Effects of raw water quality on the adsorptive removal of 2-methylisoborneol by powdered activated carbon under non-equilibrium conditions

Taste and odor (T&O) events in drinking water related to 2-methylisoborneol (2-MIB) and geosmin, both of which are produced by algae and bacteria, have become a frequent problem worldwide (Devi et al. 2021). A nationwide survey on Japanese drinking water-treatment plants performed during 2010–2012 revealed that approximately 36% of drinking water-treatment plants suffered from T&O problems (Kishida et al. 2015). These compounds do not affect human health; however, T&O events result in public concern regarding human effects from exposure to drinking water. Drinking water-quality standards in Japan require maintaining the concentrations of 2-MIB and geosmin below 10 ng/L.

It is generally difficult to remove 2-MIB and geosmin through physical water-treatment processes, such as coagulation, sedimentation, and filtration. In addition, chlorination, which is the final disinfection process during drinking water production in Japan, is not effective for the removal of these compounds. Therefore, advanced water-treatment processes such as oxidation treatment including ozonation and activated carbon adsorption are essential for removing these compounds (Faruqi et al. 2018; Ma et al. 2019; Fakioglu et al. 2020; Cerón-Vivas et al. 2022).

Powdered activated carbon (PAC) is widely used for the successful removal of these compounds. Regarding the adsorptive removal of 2-MIB and geosmin from raw water, it is necessary to address the problem associated with the competitive adsorption between these compounds and natural organic matter (NOM). NOM is inevitably present in raw water, and some fractions of NOM have strong inhibitory effects on the adsorptions of 2-MIB and geosmin onto activated carbon (Newcombe et al. 1997a, 1997b; Matsui et al. 2012, 2013; Inoue et al. 2020; Wang et al. 2020). When the concentrations of NOM with high inhibition potential increase, certain countermeasures are required in drinking water-treatment plants. In most cases, the PAC dose is increased during periods of high inhibitory organic matter concentrations; however, such an increase in PAC dose imposes additional operating costs. Developing appropriate strategies for determining the optimal PAC dose is important for the sustainable operation of drinking water-treatment plants that suffer from T&O problems associated with 2-MIB and geosmin.

To develop such strategies mentioned earlier, accurate monitoring of the concentration of inhibitory organic matter is required. Various reports have indicated that certain fractions of NOM are more likely to cause inhibition through competitive adsorption. By investigating the molecular weight fractionation using ultrafiltration, Newcombe et al. (1997a, 1997b) indicated that the smallest NOM fraction, which had less than 500 Da molecular weight, caused the greatest 2-MIB and geosmin adsorption inhibition. Matsui et al. (2012, 2013) reported that the NOM responsible for the competitive inhibitions of 2-MIB and geosmin is the chromophoric organic matter with less than 230 Da molecular weight; they also estimated that such organic matter accounted for less than 2% of the NOM. The importance of the low molecular weight organic matter which adsorbs ultraviolet in the competitive adsorption has also been reported by Wang et al. (2020). Inoue et al. (2020) observed the greatest correlation between 2-MIB removal rates in non-equilibrium adsorption processes and the scores associated with fluvic acid-like organic matter after performing parallel factor (PARAFAC) analysis on results obtained through an excitation–emission matrix (EEM). These findings indicate that it is critical to establish an adequate water-quality index representing the behavior of a specific organic fraction with a high competitive adsorption potential.

Research has revealed the characteristics of organic fractions with competitive adsorption, while continuous monitoring of such fractions should optimize the PAC adsorption process (e.g., through selecting optimum PAC doses). However, from the viewpoint of water practitioners, continuous monitoring of costly parameters may be suboptimal. Therefore, it is important to elucidate easily monitored water-quality indices, in which the behaviors of organic compounds significantly contributing the competitive adsorption are reflected. By using such indices, the competitive adsorption potentials of NOM can be reasonably estimated.

