Oxidation of natural organic matter with processes involving O₃, H₂O₂ and UV light: formation of oxidation and disinfection by-products

Abstract

This study investigates the effects of UV photolysis, ozonation and different advanced oxidation processes (O₃/UV, H₂O₂/UV and O₃/H₂O₂/UV) on the oxidation of groundwater natural organic matter (NOM) and by-product formation. Although the investigated treatments only slightly reduce the total organic carbon content (4–15%), the NOM character was changed significantly. The fulvic acid fraction decreased and the content of the hydrophilic acid fraction increased in ozone treated water and even more noticeably in water treated by O₃/H₂O₂/UV. All treatments led to significant increases in polar oxidation by-products such as aldehydes (up to 8 times) and carboxylic acids (up to 34 times), with no clear relationship between the changes in concentrations of these by-products and the addition of H₂O₂ and the UV dose. Statistical analysis showed a good correlation between carboxylic acids with ozone applications and carboxylic acids and UV₂₅₄. Trihalomethane and haloacetic acid formation potentials were reduced best (43% for THMFP and 68% for HAAFP) during the O₃/H₂O₂/UV process (0.5 mg O₃ per mg DOC; 10 mg H₂O₂ per L: 600 mL cm⁻²) using the lower UV dose, and were also well correlated (R = 0.847) during all water treatments. Bromate formation was observed only in the processes involving ozone.

---

1. Introduction

Natural organic matter (NOM) is a complex mixture present in most water supply sources. NOM is a heterogeneous mixture of naturally occurring components which originate from plants, animals and microorganisms, and from their degradation products. Recently, NOM concentrations in water sources have been observed to increase in many countries due to climate change, changes in soil acidification and more intensive rain events.¹ The molecular size and properties of NOM vary significantly depending on the water source and the time of year.²

Generally NOM significantly influences water treatment processes such as oxidation, coagulation and adsorption as well as the application of disinfectants.^1–3 A common drinking water treatment goal is to remove NOM as the major source of unwanted disinfection by-products (DBPs) precursors. To date, several hundred DBP species have been identified. Drinking water frequently contains the following types of DBPs in addition to the most abundant trihalomethanes (THMs) and haloacetic acids (HAAs): haloacetonitriles (HANs), haloketones (HKs), trichloronitromethane, N-nitrosamines, aldehydes and carboxylic acids, where many have been found to be hazardous to human health, in particular showing carcinogenic effects.^4–6

In order to identify the contribution and reactivity of different NOM fractions toward DBPs formation, different approaches and fractionation procedures using different resins have been developed, allowing the separation of hydrophobic humic and hydrophilic NOM fractions.^7,8 Additionally, polarity rapid assessment method is a novel approach using solid-phase extraction, and also shows promise for the isolation of some more toxic nitrogenous by-product precursors.⁹

In recent years, much drinking water supply research has focused on the use of advanced oxidation processes (AOP) for the control of organic compounds in water, in order to remove NOM and minimize DBPs formation in drinking water.^2,10 In AOP the hydroxyl radicals (˙OH) and other oxidative free radicals are generated, e.g. by the UV photolysis of oxidizing agents such as H₂O₂.^11,12

Previous studies reported that UV/H₂O₂ treatment can reduce NOM aromaticity, leading to smaller molecular size NOM and creating more biodegradable compounds.^12–15 Longer irradiation times combined with high H₂O₂ concentrations result in strong oxidation conditions capable of mineralizing NOM, which is reflected in decreasing total organic carbon contents.^13,16 During the UV/H₂O₂ process, the THM formation potential (THMFP) may be reduced, and the distribution of DBPs can be shifted towards more brominated species.^2,13,17 However, THMFP have also been observed by some authors to increase during the UV/H₂O₂ process.^18–20 This contrasting behaviour of NOM in water is likely due to differences in the process conditions (UV fluence and H₂O₂ dose) and the content and characteristics of the NOM present.

Ozone is commonly applied for oxidation and advanced oxidation processes, including O₃/UV, O₃/H₂O₂, O₃/H₂O₂/UV. Ozone causes the formation of low molecular weight oxygen-containing organic by-products due to the oxidative breakdown of complex natural organic matter. Aldehydes, carboxylic acids, benzoic compounds, ketones, keto-acids and bromate are all typical NOM oxidation by-products by ozone.^21–24 However, ozonation generally reduces the formation potential of hazardous chlorination by-products, such as THMs and HAAs due to the oxidation of their precursors.²⁵

This issue continues to be the focus of research, with the latest studies in this area indicating the importance of the formation of oxidation/disinfection by-products due to their toxicity and frequent detection in treated water.^6,22,26,27 Based on the literature review there is a lack of data which incorporates the application of different oxidation processes on the formation of a wide range of toxicologically important oxidation by products. Thus the aim of this study was to investigate the effects of the application of O₃, O₃/UV, O₃/H₂O₂/UV processes and UV photolysis on the formation of carboxylic acids, aldehydes, bromate, while trihalomethanes and haloacetic acids were measured after chlorination in water rich in NOM and with a high bromide content. The changes in the NOM structure during these treatments as a consequence of the different oxidation mechanisms was also investigated.

