Ozonation of natural organic matter and aquatic humic substances: the effects of ozone on the structural characteristics and subsequent trihalomethane formation potential

Natural organic matter (NOM) is a complex and heterogeneous mixture of organic compounds with varying functional and structural characteristics (Lamsal et al. 2011). It consists of particulate organic matter and dissolved organic matter (DOM); with DOM typically measured in terms of dissolved organic carbon (DOC). DOC, the poorly defined complex and heterogeneous mixture of organic compounds with varying functional and structural characteristics, is separated into two major classes: humic and non-humic. Humic substances (HSs) are defined as a diverse group of organic materials resulting from the degradation of plant and animal tissues (Yu et al. 2005). The composition of HS is site specific, strongly impacted by the local environment, and influenced by parameters such as water temperature, alkalinity, and pH (Chow et al. 2005). HSs form the major fraction of DOC in surface waters, representing 40–80% by some estimates (Thurman 1985; Zhao et al. 2006). Our recent studies of the DOC composition in two major rivers in southern Manitoba, the Red River and Assiniboine River, have found them to be rich in aquatic HSs (AHSs) (Sadrnourmohamadi et al. 2013; Sadrnourmohamadi & Gorczyca 2015a, 2015b). AHSs consist of humic acids (HAs), which are compounds insoluble at a pH of less than 2 and fulvic acids (FAs), which are soluble at all pH levels (Zhao et al. 2006; Rodríguez & Núñez 2011). HAs are characterized by their aromatic structure with extended conjugation and can be described as a heterogeneous polyfunctional polymer with a polymeric network that can entrap volatile organic compounds (Kerc et al. 2003). FA comprises the dominant fraction of HSs with lower molecular weight and lower levels of aromaticity, as compared to their HA counterparts (Amy et al. 1986; Matilainen et al. 2011; Brezinski & Gorczyca 2019a). Recently, increased attention has been paid to the removal of the HS (FA and HA) fraction of DOC, since they are known to be the main precursors toward the formation of carcinogenic trihalomethanes (THMs) (Kerc et al. 2003; Zhao et al. 2006; Brezinski & Gorczyca 2019b).

Ozonation has been reported to be an effective treatment for THM reduction in drinking water, even without a significant decrease in the initial total organic carbon content (TOC). In these cases, ozone transforms high molecular weight organic compounds into smaller ones, which are resistant to reactions involving chlorine as a disinfectant (Camel & Bermond 1998; Galapate et al. 2001; Chiang et al. 2002; Black & Bérubé 2014; Mao et al. 2014). However, several studies have demonstrated cases where ozone had a deteriorative effect on THM reduction (Toor & Mohseni 2007; Chowdhury et al. 2008). It has been reported that DOC composition and distribution in natural waters determine the efficiency of ozone in terms of the reduction in THM precursors (Camel & Bermond 1998; Galapate et al. 2001). Moreover, more recent studies have investigated several other parameters associated with THM precursors, such as NOM polarity (Yan et al. 2010), which are impacted prior to ozonation. Conventional ozonation has even been shown to be preferable over more advanced techniques such as electro-peroxone pretreatment, on waters with high NOM and high UV₂₅₄ absorbance (Mao et al. 2018). This is in line with others who note that the hydrophobic fraction of NOM is mainly targeted by ozone, contributing almost twice as much to THM formation potential (THMfp) as the hydrophilic fraction (Chiang et al. 2009).

The reactions following the application of ozone on natural waters, especially those with high levels of DOC and calcium hardness as found in Canadian Prairies waters, are especially complex. Moreover, the heterogeneous and complex structure of DOC makes ozone treatment extremely challenging. Therefore, fundamental studies tracking the effects of different parameters, such as ozone dose and DOC type under controlled laboratory conditions, are needed. Studies reporting similar experimental conditions have tested DOC levels as high as 7.8 mg L⁻¹ (Chiang et al. 2009), with other studies not exceeding 3.2 mg L⁻¹ DOC in their raw and synthetic waters (Yan et al. 2010; Deeudomwongsa et al. 2017; Mao et al. 2018). Additionally, the work of Yan et al. (2010) was the only such study that looked at alkalinity exceeding 75 mg L⁻¹ as CaCO₃. The transformation in the functional characteristics of NOM after ozonation at varying ozone doses needs to be further investigated for high DOC waters, as it will lead to a better understanding of how ozone impacts DOC concentration, and consequently, THM formation for prairie waters.