Apart from the characteristics of organic matter involved in the inhibition, contact time for adsorptive removal by PAC is one of the other important factors. Expanding a contact time is beneficial for improving the adsorptive removal by PAC. Most of the previous studies mentioned earlier investigated the inhibitory effects of NOM under equilibrium conditions (Newcombe et al. 1997b; Matsui et al. 2012, 2013; Wang et al. 2020). In real drinking water-treatment facility, however, information on the inhibitory effects under non-equilibrium conditions is particularly important. This is because that sufficient contact times to achieve equilibrium conditions are not adopted in many existing drinking water-treatment plants. In Japan, many drinking water-treatment plants cannot ensure a long contact time (i.e., >1 h) with PAC. The PAC doses are limited owing to effluent regulations in drinking water plants. Therefore, understanding the relationship between water-quality indices and the removal efficiencies of T&O compounds, such as 2-MIB and geosmin, by PAC treatment during a short contact time (i.e., <1 h) is essential for the effective removal of these compounds by PAC. Studies have reported contradictory results regarding the indices representing NOM with high competitive adsorption potential under non-equilibrium conditions. Zoschke et al. (2011) reported that the difference in the intensity of inhibition of the 2-MIB and geosmin adsorptions onto PAC could not be explained by the difference in dissolved organic carbon (DOC) concentrations; in contrast, correlations between the total organic carbon (TOC) or DOC concentrations and 2-MIB removal rates were reported by Inoue et al. (2020). Such contradictions indicate the need for further investigations on adequate water-quality indices, by which the intensity of competitive adsorption under non-equilibrium conditions can be reasonably estimated.

The main objectives of this study were to investigate the relationships between the removal rates of T&O compounds by PAC and water-quality indices in raw water under non-equilibrium conditions and to propose water-quality indicators that can be used for predicting the removal of T&O compounds by PAC in real drinking water-treatment plants. We collected raw water from three existing drinking water-treatment plants, and the relationships between the characteristics of such raw water and intensity of inhibition were investigated. With regard to the T&O compounds examined, the experiments were only performed for 2-MIB as PAC generally removes 2-MIB less effectively than geosmin (Lalezary et al. 1986). On the basis of the results obtained, approaches for predicting inhibition of 2-MIB adsorption onto PAC are discussed.

Taking the seasonal variations in raw water qualities into consideration, the experiments with a constant sampling interval should be continued for a long period including both summer and winter. On this basis, we collected raw water from water-treatment plants twice a month from June 2020 to March 2021 from a dam lake in the Tohoku area (Plant A), the lower reaches of the river in the Kanto area (Plant B), and the lake in the Kansai area (Plant C), Japan. In these water-treatment plants, T&O problems in drinking water related to 2-MIB and geosmin are frequent. During collection, only odor problems related to geosmin occurred.

To attribute the difference in the experimental results solely to the difference in the raw water quality, the experimental conditions other than the raw water were fixed identical in the PAC adsorption experiment. Stock 2-MIB solutions (1.0 mg/L) were prepared by diluting a standard 2-MIB solution (100 mg/L, Kanto Chemical Co., Inc.) with ultrapure water. The stock 2-MIB solution was spiked into the raw water samples (final 2-MIB concentration: 100 ng/L). PAC adsorption experiments were conducted using 50 mL of spiked 2-MIB samples in capped glass tubes. The PAC used in this study had been dried at 110 °C for 3 h prior to the experiment and stored in a desiccator until the use. Dried PAC (5LPD, Swing Corporation, Japan), which conforms to the standards set by the Japan Water Works Association and is widely utilized in existing Japanese drinking water-treatment facilities, was suspended in ultrapure water using a stirrer. Stirred PAC suspensions were added to spiked 2-MIB samples (final PAC concentration: 10 mg/L (Inoue et al. 2020)). Ultrapure water was added to spiked 2-MIB samples instead of the PAC suspension as a control sample. These suspensions were shaken at 150 rpm for 30 min under room temperature (20 °C) and then filtered through a 0.45 μm polytetrafluoroethylene (PTFE) syringe filter (Membrane Solutions Limited, USA). As mentioned previously, we investigated the intensity of inhibition under non-equilibrium conditions. Time-course changes in 2-MIB removal rate in a PAC adsorption experiment using raw water collected from Plant A are provided in Figure S1. The results presented in Figure S1 indicate that 2-MIB adsorption did not apparently reach equilibrium conditions at a contact time of 30 min.

2-MIB concentrations were measured using solid-phase micro-extraction (SPME) coupled with gas chromatography-mass spectrometry (GC–MS) (Table 1). 2,4,6-Trichloroanisole-d3 (TCA-d3) standard (Kanto Chemical Co., Inc., Japan) was added to each sample as an internal standard (100 ng/L), while the standard 2-MIB range was from 0.5 to 100 ng/L.

