Four electrolysis treatment systems were tested under three salinity conditions—i.e., seawater, brackish water, and fresh water—at a type approved land-based facility. Each test was conducted separately. For the start test (ballasting), 500 tons of test water were used as control (250 tons) and treated water (250 tons). The end test (deballasting) consisted of discharging water from the control and treated water tanks after a predetermined storage time. The challenge water was prepared by adding glucose as a dissolved organic carbon (DOC) source, starch as a particulate organic carbon (POC) source, silicates as total suspended solids, and organisms (microorganisms, phytoplankton, and zooplankton) to 500 tons of natural water under appropriate salinity conditions according to the IMO D-2 regulations for type approval tests. Supplementary Table S2 lists the required conditions for the challenge water according to the Ballast Water Performance Standards. Table 1 presents the specific parameters of the water used and detailed information on the land-based tests conducted in this study.

Two types of electrolysis treatment were performed: i) an indirect side-stream method involving electrolysis of seawater to produce high-concentration HOCl, which disinfects by diluting the ballast water; ii) an in-stream electrolysis treatment method involving electrolysis of the entire ballast water and treatment with radicals, potential difference, and low-concentration HOCl. In both types, the total residual oxidant (TRO) concentration, which is the sum of free oxidants (such as Cl₂) and combined chlorine, was maintained at 10 mg/L. The two types of electrolysis treatments performed in this study are shown in Figures 1A, B.

To evaluate the generation characteristics and behavior of DBPs in treated ballast water, the IMO G9 requires evaluation of the changes in DBPs according to storage time. In this study, the occurrence and pattern of DBPs were investigated for 120 h after electrolysis treatment, assuming 5 days of ship operation. The samples were collected from the challenge and treated water immediately after electrolysis (ballasting process, day 0), 24 h later (day 1), and 120 h later (day 5) in each salinity zone (seawater, brackish water, and freshwater). On day 5, treated samples were collected both before and after neutralization (referred to as N-treated water) in the deballasting mode. Neutralized treated water was mixed with an appropriate amount of neutralizer to remove residual TROs to attain the maximum allowable discharge limit for TROs (<0.2 mg/L) set by the IMO before being discharged into the receiving water. Table 2; Figure 1C present the details of the sampling points. All analytes were injected with an appropriate preservation reagent for each analysis to prevent changes in properties. Especially for DBPs, sampling preservation and pretreatment were considered to have high TRO concentrations and salinity. The whole procedure from sampling to reporting was performed based on modified United States Environmental Protection Agency (US EPA) methods, ISO standards, or other verified methods while considering a matrix effect for components such as high salinity and particulate matter. For THMs and HAAs, sodium thiosulfate was added to the samples. Ammonium chloride was added to samples for the determination of HAN and chloropicrin (CP). The pH of the samples was adjusted using sulfuric acid for THMs, DOC, and POC, and phosphate buffer for HANs and CP. The samples were transported into the laboratory, stored at < 4°C, and analyzed within 14 days. The details of sampling points and parameters for land-based tests are presented in Figure 1; Table 1, respectively.

The target compounds for this study were divided into causative substances and DBPs. As causative substances, TRO, free residual oxidant (FRO), DOC, and POC contents as well as ultraviolet absorbance at 254 nm (UV₂₅₄) were analyzed, and the specific ultraviolet absorption (SUVA) was calculated from the UV₂₅₄ and DOC values. The TRO and FRO contents were measured to confirm the concentrations of active substances, such as NaOCl, HOCl, hypochlorite (OCl⁻), and bound chloramine generated in the electrolysis process. DOC and POC contents were measured to determine the concentrations of organic matter in the test waters. UV₂₅₄ and SUVA were analyzed to investigate the characteristics of organic matter. THM₄, HAA₁₀, HAN₉, and CP were selected as DBP response variables (Supplementary Table S3).