The effect of ozone on the structural composition of THM precursors depends on several parameters such as DOC composition and ozone dosage. This study investigated the effects of ozone on THMs precursors of NOM and both HA and FA separately. The results were also linked to the change in the formation potential of THMs after ozonation.

The effect of ozone on the functional and structural properties of each source of organic compound was assessed by different physical (DOC and ultraviolet absorbance at 254 nm (UV₂₅₄)) and spectroscopic techniques (Fourier transform infrared coupled to attenuated total reflectance (FTIR-ATR), ultraviolet/visible (UV/Vis), and fluorescence spectroscopy).

Humic species are found to be the dominant fractions of DOC in two Manitoba surface waters (Goss & Gorczyca 2012; Goss et al. 2017). It has been reported that 50–70% of Manitoba potable water treatment plants that use these surface water sources are not in compliance with THM regulations set by provincial regulators. Therefore, the results of this study would be beneficial for local drinking water treatment operators and industry alike, in order to help meet requirements set by provincial regulators.

Synthetic water samples prepared in the laboratory were used in this study. Three different commercial organic compounds, including Suwannee River NOM, Suwanee River HA, and Suwanee River FA standards – provided by the International Humic Substance Society (IHSS) (St. Paul, Minnesota, USA) – were used to simulate the three different types of organic compounds. These standards, considered as the reference material, were selected because of their widespread use in the literature (Averett et al. 1994; Leenheer et al. 1995; Ratpukdi et al. 2009); and their ability to act as representative surrogates for NOM based on their acidity content (Ritchie & Perdue 2003; Adusei-Gyamfi et al. 2019) and size and conformation (Baalousha et al. 2006).

As a result, this work can be compared with those found in the literature and with future studies. Synthetic water was prepared by dissolving NOM, HA, and FA IHSS standards in deionized water and diluting to approximately 15 mg L⁻¹ with the pH adjusted to 7.0 – a typical DOC and pH range of source water used by Manitoba water treatment plants. The solution was stirred overnight and then filtered through a 0.45 μm nitrocellulose filter paper to remove any residual, non-dissolved NOM.

Ozonation was conducted on a semi-batch basis by bubbling the desired amount of ozone through the 2 L water sample following the method described in Sadrnourmohamadi & Gorczyca (2015a). Ozone was generated by a laboratory generator (Model OZO 1VTTL, Ozomax Ind., Canada); this unit is employing the corona discharge method using dehumidified atmospheric air as the feed gas to generate ozone. All experiments were conducted at a constant flow rate of 0.23 L/min, and ozone dosage was adjusted by changing the ozonation time.

Ozone output from the generator and ozone concentration in the reactor off-gas were determined by the iodometric method (APHA 2012), while the indigo method was used for ozone measurement in the aqueous phase (Bader & Hoigne 1981). Nitrogen gas was bubbled into the water sample after each ozonation runs for at least 10 min to purge the unreacted ozone gas.

Throughout this study, ozone dose refers to the transferred ozone dose (the difference of the mass of applied ozone and the mass of ozone in the off-gas). The most commonly used ozone doses in drinking water pretreatment are <1.0 mg ozone/mg DOC (Agbaba et al. 2004; Bose & Reckhow 2007). This work investigates ozone doses of 0.2, 0.6, and 1.0 mg ozone/mg DOC.