Turbidity was measured using a turbidity meter (PT-200; Nittoseiko Analytech Co., Ltd, Japan). Polystyrene latex (PSL) was used as the turbidity standard. The electrical conductivity and pH were measured using a Horiba Laqua pH/conductivity/ion meter (F-73; Horiba, Japan). The ultraviolet absorption of water at a wavelength of 254 nm (UV254) was measured using a spectrophotometer (UV-1800; Shimadzu, Japan). The TOC concentration was determined using a TOC analyzer (TOC-L; Shimadzu, Japan). EEM spectral analysis was conducted using an Aqualog (Horiba, Japan). Measurement and normalization of EEM were performed according to Gilmore & Cohen (2013). The EEMs were generated by scanning the excitation wavelengths from 239 to 800 nm at 3 nm intervals, with 2.33 nm increments of the emission wavelengths from 244.36 to 824.45 nm. The EEM of ultrapure water was subtracted from that of each sample, and the EEMs were normalized to quinine sulfate units using 0.05 mol/L quinine sulfate monohydrate in H₂SO₄.

Correlation coefficients between water-quality indices and 2-MIB removal rates were calculated. Two-sided p-values were considered significant at p < 0.05.

PARAFAC analysis was conducted using the Solo software (Eigenvector Research Offers). Each dataset for PARAFAC analysis in the three water sources was composed of 54 samples of raw water and water treated by PAC. The accuracy of excitation wavelengths from 240 to 250 nm is low (Yamashita & Tanoue 2003). Therefore, EEM at these wavelengths were excluded in the PARAFAC analysis.

Table 4 summarizes the major results of the correlation coefficients between the water-quality indices and 2-MIB removal rates. Unfortunately, there was no significant correlation (|R| > 0.7, p < 0.05) between the water-quality indices and 2-MIB removal rates when the correlations were evaluated for the overall experimental period. The correlations between the 2-MIB removal rates and water-quality indices were also evaluated separately for two shorter periods (i.e., high and low water-temperature periods based on median water temperature). In Plants A and B, there is no significant correlation between water-quality indices and 2-MIB removal rates during both high and low water-temperature periods. At Plant C, the negative correlations between TOC and 2-MIB removal rates during the high and low water-temperature periods were significant (i.e., |R| > 0.70, p < 0.05); however, the correlation between UV254 and 2-MIB removal rates was evident only in the high water-temperature period.

Table 5 and Figure S2 summarize the correlation coefficients between the 2-MIB removal rates and FRI values. With regard to the correlation during the overall experimental period, significant correlations were found between FRI values determined for Regions III and VI and 2-MIB removal rate; among them, the absolute value of the correlation coefficient between 2-MIB removal rates and FRI values determined for Region III was the highest (i.e., R = −0.58). Nevertheless, the absolute values of all the correlation coefficients were below 0.7, indicating that no FRI values were strongly correlated with the 2-MIB removal rates during the overall experimental period, even at Plant C. FRI values determined for Region VI were somewhat difficult to interpret because information on the representative compounds in this region was limited. For the other treatment plants, significant correlations were not found between FRI values and 2-MIB removal rates for the overall experimental period at Plants A and B.

The correlation coefficients determined separately for the high and low water-temperature periods are also presented in Table 5. At Plant C, significant correlations were only apparent during the high water-temperature period; FRI values determined for Regions III, IV, VI, VII, and VIII were significantly correlated with the 2-MIB removal rates. In Plant A, significant correlations were not found during both high and low water-temperature periods. At Plant B, only the FRI values of Region VI during the low water-temperature period were significantly correlated with the 2-MIB removal rates, although this result is somehow difficult to interpret as mentioned earlier. Such correlations were not discoverable by evaluating the correlation for the overall experimental period (Table 5).

The PARAFAC components differed among the three plants (Figure S3, Table S1). At Plant A, two different components were identified: one aromatic protein II and one humic acid-like organic matter. At Plant B, three different components were identified: one soluble microbial product (SMP)-like organic matter, one fulvic acid-like organic matter, and one humic acid-like organic matter. At Plant C, three different components were identified: one SMP-like organic matter, one fulvic acid-like organic matter, and one AOM (i.e., chlorophyll and phycocyanin).