2. Materials and methods

2.1. Samples and reagents

For the water matrix, this investigation utilized groundwater from a 100 m deep well in Vojvodina, Republic of Serbia. THM (in methanol) and HAA (in methyl tert-butyl ether) standards were purchased in concentrations of 2000 μg mL⁻¹ from Supelco.

The THM standard contained chloroform (CF), bromodichloromethane (BDCM), dibromochloromethane (DBCM) and bromoform (BF), and the HAA standard contained monochloroacetic acid (MCAA), monobromoacetic acid (MBAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), bromochloroacetic acid (BCAA) and dibromoacetic acid (DBAA). Internal standards fluorobenzene (THM analysis, 2000 μg mL⁻¹ in methanol) and 1,2,3-trichloropropane (HAA analysis, 1000 μg mL⁻¹ in methyl tert-butyl ether) were also purchased from Supelco.

Standards for aldehydes (formaldehyde, acetaldehyde, glyoxal and methylglyoxal), bromide, acetate, formate and potassium oxalate were purchased from Fluka.

Bromate standard was purchased from Ultra Scientific. Solvents used were for organic residue analysis and were obtained from J.T. Baker. H₂O₂ (30% w/w, reagent grade) was purchased from POCH S.A.

The NOM fractionation was carried out on Supelite™ DAX-8 (Supelco) and Amberlite® XAD-4 (Fluka) resins. All other chemicals were analytical grade and were used without further purification.

2.2. UV photolysis

The photochemical reactor used for UV photolysis and the AOPs which included UV light was made of stainless steel. Inside the quartz reaction vessel at the centre of the 0.7 L reactor was a low pressure mercury lamp (Philips TUV 16W) with a maximum emission at 253.7 nm. The design of the reactor was presented in more detail in Molnar et al.²⁸ The applied UV doses for the UV experiments were 600 mJ cm⁻² and 3000 mJ cm⁻² in batch mode.

2.3. Ozonation and O₃/UV treatment

The raw water (RW) was ozonated in a 2 L glass column (85 mm diameter). Ozone generated electrochemically by a 1 g h⁻¹ Argentox ozone generator was introduced via a ceramic diffuser at the bottom of the column. The transferred ozone dose applied was 0.5 mg O₃ per mg total organic carbon (TOC), as commonly used during drinking water pretreatment and based on the determination of ozone demand of water. The residual ozone concentration in the water after treatment was 0.5 ± 0.08 mg L⁻¹.

For the O₃/UV treatment, ozonated water (as described above) was immediately subjected to the UV treatment in a photochemical reactor of stainless steel using 600 and 3000 mJ cm⁻² UV doses.

2.4. UV/H₂O₂ treatment

During the UV/H₂O₂ experiments, 10 mg H₂O₂ per L was added to the water prior to UV irradiation (UV doses of 600 and 3000 mJ cm⁻²). The UV doses applied were comparable with the literature,^13,15,19 and were varied to investigate their influence on the NOM characteristics and reactivity, in combination with H₂O₂.

2.5. O₃/H₂O₂/UV treatment

For the O₃/H₂O₂/UV treatment, ozonated water (Section 2.3.) was spiked with H₂O₂ solution to obtain 10 mg L⁻¹ in water and was subjected to the UV light (600 and 3000 mJ cm⁻²).

After each treatment, the water samples were analyzed for TOC content and UV₂₅₄ absorbance and the SUVA values were calculated. Formation of oxidation by-products such as aldehydes and carboxylic acids as well as THMFP and haloacetic acids formation potential (HAAFP) was also investigated.

The various applied treatment combinations are presented in Table 1. All processes were carried out in duplicate and the results are presented with standard deviations.