The effectiveness of ozonation was evaluated by measuring DOC and UV₂₅₄ values. Water samples were filtered through a 0.45 μm nitrocellulose filter paper, and DOC concentrations were measured using a total organic carbon analyzer (Phoenix 8000, Tekmar Dohrmann, USA). UV₂₅₄ of the filtered samples were measured using an Ultrospec 2100 pro UV/visible spectrophotometer, (GBC Scientific Equipment, Australia) with a 1 cm quartz cell.

THM formation potential measurements were conducted according to Standard Methods 5710B (APHA 2012), with further details of this procedure having been described elsewhere (Sadrnourmohamadi et al. 2013; Sadrnourmohamadi & Gorczyca 2015a). 50 mg L⁻¹ of sodium hypochlorite was added to each filtered sample to ensure that there was a sufficient residual amount of chlorine available to react with all organics present in the water sample. All samples were buffered to a pH of 7 using a phosphate buffer. Sample vials were sealed with TFE caps and were kept in the dark at 4 °C for 7 days. After 7 days, all samples were found to have >1.0 mg L⁻¹ (3–5 mg L⁻¹) free chlorine, which shows that all THM formation reactions were driven to completion. An alteration to the 5710B procedure was the chlorinated water incubation temperature. In this study, chlorinated water samples were incubated at 4 °C to simulate the water temperature in distribution systems, which is representative of the cold Canadian Prairie region.

THMs concentrations were determined with a liquid–liquid extraction method according to Standard Methods 6232B (APHA 2012). All experiments were run in triplicate to ensure experimental reproducibility.

UV/Vis spectra of water samples were recorded on an Ultrospec 2100 pro UV/visible spectrophotometer, (GBC Scientific Equipment, Australia) using 1 cm cuvettes in the 190–800 nm wavelength range.

All FTIR-ATR spectra were recorded on freeze-dried samples using a Bruker Alpha FTIR spectrometer (Bruker, MS, USA) equipped with ALPHA's Platinum ATR single reflection diamond ATR module, recorded in the 4,000–500 cm⁻¹ region.

Fluorescence spectra in the synchronous scan mode were obtained by a Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies, CA, USA). Synchronous scan excitation spectra were acquired by recording the fluorescence intensity, while simultaneously scanning over both the emission (λ_em) and excitation (λ_ex) wavelengths and keeping a constant optimized wavelength difference (offset: Δλ = λ_em − λ_ex) (Chen et al. 2003; Rodríguez et al. 2014a). Synchronous excitation spectra can improve peak resolution and reduce overlapping interferences. It also can identify the effect of ozone on structural signatures such as intensity, position, and shift of the fluorescence peak for each sample (Uyguner & Bekbolet 2005a, 2005b; Zhang et al. 2008). Synchronous scan spectra were recorded in the excitation wavelength range of 360–650 nm using the Δλ = 44 nm bandwidth between the excitation and emission monochromators. Although most studies on HS use a synchronous fluorescence of Δλ = 18–20 nm, some researchers have reported that an offset value of 44 nm acts as a powerful tool for DOC characterization before and after oxidation with chlorine dioxide and ozone (Swietlik & Sikorska 2004; Zhang et al. 2008). In this study, both the excitation and emission slits of the instrument were set at a 10 nm bandwidth, and a 600 nm/min scan speed was used.

Figure 1 shows the effect of ozonation on the removal efficiencies of DOC and UV₂₅₄, for all water samples. Generally, increased ozone dosage resulted in an increased reduction of DOC and UV₂₅₄ values for all water samples. However, HA water samples show the highest reduction of these parameters compared to NOM and FA standards. For this sample, 1 mg ozone/mg DOC resulted in a 42% reduction of DOC and a 96% reduction in UV₂₅₄, respectively_. Generally, more available ozone molecules and hydroxyl radicals are produced as ozone dosage increases – leading to an increase in organic carbon removal (Ratpukdi et al. 2010). Chang et al. (2002) had investigated the effect of ozone on HA removal at different ozone doses, in which they reported a 47% DOC reduction at an ozone dosage of 1.5 mg ozone/mg DOC. Tuhkanen et al. (1994) and Molnar et al. (2012) have both shown DOC reduction ranged between 5% and 25% under typical ozone doses of 0.4–1 mg ozone/mg DOC. Pryor et al. (2002) have also reported up to a 25% DOC reduction at lower applied ozone doses of 0.3–0.5 mg ozone/mg DOC. Water samples used in these studies were natural waters; however, our reported DOC removal values can still be comparable with these published works.