Table 6 summarizes the correlation coefficients between PARAFAC scores and 2-MIB removal rates. At Plant C, only the PARAFAC scores of component 3 were significantly correlated with the 2-MIB removal rates during the entire experimental period, although the value of correlation coefficient was relatively low. The trends of correlation coefficients between PARAFAC scores and 2-MIB removal rates were different between the high and low water-temperature periods at Plant C. During the high water-temperature period, PARAFAC scores on components 2 and 3 were significantly correlated (i.e., |R| > 0.70, p < 0.05) with 2-MIB removal rates. Conversely, no correlation between PARAFAC scores and 2-MIB removal rates was apparent during the low water-temperature period. At Plants A and B, no significant correlation was found between PARAFAC scores and 2-MIB removal rates.

NOM is known to hamper the adsorptive removal of 2-MIB by PAC. Inoue et al. found a negative correlation between the TOC concentration and 2-MIB removal rate during PAC experiments using raw water collected from 21 drinking water-treatment plants (Inoue et al. 2020). This negative correlation was verified also by this study but was limited to the cases of Plant C during both the high and low water-temperature periods. The other TOC concentration datasets did not exhibit significant correlations with 2-MIB removal rates, indicating that the potential of inhibiting the adsorption of 2-MIB differs depending on the characteristics of the organic matter, as pointed out by Zoschke et al. (2011). In the present study, the significant correlation between UV254 and the 2-MIB removal rates was only found during a high water-temperature period at Plant C. These results indicate that it is difficult to universally evaluate the effects of competitive adsorption using comprehensive indicators of organic matter, such as TOC and UV254. Analysis of the characteristics of the organic fractions is required for evaluating the effects of competitive adsorption.

In the present study, we evaluated the applicability of FRI and PARAFAC scores for monitoring the intensity of competitive adsorption. During the overall experimental period, a significant correlation was only found between the FRI values determined for Region III at Plant C. With regard to the PARAFAC analysis, most of the PARAFAC scores did not significantly correlate with 2-MIB removal rate during the overall experimental period, except component 3 at Plant C. However, correlation trends between the EEM-related indices (i.e., FRI values and the PARAFAC scores) and 2-MIB removal rates were different between the high and low water-temperature periods (Figures S4 and S5). These results clearly indicate that the impact of organic matter reflected in each fluorescence region differs depending on water temperature; therefore, separate evaluation of the correlation with the 2-MIB removal rate is important. As can be seen in Figures S4 and S5, the relationships between 2-MIB removal rates and many water-quality indices were highly scattered. These results suggest that the impacts of the changes in such water-quality indices on the intensity of inhibition of 2-MIB adsorption were limited in comparison with the other indices. This fact emphasizes the importance of elucidating adequate water-quality indices representing the intensity of inhibition, which is one of the main objectives of the present study.

Significant correlations were evident between the EEM-related indices, namely the FRI values and the PARAFAC scores, and 2-MIB removal rates during the high water-temperature period at Plant C. With regard to FRI, the values determined for Regions VI exhibited the strongest correlation, followed by Regions IV, VIII, VII, and III. Representative compounds for Regions IV (i.e., SMP-like organic matter), VII (i.e., AOM dominated by chlorophyll), and VIII (i.e., AOM dominated by phycocyanin) were associated with biological activity. Conversely, as mentioned earlier, representative compounds for Region VI have not been fully elucidated at present. Therefore, the discussion on the potential characteristics of organic matter causing significant inhibition based on the results on the Region VI is difficult, although the FRI values determined for this region had the strongest correlation with the 2-MIB removal rate. The PARAFAC scores of components 2 and 3 were also significantly correlated with the 2-MIB removal rate. The components 2 and 3 were assumed to have SMP-like organic matter and AOM (phycocyanin), respectively. Cyanobacteria produce phycocyanin; at high water temperatures, the cyanobacteria Dolichospermum spp. were found in the water sources of Plant C. The significant correlation between these indices and 2-MIB removal rates suggests that the behaviors of the dominant inhibitory compounds are closely related to biological – especially algal – activity. These findings indicate that FRI and PARAFAC analyses are useful for evaluating competitive adsorption in lakes with algae bioactivity. The FRI values determined for Region III reflect the existence of fulvic acid-like organic matter. Fulvic acid-like organic matter has also been identified as a major compound with high competition potential (Inoue et al. 2020). Interestingly, a significant correlation between the FRI values determined for Region III and 2-MIB removal rates was only apparent during the high water-temperature period at Plant C. This fact suggests that the characteristics of fulvic acid-like organic matter would also be determining the inhibition potential.

