Incidence of chlorination by-products in an institutional drinking water distribution network, Islamabad, Pakistan, using response surface methodology

Chlorination of drinking water supplies is the most common practice of water disinfection in Pakistan, preventing the transmission of many waterborne diseases due to inadequate disinfection in water distribution networks (Amjad et al. 2013). Despite its role in disinfection, chlorine is also responsible for oxidizing natural organic matter (NOM) in water, leading to the formation of various disinfection by-products (DBPs) such as trihalomethanes (THMs) and haloacetic acids (HAAs) (Richardson et al. 2007).

Among these by-products, THMs such as chloroform (CF), bromodichloromethane (BDCM) and bromoform (BF) are of major concern due to their severe toxic and carcinogenic effects in humans (USEPA 2003; Richardson et al. 2007). Depending on the potential health risks, the United States Environmental Protection Agency (USEPA) have set limits for total THMs (i.e. sum of chloroform, bromoform, bromodichloromethane and dibromochloromethane) at 80 μg/L in potable water (USEPA 2003).

Besides brominated and chlorinated THMs, iodinated THMs (I-THMs) may also be formed when iodide ions are present in water. New concerns regarding human health with respect to I-THMs were raised by Plewa et al. (2004), who reported I-THMs, particularly iodoform, as more toxic than brominated and chlorinated THMs. The taste and odor threshold for iodoform is in the range of 0.02–5.00 μg/L, and when surpassed may prompt organoleptic issues and consumer complaints. Currently, there is no published standard analytical method for I-THMs in water (Allard et al. 2012). Therefore, there is a dire need to constantly monitor the levels of THMs and their precursors in drinking water supplies (Zia et al. 2005).

In Pakistan, there is a lack of reported work regarding identification of toxic DBPs in drinking water supplies. In a study by Amjad et al. (2013), average total THMs concentration (TTC) was found to be approximately 143 and 259 μg/L for Rawalpindi and Islamabad, respectively. Solid phase microextraction technique (SPME) is a rapid and sensitive technique for THMs determination. The sensitivity of the technique depends on a number of parameters that affect extraction of analytes from water, such as fiber type, sample volume, stirring, salt addition, extraction/desorption time and temperature (Bahri & Driss 2010). The conventional approach for process optimization is time consuming, requires a large number of experiments to be performed and is also expensive. According to Dos Santos et al. (2011), extraction temperature was the most important factor in THMs extraction from soft drinks.

Therefore, it has been challenging to develop and optimize various variables and different conditions to estimate maximum THMs extraction from water. Recently, various statistical experimental designs have been used for this purpose in water distribution networks (Rosero et al. 2012). Response surface methodology (RSM) is a useful technique for designing experiments, building models, and analyzing and optimizing effects of several independent variables. It also analyzes the relationship between independent variables and resulting responses (Rasheed et al. 2016). RSM combined with central composite design (CCD) is an efficient tool to study the simultaneous effect of various variables, which influence the responses, with a limited number of experiments by eliminating non-significant interactions of variables (Guimarães et al. 2008). Aguirre-González et al. (2011) optimized the extraction conditions for THMs from water samples in an RSM study using a composite 2⁵ factorial design. It was found that the extraction temperature and desorption times are the most influential conditions in the process.

The present study was designed to investigate the incidence of chlorination by-products (CBPs), mainly THMs, in an institutional drinking water distribution network in Islamabad, Pakistan. Objectives of the study are: comparison of headspace (HS)-SPME method and liquid-liquid extraction (LLE) technique to achieve maximum THMs extraction from water; optimization of analytical conditions using RSM and CCD for THMs determination; and subsequently, application of the optimized method for THMs detection and quantification in drinking water distribution networks using gas chromatography.

Standard analytes (iodoform, chloroiodomethane, chloroform, dibromochloromethane, bromodichloromethane and bromoform) and solvents were purchased from Sigma Aldrich (USA) and Merck (Germany), respectively, with 99% purity. The SPME fiber (75 μm Car-PDMS) was obtained from Supelco (USA).

The water treatment plant within the institution mainly consisted of underground tanks for water storage. From here water is pumped to overhead reservoirs for further distribution to all the filtration plants throughout the campus. Prior to distribution, water is treated with chlorine on a regular basis and its concentration is also monitored regularly to ensure safe water quality to consumers.