UV₂₅₄ is a parameter that indicates the unsaturated and/or aromatic carbon content of water samples (Edzwald & Tobiason 1999). For all the water samples studied, UV₂₅₄ values decreased following ozone application. This can be due to oxidation of unsaturated and/or aromatic carbon molecules, which is in accordance with previously reported results (Gilbert 1988; Camel & Bermond 1998; Swietlik et al. 2004; Sadrnourmohamadi & Gorczyca 2015a). Figure 1 shows a higher reduction in UV₂₅₄ values compared to DOC. This indicates the oxidation of a large portion of conjugated species by ozone, even with a low reduction in DOC. (Chang & Singer 1991; Chang et al. 2002; Ratpukdi et al. 2010). This is indicative of a targeting of aromatic, conjugated species rather than aliphatic groups.

Table 1 indicates the results of total THMfp (TTHMfp) for different water samples pre- and post-ozonation. In Table 1, it is clear that chloroform (CH₃Cl) is the major detected species of THMs since NaOCl was used for chlorination. No brominated species were detected since the simulated water samples did not contain bromide. Results indicate that prior to ozonation, HA yielded a higher TTHMfp, followed by NOM and FA standards. Upon ozonation, this value reduces significantly, with the highest reduction (89%) observed for HA water samples (Table 1). A lower ozone dose accomplished a relatively significant reduction in TTHMfp; however, by increasing the ozone dose, a small incremental reduction was achieved. This indicates that ozone can eliminate a large portion of the precursors that promote THM formation quickly, while it has a lesser effect on other portions of the precursors under these experimental conditions (Amy et al. 1986). As shown in Figure 1, ozone can oxidize/decompose aromatic, UV absorbing moieties of NOM resulting in a higher reduction of THMfp relative to DOC, which is consistent with previous studies (Amy et al. 1986, 1991; Galapate et al. 2001). Improved reduction of TTHMfp at higher ozone doses is related to an increased reduction of HS as the main precursor to form THMs (Rodríguez & Núñez 2011). HAs are reported to have high levels of aromaticity, and consequently, a high UV absorbance (Amy et al. 1986). As reported, UV₂₅₄ removal can be related to TTHMfp (Uyak & Toroz 2007). Figure 2 shows a plot of UV₂₅₄ removal against TTHMfp data from this study. A good correlation of R² = 0.82 was achieved between UV₂₅₄ removal and TTHMfp reduction.

Figure 3 shows the UV/Vis spectra of each water sample before and after a series of ozone applications. All non-ozonated samples show quite similar UV/Vis absorption spectra with a featureless decrease in absorption intensity with increasing wavelength. Due to the complex structure of organic compounds in each water sample, the UV spectra did not exhibit many distinctive features. However, in the case of non-ozonated samples, a shoulder in the wavelength range of 260–290 nm indicates the presence of π-π* electron transitions from polycyclic aromatic hydrocarbons, benzoic acids, and phenolic substances (Fukushima & Tatsumi 2001). It can also be related to general unsaturated and aromatic compounds, attributed to the broader conjugated structure (Chin et al. 1994; Purmalis 2013). Moreover, UV₂₈₀ provides important clues regarding the extent of humification, degree of aromaticity, and possible estimations of molecular weight (Chin et al. 1994). Figure 3(b) shows that non-ozonated HA has the highest UV₂₈₀ absorbance values compared to those recorded by NOM and FA. This behavior is expected since HA contains a relatively high amount of aromatic and functional groups, and hence UV absorbing moieties. The difference decreases with an increase in wavelength (Chen et al. 2002).