During the high water-temperature period at Plant A, the correlation coefficients between the FRI values determined for Regions V and VII and 2-MIB removal rates were relatively high, although the correlations were not significant. A similar relationship was also seen between the PARAFAC score of component 1 and 2-MIB removal rates. These relationships may be attributed to the insufficient number of analyses, which is one of the limitations of the present study. By increasing the data set, it is highly possible that a significant correlation between these indices and 2-MIB removal rates could be established. The FRI values of Region V appeared to be related to the PARAFAC scores of component 1. Humic acid-like organic matter affected the 2-MIB removal by PAC during the low water-temperature period. The FRI values of Region VII revealed the presence of AOM, including chlorophyll. This result suggests that an increase in algal activities negatively affects the 2-MIB removal. In addition to terrestrial plants, aquatic organisms including algae could also be sources of humic acid-like organic matter. Although it is difficult to distinguish their source from the results of the EEM analysis, it is possible that the behavior of humic acid-like organic matter is also affected by the algal activities.

Neither the FRI values nor PARAFAC scores significantly correlated with 2-MIB removal rates at Plant B during both the high and low water-temperature periods except the FRI values determined for Region VI during the low water-temperature period. Because the information on the organic compounds in this region is limited at present as mentioned earlier, representative organic compounds involved in the inhibition of 2-MIB adsorption onto the activated carbon were not well characterized. Several wastewater-treatment plants are located upstream of Plant B; therefore, organic matter contained in the effluent from wastewater-treatment plants also determines to a certain extent the characteristics of NOM in the raw water of Plant B. Our results suggest that the investigated water-quality indices may not be suitable for evaluating the effects of organic matter derived from wastewater-treatment plants on competitive adsorption.

PARAFAC analysis has been widely applied for interpreting results obtained from EEM analyses; it is a powerful tool for attributing the EEM spectra changes to several organic matter groups and therefore facilitates the more informative results from EEM analysis. Conversely, the FRI analysis is certainly advantageous compared to the PARAFAC analysis, since it can interpret results even when the number of available results is limited; in extreme cases, FRI analysis can be performed even for one result of EEM analysis. In the present study, the maximum number of components identified was three, mainly because of the limited number of EEM analyses. In such situations, information on the entire EEM spectrum can be reasonably interpreted by performing an FRI analysis.

As already discussed, determining FRI values in the regions in which the behavior of AOM is reflected would provide useful information on the status of organic matter with a high potential for inhibiting 2-MIB adsorption on activated carbon. The fact that the EEM analysis is capable of detecting organic matter associated with photosynthesis (e.g., chlorophyll) has made it applicable in other research fields, such as food chemistry (Tan et al. 2000; Guimet et al. 2004). However, in water environment research, the regions in which chlorophyll and other organic matter associated with algal activities (i.e., emission range greater than 600 nm) are seldom included in EEM analyses, even in studies investigating changes in Chlorophyll a concentrations (Singh et al. 2010; Yamashita et al. 2017). Including regions reflecting the behavior of organic matter associated with algal activities in both PARAFAC and FRI analyses, as performed in this study, may open a new door for detailed characterization of aquatic organic matter; it may be essential not only for improved management of competitive inhibition in the activated carbon adsorption process but also for other research fields associated with the water environment.

We investigated water-quality indices that can evaluate the inhibition intensity of the adsorptive removal of 2-MIB onto PAC by monitoring raw water in three drinking water-treatment plants for 10 months. We found no universal inhibition indicator that could explain the change in the 2-MIB removal rate during the overall experimental period using a single water-quality index. Dividing the experimental period into high and low water-temperature periods improved the correlation between 2-MIB removal rates and changes in several water-quality indices. This finding suggests that the dominant inhibitory constituents differ between high and low water-temperature periods. In lake water, biologically associated organic matter – especially AOM – evaluated by EEM analysis, exhibited a strong correlation with 2-MIB removal rates in certain situations. By expanding the region to be involved in the calculation of the excitation and emission wavelength ranges from 600 to 880 nm, both the PARAFAC analysis and FRI method were effective at detecting the variation in the organic fraction associated with algal activities. Our results highlight the importance of including the regions reflecting the behavior of organic matter associated with algal activities to expand the possibility of using water-quality indices and evaluate the intensities of inhibition in PAC adsorption processes.

We are very grateful to the employees of the three drinking water-treatment plants in Japan for their great efforts in providing raw water samples. This study was partially supported by a Health Labour Sciences Research Grant (H30-Kenki-Ippan-004, 21LA1004) from the Ministry of Health, Labour, and Welfare, Japan.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.