Table 1 Description of the water treatment conditions applied

Label	Treatment	Conditions
WT1	Ozonation (O3)	0.5 mg O3 per mg DOC
WT2	UV photolysis	600 mJ cm^−2
WT3	UV photolysis	3000 mJ cm^−2
WT4	O3/UV	0.5 mg O3 per mg DOC; 600 mJ cm^−2
WT5	O3/UV	0.5 mg O3 per mg DOC; 3000 mJ cm^−2
WT6	O3/H2O2/UV	0.5 mg O3 per mg DOC; 10 mg H2O2 per L; 600 mJ cm^−2
WT7	O3/H2O2/UV	0.5 mg O3 per mg DOC; 10 mg H2O2 per L; 3000 mJ cm^−2
WT8	H2O2/UV	10 mg H2O2 per L; 600 mJ cm^−2
WT9	H2O2/UV	10 mg H2O2 per L; 3000 mJ cm^−2

---

2.6. NOM fractionation

NOM fractionation was carried out, before and after the applied treatments, according to Mergen et al.:⁸ water samples were passed through glass columns (85 cm long, 2 cm diameter) filled with XAD resin, allowing both the hydrophobic fractions (humic and fulvic acid fractions) and hydrophilic fraction (hydrophilic acid and non-acidic hydrophilic fraction) to be isolated. Samples are filtered though a 0.45 μm filter and acidified to pH 2. 2 L of raw water is passed through the DAX-8 resin column. The adsorbed substances are eluted from the DAX-8 resin with 0.1 M NaOH (eluate is the hydrophobic fraction), after which the eluate is acidified to pH 1, left to settle, and centrifuged after 24 h. The supernatant represents the fulvic acid fraction (FAF). After centrifugation, the residue is dissolved in 0.1 M NaOH (HAF-humic acid fraction). After passing through the DAX-8 resin, the water sample is also passed through XAD-4 resin. Elution with 0.1 M NaOH yields the hydrophilic acid fraction (HPIA). The residual fraction after passing water through both resins is the non-acidic hydrophilic fraction (HPI-NA).¹⁰

2.7. Analytical methods

TOC contents in the water samples were analyzed by ElementarLiquiTOCII, using oxidation by platinum-catalysed combustion at 850 °C.

UV₂₅₄ absorbance was measured in a 1 cm quartz cell according to standard methods²⁹ on a CINTRA 1010, GBC Scientific Equipment spectrophotometer at a wavelength of 254 nm. The specific UV absorbance (SUVA, L mg⁻¹ m⁻¹) was calculated as:

The amount of ozone transferred to the water was calculated from the difference in the input and output ozone concentrations in the gas phase, measured by iodometric titration under standard conditions (273 K and 101.3 kPa).²⁹ The residual concentration of ozone in water was determined by standard indigo method.²⁹

THMFP and HAAFP were determined according to standard method.²⁹ Samples were dechlorinated after a seven day reaction period. Trihalomethanes were directly analyzed by purge and trap system (Tekmar 3100) coupled to a GC/MS (Agilent Technologies 7890A/5975C), based on USEPA method 5030B.³⁰ The practical quantitation limits for the THMs ranged from 0.34–0.74 μg L⁻¹. HAA precursor contents were measured according to USEPA Method 552.³¹ Analysis of HAAs was performed on a GC/μECD (Agilent 6890N) using a DB-608 column. The practical quantification limits for the HAAs were 0.30–2.80 μg L⁻¹.

Aldehydes (formaldehyde, acetaldehyde, glyoxal and methylglyoxal) in water were analysed after derivatization using O-(2,3,4,5,6-pentafluorobenzyl)-hydroxylamine reagent and extraction with hexane, in accordance with USEPA method 556.³² The practical quantification limits for aldehydes were 0.58–1.02 μg L⁻¹.

A Dionex ICS-3000 Ion Chromatography System was used to determine the bromide, bromate and anions of carboxylic acids (acetate, formate and oxalate) contents according to DIONEX Application Note 154.

pH was measured by portable instrument (WTW InoLab pH).

p- and m-alkalinity was measured in accordance with the standard method.²⁹

The statistical analysis was performed using the Design-Expert 9 (Stat-Ease) software.

3. Results and discussion

3.1. Raw water characteristics

The raw water investigated (general characteristics shown in Table 2) has a high NOM content (5.82 ± 0.39 mg C/L TOC). The values for UV absorbance at 254 nm and the specific UV absorbance (SUVA) are good indicators for NOM character, indicating the presence of both hydrophobic and hydrophilic NOM structures (3.96 ± 0.10 L mg⁻¹ m⁻¹). The total NOM content, based on the TOC, UV₂₅₄ and SUVA values, is also presented in Fig. 1, to better compare the characteristics of the raw and treated waters. NOM characterization according to the SUVA value is in compliance with the NON fractionation using XAD technique with a pair of resins. The results of the NOM fractionation in the raw and treated waters are presented in Fig. 2. The most abundant NOM fraction was the hydrophobic fraction of fulvic acid (75%), followed by similar amounts of the hydrophilic acid and non-acidic hydrophilic fractions, accounting for 12 and 13%, respectively.