The samples collected from the drinking water source and consumer end within the university (Table 1) were analyzed for physicochemical contamination (free chlorine, UV₂₅₄, pH, total dissolved solids (TDS), dissolved oxygen (DO), turbidity, electrical conductivity (EC), alkalinity, hardness, etc.) which showed that all the parameters were within USEPA and World Health Organization (WHO) limits. It was observed generally that free chlorine concentration at source was in the range 0.5–1.5 mg/L, whereas at consumer taps it ranged from 0.23 to 0.46 mg/L, which lies within the optimum range prescribed by WHO.

Sampling was conducted from the main water reservoir and consumer taps from the National University of Sciences and Technology (NUST), H-12 campus premises. Samples were collected from each site in duplicate (Table 1). Freshly prepared ascorbic acid solution (0.142 M) was added to each 40 mL vial as a chlorine quenching agent prior to sampling. Samples were analyzed as per standard methods (APHA 2012).

A THM stock solution of 1,000 μg/L was prepared in methanol as per EPA Method 551.1 (USEPA 1995). Working standard solutions were prepared to obtain linear calibration curves and detection limits. For spiking THMs standard solution, carbon tetrachloride (CCl₄) solution was prepared in methanol (1,000 μg/L) to obtain a reproducible chromatogram (Figure 1).

THMs analysis was conducted using a Shimadzu 2010 gas chromatography system with fused silica capillary column (30 m × 0.32 mm × 1 μm) equipped with an electron capture detector (ECD). Initial oven temperature was 50 °C, which was then increased at a rate of 15 °C/min to 200 °C. The constant flow of helium carrier gas was maintained at 4 mL/min.

Distilled water (30 mL) was placed in a glass vial, then THMs standard (10 μL), Na₂SO₄ salt and CCl₄ (internal standard) were added. The sample was stirred at 300 rpm and the SPME fiber was injected into the headspace at 50 °C. The fiber was retracted back and transferred without delay to the GC injection port at 220 °C (Allard et al. 2012).

THMs standard (10 μL) was added to 35 mL distilled water followed by Na₂SO₄ salt and t-butyl methyl ether (MtBE) as extraction solvent. The vial was sonicated for 5 min, and 500 μL of the organic layer formed was transferred into a gas chromatography (GC) vial containing 10 μL of CCl₄ (internal standard). The extract was then injected into a GC column for analysis (Allard et al. 2012).

Conventional LLE-GC-ECD and HS-SPME-GC-ECD techniques were compared using unpaired t-test in order to achieve maximum response. The results indicated an increase in peak areas by using HS-SPME as compared with LLE (Figure 2).

The recovery efficiencies were also calculated for these methods to evaluate the method performance for THMs. According to USEPA, percent recoveries must fall in the range of 70 to 120 for THMs. The HS-SPME gave acceptable recovery values for all THMs (Table 2). As a result, HS-SPME is proposed as a reproducible, faster and more accurate technique for THMs analysis compared with LLE. Cancho et al. (1999) determined the recovery values close to 100% by spiking the water samples at a concentration of 10, 1.5 and 0.5 mg/L with I-THMs.

Distilled water was fortified with THM (1,000 μg/L) and salted with 1 g of Na₂SO₄ and NaCl separately, whereas the control was without salt. The results showed that an increase in salt concentration resulted in high THMs extraction. It can be related to the fact that the salt addition amplified the ionic potency of the solution and resulted in dispersion of analytes into the headspace (Takamatsu & Ohe 2003). As shown in Figure 2, Na₂SO₄ was observed to have a considerable effect on the THMs extraction as compared with NaCl and control. Therefore, use of Na₂SO₄ for THMs analysis was preferred over NaCl due to fewer impurities (USEPA 1995).

Accuracy of the HS-SPME technique was evaluated by plotting calibration curves. An acceptable linear range with regression coefficients (R²) higher than 0.93 was obtained for all THMs, which correlated with the findings of Stack et al. (2000), who indicated 0.9920 to 0.9959 R² value at THMs concentration ranging from 10 to 160 mg/L. The validity of HS-SPME technique was estimated in terms of limit of quantification (LOQ) and limit of detection (LOD) (Table 3). The LOQs ranged between 4 ng/L for CHI₃ and 68 ng/L for CHCl₃. Repeatability and reproducibility values were found to be less than 11%. The results demonstrated that the proposed HS-SPME technique is appropriate for measuring THMs at μg/L levels in drinking water.

