane industry. SHMP is inexpensive but unstable compared to polymeric organic scale inhibitors. Organophosphonates are more effective and stable than SHMP. They act as antifoulants for insoluble aluminum and iron, keeping them in solution. Polyacrylates (high molecular weight) are generally known for reducing silica scale formation via a dispersion mechanism. During the past two decades, new generations of antiscalant have emerged commercially, in which the active ingredients are mostly proprietary mixture of various molecular weight polycarboxylates and polyacrylates. The common antiscalant used in Rajasthan plants is ‘Aventura’ make, as informed by the manufacturer, belongs to polyacrylates category and has been supplied to the plants in Bharatpur for its known benefits against organic fouling as well as carbonate scaling, which were duly reflected in our results. However, many plants in that area receive raw water that contains high sulfate concentration, which does not appear to respond to this antiscalant as evidenced by our observations. A probable remedy could be to use poly citric acid-based chemicals to tackle such scales (Zhao et al. 2016). Therefore, it is important to optimize the antiscalant treatment with respect to its type and dosage, identifying the dosage – induction type relationship for the extended level of super saturation. In RO plants operating on seawater with TDS in the range of 35,000 mg/L, scaling is not as much of a problem as in brackish water plants because the recovery of sea water plants is limited by the osmotic pressure of the concentrate stream to 30–45%. For safety reasons, however, a scale inhibitor is recommended when operating above a recovery of 35%. On the contrary, the plants treating brackish water aim to recover typically 40–60% of water and sometimes even higher, which makes them highly prone to scaling (Amjad 1996; DOW Water & Process Solutions 2011).

MAHEGAYEN and PIDHYANI plants were operating at very high pressure levels of 27 and 17 kg·cm⁻² and low recovery ratio 33.33% and 27.5%, respectively, as shown in Table 1. This high pressure and low recovery required more frequent cleaning, but the routine citric acid or HCl cleaning did not show good efficacy. Reports are available which show that, compared to citric acid, poly citric acid presents a much better option for cleaning membranes affected by sulphate scales and an alkali followed by an acid cleaning cycle is a superior option than plain acid cleaning in such cases (Zhao et al. 2016). This underscores the importance of employing the right antiscalant. Inappropriate dose following the stoichiometry of reactions, besides changing the operational parameters of the plants and their cleaning cycle to enhance the life of the membrane.

In a similar study by our group on a 1,000 L/h capacity community RO plant at Siroli, Jaipur, Rajasthan, established under corporate social responsibility by Bosch India, a detailed cost analysis was carried out for the treated water in an assumed life cycle of 3 years, which was the average life of RO membranes in that plant (Ray et al. 2020). The total operating cost per litre of water produced was 0.1917 rupees/L (equivalent to USD 2.62 per 1,000 L assuming a conversion of 1 USD = 73 Indian rupees), which had a cost breakup among various components as follows: energy cost, 17.21%; cost of procuring raw water, 18.85%; cost of antiscalant, 2.39%; membrane replacement cost, 9.16%; pre-filter replacement cost, 2.62%; operator's salary cost, 44.83%; other miscellaneous costs including minor repairs and replacement of parts, 4.95% of the total cost. Thus, the cost of antiscalants constituted a low fraction of the total operating cost associated with the treated water.

The removal was good for all monovalent forms and the tendency to deposit on the membrane was minimal as was exemplified by the Tables for all the four plants. NaCl has a high solubility in water because water is a polar molecule and sodium and chloride form a strong ionic bond with each other, it plays a negligible role in chemical scaling (Pinho & Macedo 2005). Another interesting observation was that the removal efficiency of fluoride was high at all the plants. Groundwater in Rajasthan generally has a high concentration of hardness and alkalinity, which bind chemically with fluoride and hence its removal increases substantially as is evidenced by the values shown in Table 2. In an alkaline environment, the fluoride ion can be easily removed from the feed due to its affinity to both calcium and magnesium; further, it does not get deposited on the membrane because OH⁻ and F⁻ ions have similar radii facilitating exchange with each other (Khan et al. 2013). Due to the greater solubility of sodium fluoride in water, higher levels of fluoride can lead to higher concentrations of sodium ions; consequently, high sodium levels result in high alkalinity of product water when associated with higher bicarbonate concentrations (Karthikeyan et al. 2000; Pinho & Macedo 2005).