Table 2 Characteristics of the raw groundwater^a

Parameter	Units	Mean value ± sd
a sd – standard deviation of 3 measurements.
pH	—	8.01 ± 0.10
Total alkalinity	mg CaCO3 per L	753 ± 24
TOC	mg L^−1	5.82 ± 0.39
UV254	cm^−1	0.231 ± 0.05
SUVA	L mg^−1 m^−1	3.96 ± 0.10
THMFP	μg L^−1	305 ± 44
CLFP		159 ± 25
BDCMFP		91.6 ± 24.7
DBCMFP		48.5 ± 2.9
BRFP		5.54 ± 2.9
HAAFP	μg L^−1	348 ± 53
DCAAFP		208 ± 59
TCAAFP		82.7 ± 11.1
BCAAFP		47.6 ± 9.0
DBAAFP		9.1 ± 4.5
Total aldehydes	μg L^−1	11.2 ± 2.5
Formaldehyde		6.59 ± 0.88
Acetaldehyde		2.61 ± 0.48
Glyoxal		1.48 ± 0.69
Methylglyoxal		0.549 ± 0.454
Carboxylic acids	μg L^−1	<10.0
Bromide	μg L^−1	623 ± 21

---

Fig. 1 Effects of water treatments on the total NOM content.

Fig. 2 Effects of ozonation and O₃/H₂O₂/UV treatment on changes in NOM characteristics compared to the raw water.

The high content of bromide ions in the raw water (623 ± 21 μg L⁻¹) increases the risk of brominated disinfection by-products and bromate formation during chemical oxidation. The hydrophobic fraction of NOM has a significant impact on the formation of DBPs, which is supported by the high THMFP (305 ± 44 μg L⁻¹) and HAAFP (348 ± 53 μg L⁻¹) values. The percentage contribution of individual disinfection or oxidation by-products was calculated based on their average content (given in Table 1) in relation to the total content of a particular group, such as THMs, HAAs and aldehydes. The dominant compounds from the group of THMs formed by chlorination of raw groundwater is chloroform accounting for 52%, while precursors of brominated THMs include DBCM (30%) followed by DBCM (16%) and BR (2%). The dominant compounds from the group of HAAs are chlorinated HAAs, particularly DCAA and TCAA (60 and 23%, respectively) followed by precursors of brominated HAA, BCAA and DBAA (14 and 3%, respectively). Precursors of MCAA and MBAA were not detected in the raw water. When it comes to the carbonyl compounds, the raw water has a total aldehydes content of 11.2 ± 2.5 μg L⁻¹, with formaldehyde dominating (59%), while acetaldehyde, glyoxal and methylglyoxal forms represent 23, 13 and 5%, respectively. Aldehyde formation could be a result of natural processes.⁶ The carboxylic acids content in the raw water were below 10 μg L⁻¹.

3.2. The effects of photolysis and oxidation treatments on the content and character of NOM in water

Changes in the NOM content based on the TOC, UV₂₅₄ and SUVA values are presented in Fig. 1. TOC reduction by the investigated water treatments (WTs) was 4–15% compared to the raw water. More noticeable reductions in NOM was observed for UV₂₅₄ (10–62%) and SUVA values (1–55%). Advanced oxidation using ozone combined with hydrogen peroxide and UV (WT7) was shown to be the most effective process for total NOM reduction. Addition of the UV light treatment and/or H₂O₂ oxidant slightly improved the degree of NOM oxidation removal (WT4–7) compared to ozonation alone. Based on the residual SUVA values, the hydrophilic character of NOM is increased after the oxidation treatments.

Full characterisation of NOM was carried out in the waters treated by selective oxidation by ozone (WT1) as well as ozone combined with H₂O₂ and UV (WT7), as the AOP treatments which achieved maximum NOM removals. Although ozonation did not result in a significant reduction in the total NOM content (only 7% TOC), ozone oxidation leads to the partial oxidation of the fulvic acid fraction and increases the amount of the hydrophilic NOM fractions compared to the raw water (Fig. 2).

In ozonated water (WT1), the share of FAF decreased by 19% while the contributions of HPI-A and HPI-NA increased by 10%. In addition, a greater degree of NOM oxidation was observed when ozone was combined with H₂O₂ and UV (WT7). Thus, the contribution of FAF decreased by 36%, and the contribution of HPI-A slightly increased by 7%, while HPI-NA increased by 29% relative to the raw water. It may be supposed that the main oxidation products of the fulvic acid fraction are hydrophilic polar compounds of low molecular weight.