Design of experiments and statistical analysis were conducted using software package DESIGN-EXPERT (trial version 9, Stat-Ease, Inc., MN). The full factorial CCD with 30 experiments was applied to optimize the level of effective variables such as salt amount, extraction temperature, extraction and desorption time for visualizing the significant THMs extraction conditions. Table 4 lists the ranges and levels of applied parameters by RSM-CCD.

Interaction between extraction temperature and extraction time was observed in a 3D response surface revealing synergetic effects of both variables for THMs (Figure 3(a)). This can be attributed to the fact that increasing the extraction temperature increases the diffusion of the analytes to the fiber surface. Consequently, the time necessary to reach the equilibrium of partition between the sample and extractor phase is reduced (Dos Santos et al. 2011). In addition, here diffusion coefficients in both water and headspace are higher; thus, diffusion of volatile analytes from aqueous phase to headspace is enhanced. Therefore, with increase in temperature during adsorption period, an increased THMs extraction rate was observed. Similar results were also reported by Deok-Hee et al. (2003).

While observing the effect of extraction time on THMs extraction in Figure 3(a), an increase in THMs extraction (z-axis) was observed with increased extraction time (B: y-axis, extraction time). The highest extraction was observed at 30 min. Here, extraction time was evaluated as an important parameter that influences partition of analytes, as HS-SPME is an equilibrium process that involves separation of analytes from aqueous phase to headspace and eventually into the fiber (Pawliszyn 1997). Acceptable equilibrium was attained for all THMs at 30 min in the present study.

The impact of desorption time (D: x axis, desorption time) and salt on the THMs extraction was evaluated in Figure 3(b). It is evident that addition of more salt resulted in higher THMs extraction. This could be attributed to the fact that the salt addition increases the ionic strength of matrix and decreases the solubility of analytes so that more analytes are dispersed into the headspace, thereby contributing to enhanced adsorption on the fiber (Takamatsu & Ohe 2003). Longer extraction time synergistically led to higher THMs extraction; desorption time of 8 min was sufficient to desorb analytes in the GC port (see Figure 5). Similar behavior for high molecular weight compounds has been reported previously (Cancho et al. 1999; San et al. 2007).

The coefficient for AB was found to be 37.92, higher than for factors C, BC and BD. Thus, the order of significance is listed as AB>BC, BD>AC, AD, CD. The standardized effects of these factors and their interactions on the THMs extraction were investigated by Pareto chart analysis. It is evident from Figure 4 that the factors AB (salt–extraction time) are the most influential factors affecting THMs extraction (p < 0.01) by the HS-SPME technique, followed by C (extraction temperature), BC (extraction time–extraction temperature) and BD (extraction time–desorption time).

Process optimization is an important step in determining values of factors for which response is at a maximum (Rasheed et al. 2016). Based on the variables selected (Table 4), numerical optimization was performed by the RSM-CCD to achieve one or more points in the factors domain that would maximize the THMs extraction. A desirability value (D) closer to 1 is considered to be significant by the RSM software. It was observed that maximum extraction efficiency was obtained when temperature was maintained at 80 °C with an extraction time of 30 min and 3.25 g salt at 8 min of desorption time at a D value of 1.0 (Figure 5).

The HS-SPME technique optimized by the RSM software was then employed for determination of THMs from the water samples of an educational institution in Islamabad, Pakistan. Samples were collected and analyzed as per standard protocol. The concentrations of total and iodinated THMs are shown in Figure 6(a) and 6(b), respectively. For TTHMs (total trihalomethanes) among all sites, sites L2A, L3A, CNM, MRC, TW8A and SMME had high concentration of TTHMs as shown in Figure 6(a). The respective chromatographic peaks for TTHMs and I-THMs can be observed in Figure 7(a) and 7(b), respectively. Figure 7 shows clearly identifiable chromatographic peaks from sites L2A and MRC. The large peak signal of chloroform showed a high content of chlorine present at these sites available to react with organic matter, which resulted in high THMs yield at sites L2A and MRC. This high concentration of TTHMs could be attributed to the presence of high UV₂₅₄ absorbance, TDS and residual chlorine. Some 88% of sites exceeded the standard values, while the highest concentration was observed to be 455.9 μg/L at site L2A. UV₂₅₄ absorbance is the indicator of NOM in water, which is one of the most significant precursors of THMs development (Singer 1999; Chang et al. 2001). At all the sampling sites, chloroform was detected in the highest ratio (Figure 6(a)), with maximum concentration of 233.4 μg/L observed at site L3A, while I-THMs were detected in approximately 85% of the samples as shown in Figure 6(b). Chloroiodomethane was the dominant species, found in 79% of the tested samples with highest mean value of 101.1 μg/L at MRC, while on the remaining eight sites, levels exceeded the threshold values of 0.2–5 μg/L. However, iodoform was detected in lowest concentration, ranging from 0.012–0.433 μg/L in 45% of the samples, whereas in other sites it was within the threshold values.