UV spectra of ozonated water samples showed that UV absorbance decreased with increasing ozone dosage. The shoulder in the range of 260–290 nm for non-ozonated water samples disappeared as a result of ozone application. Also, the absorptivity at 280 nm decreased with increasing ozone dosage. This is to be expected since ozone can remove or oxidize the moieties responsible for UV absorbance in this wavelength range (Rositano et al. 2001). Ozonation is found to target NOM at all wavelengths, with a sharp decrease at shorter wavelengths which is in agreement with the previous work (Rositano et al. 2001; Ho et al. 2013). The absorptivity of ozonated water samples approaches zero within the visible range (400–700 nm).

The effect of ozonation on NOM, HA, and FA samples was also studied by synchronous scan spectra recorded at Δλ = 44. Figure 4 shows the spectra of non-ozonated samples, as well as ozonated samples at 0.2, 0.6, and 1 mg ozone/mg DOC doses. The three non-ozonated samples showed a well-defined maximum peak at λ = 380–400 nm and a small shoulder-like peak at λ = 260. The non-ozonated HA sample exhibited a fluorescence peak with half intensity compared to the FA sample, which is in agreement given the higher molecular weight of HA. The intensity of the fluorescence peak is affected by the molecular structure of NOM and decreases with an increase in the molecular weight of the analyzed organic matter fractions (Chen et al. 2003; Swietlik & Sikorska 2004; Rodríguez et al. 2014a). Although HA is known to be rich in aromatic character, its low fluorescence intensity can also be related to HA inter-, and intra-molecular bonding, and re-absorption of the emitted radiation by other absorption centers of humic macromolecules (Miano et al. 1988; Senesi 1990; Chen et al. 2003). Moreover, the peak of the maximum intensity of HA occurred at a higher excitation wavelength compared to FA. According to Rodríguez et al. (2014a, 2014b) the maximum peaks located at higher wavelengths are indicative of aromatic or conjugated unsaturated systems, characteristics that fit well with HA. Peaks of maximum intensity at lower wavelengths are related to a low degree of aromatic and conjugated chromophores and are in agreement with the typical characteristics of the FA. Some additional small sub-peaks also can be seen at an excitation λ of 270–290 nm and 270–290 nm for HA and FA, respectively. These peaks are indicative of the polydispersive structure of HSs (Rodríguez et al. 2014a). Upon ozonation, the fluorescence intensity of the maximum peak as well as the whole spectra decreased for all samples. This can be attributed to a lower degree of aromaticity and a lower level of conjugation (Uyguner & Bekbolet 2005a; Zhang et al. 2008; Rodríguez et al. 2014a).

A protein-like peak at a shorter wavelength of 260 nm with low intensity increased when 0.2 mg ozone/mg DOC ozone dosage was applied. According to the literature, this peak can be attributed to aromatic amino acids (Coble 1996; Zhang et al. 2008; Rodríguez et al. 2014a). By increasing the ozone dose, the intensity of this peak decreased significantly, even below the non-ozonated sample due to the extensive structural degradation of the humic macromolecule at high ozone dosages (Rodríguez et al. 2014b). The results obtained in this study can be compared to those achieved by Zhang et al. (2008) (ozone dose = 1 mg ozone/mg TOC and synchronous spectra recorded at Δλ = 66 nm). A reported increase in the fluorescence of the protein-like peak was followed by a decrease in the fluorescence of the maximum peak of the spectrum for both FA and HA extracted from river water samples. The disappearance of these characteristic peaks, which is accompanied by a reduction in DOC, UV₂₅₄, and TTHMfp, can be attributed to the destruction of aromatic moieties (Uyguner & Bekbolet 2005a, 2005b).

The FTIR-ATR spectra of HS and FA samples obtained pre- and post-ozonation are shown in Figure 5. The change in functional group composition after ozone application can be the main case for THM reduction (Chang et al. 2002). All the spectra for the two non-ozonated HSs (HA and FA) were similar. The absorption bands of all samples were interpreted based on the literature (Senesi et al. 1989; Silverstein & Webster 1998).