A greater degree of NOM oxidation was achieved by the synergistic effects of O₃, H₂O₂ and UV combination (WT7) compared to ozone alone. It can be assumed NOM oxidation proceeds mainly via hydroxyl radical (HO˙) attack, whereby these radicals, formed by the decomposition of ozone and H₂O₂ in the presence of UV light, are stronger oxidation species than ozone alone.^10,33 In order to identify the by-products of NOM oxidation, aldehydes and carboxylic acids contents were followed after all applied treatments.

3.3. The effects of photolysis and oxidation treatments on the content and character of NOM in water

The effects of UV photolysis and the investigated oxidation treatments (based on ozone, H₂O₂ and UV) on the formation of aldehydes are presented in Fig. 3. It is evident that all water treatments led to an increase in total aldehydes content compared to the raw water (11.2 ± 2.5 μg L⁻¹). Water subjected to ozonation had significantly increased total aldehyde contents, about 7 times higher than the RW. In the ozone based AOPs, there is no clear relationship between addition of H₂O₂ and increasing UV dose and the resulting increase in aldehydes. Total aldehydes contents were very similar in water treated by O₃ (WT1), O₃/UV (WT4) and O₃/H₂O₂/UV (WT6) with lower UV doses, and were in the range 70.8–74.1 μg L⁻¹. Increasing in UV dose resulted in different behaviours in the O₃/UV and O₃/H₂O₂/UV processes, where in the first case total aldehydes dropped to 24.8 ± 5.5 μg L⁻¹ (WT5) while in the second case the maximum 90.6 ± 9.9 μg L⁻¹ (WT7) was observed (8 times higher than the RW). For UV photolysis (WT2,3) and the H₂O₂/UV process (WT8,9), increasing the UV dose definitely increases the yield of total aldehydes (up to 3 times higher).

Fig. 3 Effects of photolysis and oxidation treatments on the formation of aldehydes in water.

In water treated by all processes, the most abundant aldehyde was formaldehyde followed by glyoxal, acetaldehyde and methylglyoxal. Glyoxal and methylglyoxal are formed the most in the ozone-based AOPs, accounting for 20–50% of total aldehydes. The presented results are significant in highlighting the increased concentration of all investigated aldehydes after the advanced oxidation treatments based on ozone and H₂O₂. Glyoxal and methylglyoxal are known toxins and are classified as carcinogens or suspected carcinogens.^23,33 In addition, apart from the aldehydes, short-chained carboxylic acids are the most common by-products of ozonation.^27,34

Carboxylic acids formation caused by advanced oxidation and UV photolysis showed similar behaviour as the aldehydes (Fig. 4). For both the aldehydes and carboxylic acids, all oxidation treatments including photolysis alone resulted in an increase in both groups of oxidation by-products (8–34 times greater concentrations, depending on the treatment). Ozonation alone increased carboxylic acids content 20 times compared to the RW, to 322 ± 28.2 μg L⁻¹. The combinations of O₃ with UV (WT4 and WT5) additionally enhanced carboxylic acids concentrations (456–534 μg L⁻¹) compared to the ozonated water, while addition of H₂O₂ had no impact on further changes in the total carboxylic acids content (WT6 and WT7) compared to the O₃/UV treatment (449–510 μg L⁻¹).

Fig. 4 Effects of photolysis and oxidation treatments on the formation of carboxylic acids in water.

As previously noted, in the combination of H₂O₂ with UV (WT8,9) and UV photolysis alone (WT2,3), increasing the UV dose promotes aldehydes formation. Moreover, carboxylic acids were formed at much higher quantities than aldehydes, which is in accordance with the observations of Nawrocki et al.³⁵ Hammes et al.³⁶ also commented on aldehyde and ketone formation, suggesting that organic acids represent a large fraction of the assimilable organic carbon formed after ozonation. In addition, in our study it was determined that by favouring the oxidation mechanism with free hydroxyl radicals generated in the combination of ozone with UV light and H₂O₂, further increases in carboxylic acid contents resulted compared to ozonation alone.