Ion-balancing calculation was performed to check whether the principle of electrical neutrality is followed in the chemical analysis. Only the most dominant anions (chloride, bicarbonate, fluoride, sulphate, and nitrate) and cations (calcium, magnesium, sodium, and potassium) were covered while performing ion-balance.

Tables 3 and 4 encompass parameters such as ionic concentration in meq·L⁻¹, cation to anion ratio and percent difference of ion sums. The calculated values were within the limit of acceptance criteria as mentioned in APHA manual (Rice et al. 2013). The results of most of the samples have an anionic concentration marginally higher than the cationic concentration, which may be due to some encountered cations such as soluble Fe, Mn, and silica that were not included during ion-balance or presence of suspended solids, which may elevate the apparent alkalinity because carbonates dissolve slowly in the presence of suspended solids during titration and adsorbent cations may exchange with H⁺ titrant.

Calcium carbonate (CaCO₃) saturation indices, abbreviated as LSI and RSI, were used to evaluate the scale-forming or dissolving tendencies based on the feed water composition. Table 5 summarizes that all the results that indicate LSI > 0, i.e. +ve, for all the plants implying that all feed waters were supersaturated and tended to precipitate calcium carbonate. The RSI value was in a range of 4.68–5.66, suggesting heavy to light deposition tendency. When the results of reject water were compared with those of feed water, it was found that LSI and RSI values were almost same at GANGARSOLI and HANTRA plants but these were significantly different from those of the other two systems at MAHAGAYEN and PIDHYANI, thus bringing out the benefits of the application of antiscalants distinctly.

As there was a low concentration of unexplained ions at MAHGAYEN and PIDHYANI plants as can be seen in Table 2, it indicates that almost all calcium carbonate started to reach the reject stream without depositing on the membrane. This was further exemplified by a significant difference in the values of LSI and RSI of feed and reject streams and represented effective functioning of the antiscalant.

Results of chemical analysis depicted that the major unaccounted salts are due to scaling of calcium. A small fraction of salts of Ca may also have deposited on the surface of membranes of the pre-filtration units and remained unaccounted, though the majority would deposit on RO membranes. From the SEM-EDS analysis of fouled membrane for surface characterization, as shown in Figure 2, it is evident that the peak of calcium almost singly dominates the deposited mass fraction justifying high percentage of unaccounted for salts of calcium indicated by the mass balances in the plants that were not using antiscalant, which can also be seen from elemental composition result of fouled RO membrane of GANGARSOLI, Bharatpur not using antiscalant (P1) in Table 6.

The dominant microstructures belonged to certain morphic forms of calcium carbonate. Rajasthan is situated in the hot arid zone and receives good sunlight yearlong with day temperature generally remaining above 40 °C in summers. As can be seen from Figure 3, Calcium carbonate exhibits reverse solubility with temperature and this perhaps is the main reason for dominant calcium carbonate scaling in this region. Calcium carbonate changes from amorphous to crystalline form above 40 °C and its microstructure presents aragonite form above this temperature while it is predominantly in the calcite form below 30 °C, both being less soluble in water (Antony et al. 2011). These two forms dominated the membrane salt deposits as is evident from the SEM image of the salts deposited on membrane shown in Figure 2. The shape of the crystals appeared rhombohedral (staircase like crystal) of calcite, which exists as the crystalline polymorphic form of calcium carbonate.

However, after using the antiscalant it was found that the SEM image was not depicted any crystal of calcium carbonate, which was crossly corroborated by the much lower peak of calcium and, moreover, a distinct peak of sulphur appeared in EDS analysis as shown in Figure 4 and Table 6. This indicates the shift in the chemical composition of the deposits as carbonates responded well to the antiscalant doses but sulfates still had the tendency to precipitate as also reinforced by the chemical analysis of samples shown in Table 2.