A broad peak at the 3,700–3,200 cm⁻¹ region is generally assigned to O–H groups, and the shoulder present in the 3,000–2,800 cm⁻¹ region can be attributed to aliphatic CH, CH₂, CH₃ stretching. The peak in the range of 1,720–1,700 is indicative of carbonyl groups (C = O stretching mainly due to carboxylic acids and ketones). The peak at 1,637 cm⁻¹ is a typical peak for aromatic C = C vibration and C = O conjugation within the aromatic ring. The peak at 1,380–1,400 cm⁻¹ is assigned to C–H deformation of aliphatic and CH₃ groups. In addition, a peak in the range of 1,150 cm⁻¹ is indicative of C–O stretching and O–H deformation of COOH (Li et al. 2008). The C–O stretching of alcoholic compounds, esters, ethers, and phenols is shown at 1,066 cm⁻¹. Also, aromatic C–H bending can be observed at 850 cm⁻¹. Carboxylic character (an O–H band at 3,400–3,300 cm⁻¹ and a C = O band at 1,730–1,700 cm⁻¹) was found for all samples. The O–H peak (3,700–3,200 cm⁻¹) is more intense for HA samples than for FA which is consistent with the literature (Senesi et al. 1989). HA shows a relatively higher absorption intensity at ∼1,620 cm⁻¹ and 2,920 cm⁻¹. These results suggest that the untreated HA sample is richer in aromatic compounds and aliphatic CH₂/CH₃ groups than the FA sample, which is consistent with our UV/Vis spectroscopy analysis. After 0.2 mg ozone/mg DOC application, the peak area of the O–H band decreases. Further increase in ozone dosage results in the removal of this peak and the shoulder located at 2,900 cm⁻¹ (attributed to C–H functional groups). A reduction is also observed for aromatic C = C vibration and C = O conjugation attached to an aromatic ring (1,600 cm⁻¹). Moreover, in HA samples, the absorbance of aromatic C–H bending (850 cm⁻¹) was reduced. In summary, FTIR results indicate that after ozonation, O–H, C–H, and CO/OH peaks are still present; however, the peak prominence of aromatic C = C and C = O conjugation with aromatic rings (1,600 cm⁻¹) and O–H (3,200 cm⁻¹) decrease remarkably. The degradation of these functional groups, which is more significant for the HA samples – in line with UV/Vis spectral data (Figure 3) – likely accounts for the TTHMfp reduction following ozonation (Chang et al. 2002; Kanokkantapong et al. 2006).

In this study, the composition and chemical characteristics of NOM and isolated HSs, HA, and FA (provided by the IHSS), before and after ozone application were studied by means of FTIR-ATR, UV/Vis, and synchronous scanning fluorescence (SSF) measurements. The main objective of this study was to show the effect of ozone oxidation on the structural and chemical composition of three types of aquatic organic standards. The results were also linked to the effects of the ozonation process on reducing THMfp of water samples with high levels of THM precursors present. The following conclusions can be made from this study:

• 1.

  Ozonation at a dose of 1 mg ozone/mg DOC was capable of reducing DOC, UV₂₅₄, and THMfp by up to 42%, 95%, and 89% for the HA water standard – the main THM precursor – respectively.

• 2.

  The results of the UV/Vis, FTIR-ATR, and SSF work reveal a consistent trend showing that ozone can alter the composition of DOC in the water samples, causing a significant reduction in aromaticity. The reduction of UV₂₅₄ for each ozonated sample also reaffirms the notion that ozone mainly targets aromatic moieties present in organic carbon molecules.

• 3.

  FTIR-ATR results showed that the reduction of unsaturated functional groups, such as aromatic rings and C = C bonds in the water standards, are the main targets of ozone application.

• 4.

  SSF results also revealed that ozonation decreases the fluorescence intensity of the maximum peak as well as for the whole spectra. Ozonation reduced the concentration of aromatic and highly conjugated compounds in the water. This is the main reason why ozonation is effective in reducing water THMfp levels, as they contain the bulk of highly reactive THM precursors.