3.4. Presence of THMs and HAAs in water by photolysis and oxidation treatments

The reactivity of residual NOM in AOPs treated water toward trihalomethanes formation is presented in Fig. 5 as trihalomethane formation potential. It is evident that the applied treatments resulted in significant changes in the THMFP, which were in the range 173–461 μg L⁻¹. Application of ozone led to slightly increased THM formation (by 22%) compared to the raw water. In ozone based AOPs (WT4–7), there is no observed relation between changing THMFP and increasing UV dose. In the O₃/UV process, slight reductions (17%) in THMFP were achieved by application of the higher UV dose. In contrast, using the O₃/H₂O₂/UV process, the lower UV dose (WT6) resulted in the lowest THMFP, with a reduction of 43%. Application of UV photolysis (WT2 and 3) did not change the content of THM precursors, while the combination of UV with H₂O₂ (WT8 and 9) led to an increase in THMFP compared to the raw water. The most abundant species of all the formed THM in the treated water were precursors of chloroform (up to 60%), followed by BDCM (up to 37%), DBCM (up to 29%) and BR (up to 7%).

Fig. 5 Effects of photolysis and oxidation treatments on the formation of trihalomethanes in water.

The contents of haloacetic acids precursors in the raw and treated waters is presented in Fig. 6. Significant changes in HAAFP were evident after the applied treatments, with final formation potentials in the range 113–460 μg L⁻¹. Oxidation by ozone lead to a 21% increase in HAAFP. Furthermore, apart from the O₃/UV treatment, in all the treatments which included UV application (WT2, 3, 6–9), a lower UV dose resulted in lower HAAFP (22–68% reductions), while the higher UV dose caused an increase in HAAFP compared to the raw water (4–32%). The best treatment for reducing HAAFP was the O₃/H₂O₂/UV treatment (WT6), removing 68% of the HAAFP compared to the raw water. The O₃/UV treatment (WT4, 5) also led to reductions in HAAFP (9–44%), but in this case, the higher removal was achieved at the higher UV dose.

Fig. 6 Effects of photolysis and oxidation treatments on the formation of haloacetic acids in water.

The given results indicate that for optimal removal of THM and HAA precursors, selective and non-selective NOM oxidation mechanisms must be applied together. It can be assumed that in AOPs involving the use of ozone: removal of THM and HAA precursors by O₃/UV and O₃/H₂O₂/UV processes initially occurred due to the selective oxidation by ozone, followed by the radical oxidation mechanism of specific precursor materials upon application of UV radiation (WT4, 5) and H₂O₂ combined with UV (WT6, 7). It can also be supposed that the formation of the optimal amount of hydroxyl radicals in the case of the O₃/UV process requires higher doses of UV radiation, while in the case of the O₃/H₂O₂/UV treatment, the synergistic effects of ozone, hydrogen peroxide and UV radiation mean the selective removal of THM and HAA precursors has already been achieved at lower doses of UV radiation. Further increases of UV dose in the O₃/H₂O₂/UV treatment initiated the formation of greater amounts of reactive radical species, leading to further oxidation of NOM, increasing the NOM hydrophilicity (indicated by SUVA and NOM fractionation results, Table 2) and the formation of new reactive precursors of THM and HAA.

Additionally, Wang et al.¹⁶ made similar observations, whereby the UV/H₂O₂ process led to an increase in THM and HAA formation in river water. Generally, the investigated treatments showed similar impact on the precursors of both HAA and THM. This observation is supported by the good correlation (R = 0.847) between HAAFP and THMFP normalized on TOC content (Table 4). Chin and Bérubé³³ indicate that ozone based AOPs (O₃/UV) are more effective in THMFP and HAAFP reduction than the ozone or UV treatment alone, which is similar to the results presented above, which additionally show that the efficacy also strongly depends on the UV dose and the additional application of H₂O₂. Dotson et al.¹⁸ suggested that the UV/H₂O₂ process leads to the THM yield increasing compared to the photolysis alone, which is also in agreement with the results given above.

3.5. The effects of photolysis and oxidation treatments on the bromate formation in water

Table 3 presents the results of bromate formation during the investigated treatments. It is evident that processes which do not involve the application of ozone do not lead to bromate formation. However, ozone based treatments cause bromate formation in concentrations 14.4–50.9 μg L⁻¹, which exceeds the regulated 10 μg L⁻¹ level.^37,38 Application of UV and/or H₂O₂ resulted in an increase in bromate formation compared to the ozonation alone (1.6–3.5 times).

Table 3 Effects of photolysis and oxidation treatments on the bromate formation

Treatment	Bromate (μg L^−1)
RW	<5.00
WT1	14.4 ± 2.7
WT2	<5.00
WT3	<5.00
WT4	50.9 ± 5.1
WT5	35.2 ± 4.3
WT6	31.3 ± 4.1
WT7	22.5 ± 3.9
WT8	<5.00
WT9	<5.00

---

Zhao et al.²⁶ indicated that during the UV/O₃ process bromate concentration could exceed levels compared to ozonation alone. They also suggest that the presence of humic acids at a certain level can inhibit bromate formation during the UV/O₃ process. In our study, the dominant fulvic acids fraction did not exhibit such an inhibitory effect toward bromate formation.