Some polymorphic forms of calcium carbonate still existed, which may also indicate the insufficiency of the antiscalant dose as the feed water of MAHGAYEN has very high hardness concentration and the dosages are randomly administered without assessing stoichiometry. Antiscalants are polyelectrolytes that have a molecular weight ranging from 2,000 to 10,000 Da. The crystallization of the calcite can be controlled by overdosing of these as has been reported by Kim & Yang (2017) and the shape of crystals was observed to be converted from rhombohedral to pyramidal as exemplified in Figure 5, which represents the SEM micrograph of PIDHYANI having much lower hardness compared to that of MAHGAYEN but the same antiscalant dose.

Consequently, it was concluded that antiscalant being used in the field is effective against alkaline scaling and responds less to the non-alkaline scaling. Therefore, antiscalant should be screened for particular foulants at that plant prior to their application and its dose also requires optimization. Membrane autopsy and chemical analysis of plants revealed that RO brackish water membranes were fouled with inorganic substances such as calcium sulphate and calcium carbonate. On the contrary, seawater reverse osmosis (SWRO) membranes are mainly fouled by biological growth, particulate materials and organic matter and require more intensive pre-treatment (Matin et al. 2016).

FTIR analysis can be used to identify unknown organic fouling and some inorganic material. Therefore, rejected membrane from of Bharatpur district was analysed to assess whether any specific functional groups and inorganic compound were present. The peak at 1,580 cm⁻¹ represents aromatic C═C type of bond, which indicates the polysulphone layer of thin-film composite membrane, similarly reported in Rahman et al. (2018) that the peaks within the 1,700–1,500 cm⁻¹ wavelength range are typical of polymer made of Udel polysulphone as shown in Figure 6 (Rahman et al. 2018).

While the characteristic peak at 1,004 cm⁻¹ wavelength was the most intense, which corresponds to the Si-O stretching band, i.e. silicate (Dos Santos Pereira et al. 2017). Besides, the peak at 460 cm⁻¹ belongs out of plane Si-O deformation (Shokri et al. 2009); 521 cm⁻¹ represents Si-O bending; and 797 cm⁻¹ presents symmetrical stretching of Si-O-Si (Ojima 2003). This is due to heavy silica scaling observed in Rajasthan due to predominantly non-cohesive sandy strata as a lot of soil gets sucked into the feed water. When the concentration of silica exceeds its solubility limit of 120 mg·L⁻¹, the solubility of silica decreases that tends to silica polymerization and formation of amorphous colloidal silica at the pH 7.5. Antiscalant may be used to reduce the silica scaling by inhibiting colloidal silica from attaching to the membrane surface but may be less effective and more expensive when silica concentration exceeds saturation limit (Bremere et al. 2000). Hence, to avoid silica scaling, we recommended an additional sand filter in the pre-filtration system, which comprised earlier micro- and ultra-filter in series. The peak at 689 cm⁻¹ and 917 cm⁻¹ probably belong to the aluminosilicate that is the major component of kaolin mineral (Chen et al. 2015). As a comparison, the seawater membrane indicates predominantly organic fouling that defines most of the peaks observed in FTIR (Rahman et al. 2018).

The study brought out the subtle differences in the tendency for inorganic scaling of RO membranes for the treatment of brackish water, which are quite different from the fouling observed in seawater desalination. The organic fouling was almost non-existent in brackish water desalination, while this is a big problem in sea water desalination. Calcium carbonate almost singly represented the scalants due to its solubility and chemical equilibria of reactions at the given environmental conditions when the antiscalants were not used resulting in lowering of membrane life. The average membrane life was found to be 2–2.5 years with the use of antiscalant and 6–18 months without using antiscalant, which was ascertained using questionnaire surveys conducted.

Application of antiscalants worked effectively against the deposition of calcium carbonate and a significant increase in lifespan was observed, however, sulfates still remained largely unaffected and started dominating the scales. This shift in the nature of scales was well explained by the saturation indices LSI and RSI as well as SEM-EDX analysis and supported by a mass balance of ions. Silica was another major inorganic scalant of geological origin that did not respond to antiscalants as indicated by FTIR analysis of fouled membrane and hence warrants strengthening of the pre-filtration system. The results of the study will be fruitful in providing a sustainable solution to the accelerated scaling observed in field plants installed for brackish water desalination.

The authors thank the WSSO, Public Health Engineering Department, Rajasthan, India for providing financial as well as logistic support for the study. Mr Ketan Dalal of M/s Orion appliances is also acknowledged for help during site visits to the RO plants studied.

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