Another reason could be attributed to the close occurrence of sampling sites to the chlorination source, as more residual chlorine was available to react with the precursor UV₂₅₄ absorbance, yielding high concentrations of TTHMs and I-THMs. Similar findings were also reported by Bergamaschi et al. (1999) and Karanfil et al. (2002).

A correlation between THMs concentration, residual Cl₂ and UV₂₅₄ absorbance was assessed by regression analysis, keeping in view the findings of Chang et al. (2001) and (Singer 1999) as mentioned above.

The results showed a strong positive linear correlation between residual Cl₂ and UV₂₅₄ concentration with THMs formation, with R² = 0.80 for both (Figure 8). These results are in accordance with literature (Chowdhury et al. 2007). Hence, it proves that the potential reason for THMs contamination in drinking water was the presence of NOM and residual Cl₂.

The present study was designed to quantify THMs in drinking water through an optimized HS-SPME technique by using GC. The outcomes of this research work are as follows.

The physical and chemical parameters (pH, EC, temperature, UV₂₅₄ absorbance, residual Cl₂, TDS, turbidity, DO etc.) of drinking water samples meet the permissible limits recommended by WHO.

The HS-SPME and LLE techniques were compared to achieve the maximum THMs response. The results showed significant (p < 0.1) increase in peak areas for HS-SPME, which is an excellent alternative extraction technique comparable to LLE.

HS-SPME technique was optimized using RSM-CCD for THMs determination. Optimum conditions for THMs extraction were 30 min extraction time at 80 °C with addition of 3.25 g Na₂SO₄ salt and 8 min of desorption time.

The optimized method was used to determine THMs in institutional drinking water samples, which revealed THMs presence in 90% of the samples, with 30% exceeding the USEPA limit, indicating the possibility of adverse public health risks such as cancer, reproductive disorders, taste/odor problems, organoleptic issues and consumer complaints. As THMs are more common in the public water systems, they are a threat to any water supply that uses chlorine; thus, a large population may be affected by this contamination, ultimately putting pressure on population dynamics.

Results revealed a strong correlation of THMs formation with UV₂₅₄ concentration (R² = 0.8) and residual Cl₂ (R² = 0.8).

The validity of the optimized HS-SPME technique was further investigated by linear range, detection limits, precision and recovery efficiency for each analyte. The results verified that this technique is suitable and applicable for drinking water analysis.

Keeping in view the need and significance of the current study, there follow some future recommendations for undertaking further research in this field: (1) epidemiological and genotoxicity studies of THMs exposure to human cells/blood may be carried out to identify toxic levels using comet assay or various other techniques; (2) different methods of chlorinated disinfection by-products (C-DBPs), mainly THMs, removal and control may also be investigated in detail to minimize the THMs formation in drinking water sources, as they are potential human carcinogens.

Furthermore, a few mitigation measures to control DBPs formation and ultimately provide safe drinking water to consumers are: granular activated carbon (GAC) adsorption may be used to remove NOM, the major precursor of DBPs; multiple alternative drinking water treatment processes, including pre-ozonation, conventional treatments (coagulation/sedimentation, pre/post-sand filtration), ozone biological activated and carbon advanced treatment, may also be investigated, depending upon the available resources and need of the particular area; and finally, new approaches may be developed to effectively control THMs formation in chlorinated drinking water by targeting intermediate aromatic halogenated DBPs.

The authors acknowledge the financial support provided by the National University of Sciences and Technology (NUST), Pakistan to pursue the research work and providing the opportunity and facilities to undertake the research. The authors are very grateful to the construction and maintenance (C & M) staff of NUST for their support during the sampling process.