3.6. Insight into correlation between investigated parameters

Table 4 presents the results of the correlations between all the parameters, as obtained by the Design Expert 9 statistical software. A strong positive correlation (R > 0.8) was observed between ozone application and carboxylic acid formation (R = 0.981) while a strong negative correlation was observed between ozone application and UV₂₅₄ reduction (R = −0.974).

Table 4 Correlation between all investigated parameters

Correlated parameters	Ozone	Hydrogen peroxide	UV light	UV254	THMFP	HAAFP	Aldehydes	Carboxylic acids
Ozone	1.000
Hydrogen peroxide	0.000	1.000
UV light	0.000	0.000	1.000
UV254	−0.974	−0.120	−0.142	1.000
THMFP	−0.604	0.303	0.316	0.469	1.000
HAAFP	−0.485	0.036	0.528	0.344	0.847	1.000
Aldehydes	0.535	0.300	0.174	−0.601	−0.113	0.198	1.000
Carboxylic acids	0.981	0.030	0.072	−0.981	−0.470	−0.321	0.621	1.000

---

Consequently, a strong negative correlation was also observed between UV₂₅₄ and carboxylic acid formation (R = −0.981). These correlations may be explained by the selective attack of ozone on the double bonds in aromatic NOM structures, resulting in UV₂₅₄ reduction and the production of polar by-products such as carboxylic acids. In addition, the formation potentials of trihalomethanes and haloacetic acids also showed a good correlation (0.847) during the investigated treatments, which might indicate that both groups of DBPs share similar precursor compounds. No correlations were found between the other presented parameters.

4. Conclusions

This work considers the effect of different oxidation processes including UV light, O₃ and H₂O₂ on the oxidation of largely hydrophobic natural organic matter and by-products formation. The following conclusions can be made:

• The treatments applied do not significantly reduce total NOM content, but considerably change its structure. The higher degree of hydrophobic NOM oxidation into hydrophilic structure achieved using O₃/H₂O₂/UV advanced oxidation compared to ozonation suggests the more effective oxidation of NOM in the presence of hydroxyl radicals.

• The synergistic effect of UV, O₃ and H₂O₂ and the role of hydroxyl radicals formed during advanced oxidation had a different influence on the formation of polar by products and the reactivity of residual NOM towards chlorinated DBPs formation.

• Aldehydes and in particular carboxylic acids, generally reached a maximum in ozone based processes, but UV photolysis and H₂O₂/UV also increased their contents.

• THMFP and HAAFP increased or decreased depending on the treatment conditions. The maximal formation was measured after the H₂O₂/UV process while the maximal reduction of these parameters was achieved by the O₃/H₂O₂/UV process at the lower UV dose.

• Bromate was formed only during oxidation processes involving ozone.

• Statistical analysis show a good correlation between: (i) O₃ application and carboxylic acid formation, (ii) O₃ and UV₂₅₄; (iii) carboxylic acid and UV₂₅₄ and (iv) THMFP and HAAFP.

The given results indicate that the oxidation treatments investigated have a great potential for NOM oxidation, and could therefore be of great significance during the pre-treatment of water with a high NOM content. The choice of oxidation treatment can be used to control the emergence of specific oxidation by-products.

Acknowledgements

The authors gratefully acknowledge the support of the Provincial Secretariat for Higher Education and Scientific Research, Republic of Serbia, Autonomous Province of Vojvodina (Project No. 114-451-2263/2016).

References

1. S. A. Baghoth, Characterizing Natural Organic Matter in Drinking Water Treatment Processes and Trains, Dissertation, CRC Press, 2012.
2. R. Lamsal, M. E. Walsh and G. A. Gagnon, Water Res., 2011, 45(10), 3263.
3. J. Molnar, J. Agbaba, B. Dalmacija, A. Tubić, D. Krčmar, S. Maletić and D. Tomašević, Chem. Eng. J., 2013, 222, 435.
4. S. D. Richardson, Anal. Chem., 2012, 84, 747.
5. A. Tubić, J. Agbaba, B. Dalmacija, J. Molnar, S. Maletić, M. Watson and S. UgarčinaPerović, J. Environ. Manage., 2013, 118, 153.
6. M. Serrano, I. Montesinos, M. J. Cardador, M. Silva and M. Gallego, Sci. Total Environ., 2015, 517, 246.
7. E. H. Goslan, D. A. Fearing, J. Banks, D. Wilson, P. Hillis, A. T. Campbell and S. A. Parsons, J. Water Supply: Res. Technol.--AQUA, 2006, 51, 475.
8. M. R. D. Mergen, B. Jefferson, S. A. Parsons and P. Jarvis, Water Res., 2008, 42, 1977.
9. X. Liao, E. Bei, S. Li, Y. Ouyang, J. Wang, C. Chen, X. Zhang, S. W. Krasner and I. H. Suffet, Water Res., 2015, 87, 292.
10. J. Molnar, J. Agbaba, B. Dalmacija, M. Klašnja, M. Dalmacija and M. Kragulj, Sci. Total Environ., 2012, 425, 169.
11. Y. Liu, X. He, X. Duan, Y. Fu and D. D. Dionysiou, Chem. Eng. J., 2015, 276, 113.
12. M. M. Bazri, B. Barbeau and M. Mohseni, Water Res., 2012, 46(16), 5297.
13. R. Toor and M. Mohseni, Chemosphere, 2007, 66(11), 2087.
14. K. E. Black and P. R. Bérubé, Water Res., 2014, 52, 40.
15. H. R. Sindelar, M. T. Brown and T. H. Boyer, Chemosphere, 2014, 105, 112.
16. G.-S. Wang, C.-H. Liao, H.-W. Chen and H.-C. Yang, Environ. Technol., 2005, 27(3), 277.
17. W. Buchana, F. Roddick and N. Porter, Chemosphere, 2006, 63(7), 1130.
18. A. D. Dotson, V. S. Keen, D. Metz and K. G. Linden, Water Res., 2010, 44(12), 3703.
19. D. H. Metz, M. Meyer, A. Dotson, E. Beerendonk and D. D. Dionysiou, Water Res., 2011, 45(13), 3969.
20. H.-C. Lin and G.-S. Wang, Desalination, 2011, 270(1–3), 221.
21. W. J. Huang, G. C. Fang and C. C. Wang, Sci. Total Environ., 2005, 345, 261.
22. A. Papageorgiou, D. Voutsa and N. Papadakis, Sci. Total Environ., 2014, 481, 392.
23. C. Liu, X. Tang, J. Kim and G. V. Korshin, Chemosphere, 2015, 125, 182.
24. J. Agbaba, J. Molnar, A. Tubić, M. Watson, S. Maletić and B. Dalmacija, Water Sci. Technol.: Water Supply, 2015, 15(1), 75–83.
25. G. Hua and D. A. Reckhow, Water Res., 2013, 47, 4322.
26. Y. Zhao, F. Xiao, D. Wang, M. Yan and Z. Bi, J. Environ. Sci., 2013, 25(11), 2207.
27. B. Jurado-Sánchez, E. Ballesteros and M. Gallego, Water Res., 2014, 51, 186.
28. J. Molnar, J. Agbaba, A. Tubić, M. Watson, M. Kragulj, S. Rončević and B. Dalmacija, Water Sci. Technol.: Water Supply, 2015, 15(1), 34.
29. AWWA-APHA-WEF, Standard Methods for the Examination of Water and Wastewater, American Public Health Association/American Water Works Association/Water Environment Federation, 20th edn, Washington, DC, 2012.
30. USEPA, Method 5030B, Purge and Trap for Aqueous Samples (Revision 2), 1996.
31. USEPA, Method 552, Determination of haloacetic acids in drinking water by liquid-liquid extraction, derivatization, and gas chromatography with electron capture detection, Environmental Monitoring Systems Laboratory Office of Research and Development U.S. Environmental Protection Agency, Cincinnati 45268, 1990.
32. USEPA, Method 556, Determination of carbonyl compounds in drinking water by pentafluorobenzylhydroxylamine derivatization and capillary gas chromatography with electron capture detection (Revision 1), National Exposure Research Laboratory Office of Research and Development U.S. Environmental Protection Agency, Cincinnati, Ohio 45268, 1998.
33. A. Chin and P. R. Bérubé, Water Res., 2005, 39, 2136.
34. A. Dąbrowska, B. Kasprzyk Hordern and J. Nawrocki, Global NEST J., 2005, 7(1), 61.
35. J. Nawrocki, J. Świetlik, U. Raczyk-Stanisławiak, A. Dąbrowska, S. Biłozor and W. Ilecki, Ozone: Sci. Eng., 2015, 25(1), 53.
36. F. Hammes, E. Salhi, O. Köster, H. P. Kaiser, T. Egli and U. von Gunten, Water Res., 2006, 40(12), 2275.
37. European Union. Official J. European Commun. L 330: Directive 98/83/EC, 1998.
38. Official paper SRJ 42/98-4, Serbian Regulation for Drinking Water, 1998.

---