Reagents unaffected by impurities during the measurement and analysis of analytes were used in this study. Potassium hydrogen phthalate (reference material for DOC and POC) and calcium hypochlorite (technical grade; reference material for TRO and FRO) were purchased from Sigma-Aldrich (St. Louis, MO, United States). A 1000 mg/L stock solution of each reagent was prepared and diluted before use. The THMs, HAAs, HANs, and CP were purchased from AccuStandard (New Haven, CT, United States). Fluorobenzene and 1,2-dichlorobenzene-d₄ were used as internal standards for THMs; decafluorobiphenyl (internal standard for HAAs and surrogate standard for HANs), 2-bromobutanoic acid (surrogate standard for HAAs), and p-bromofluorobenzene (internal standard for HANs) were purchased from AccuStandard (New Haven, CT, USA). Artificial seawater was made using Daigo’s artificial seawater SP (Nihon Pharmaceutical, Tokyo, Japan). High-performance liquid chromatography grade solvents [methyl tert-butyl ether (MtBE), methanol, acetone, and n-hexane] were purchased from Burdick & Jackson (Muskegan, MI, United States). Tertiary distilled water was produced using the Milli-Q system (Millipore, Billerica, MA, United States). All reagents used in this study were of guaranteed reagent grade.

TROs and FROs are characterized by high reactivity immediately after generation and their concentrations decrease over time. It is therefore crucial to obtain reliable data and minimize the effects of interference. Consequently, TROs and FROs were immediately analyzed in the field using a CLX Online Residual Oxidant and Chlorine Monitor (HF Scientific; Salt Lake City, UT, United States) using the N,N-diethyl-1,4-phenylenediamine method (ISO 7393-2). DOC and POC were analyzed using a total organic carbon analyzer (TOC-VCPH; Shimadzu; Kyoto, Japan) according to ISO 8245. UV₂₅₄ was measured using a UV/Vis spectrophotometer (V-650; Jasco, Easton, MD, United States) according to US EPA 415.3, and the SUVA was calculated by dividing the UV₂₅₄ by the DOC values. THM₄ was analyzed using purge and trap equipment with a gas chromatograph/mass spectrometer (GC/MS) (7890B GC-5977A MSD; Agilent, Santa Clara, CA, United States) according to US EPA 524.2. HAA₁₀, HAN₉, and CP were extracted using MtBE according to US EPA 552.3 for HAA₁₀, and US EPA 551.1 for HAN₉ and CP. For HAA₁₀, derivatization was performed using a 10% sulfuric acid/methanol solution. The final extracts for HAA₁₀, HAN₉, and CP were determined using a gas chromatograph/electron capture detector (7890A GC; Agilent, Santa Clara, CA, USA) and GC/MS (7890B GC–5977A MSD).

Quantitative analysis of the analytes was performed using the internal standard method, and quality control was performed in accordance with the certification criteria for each item. Calibration curves were drawn with at least five concentrations. The coefficient of determination of the calibration curve was set to >0.99; the standard deviation of the relative sensitivity coefficient was set to <20%. Supplementary Table S3 lists the verification results, including the method detection limits, accuracy, and precision (relative standard deviation). The accuracy and precision were within the acceptable ranges (accuracy: 81.6%–123%, precision: 0.31%–16.0%). Blank, duplicate, and calibration check samples were analyzed by binning 20 samples together as one batch.

We assessed the sum of DBPs (Figure 2) and variations in the concentrations of TROs, FROs, and DBPs (Figure 3) in electrolyzed water samples according to holding periods under each salinity condition. The concentrations of THM_4, HAA₁₀, and nitrogenous DBPs (N-DBPs) are listed in Supplementary Tables S4–S6. As shown in Figure 3A, during the 5 days of storage, as proposed by the GESAMP Ballast Water Working Group on Active Substances, DBPs formed in the treated water samples after electrolysis via the reaction of TROs with organic matter over time. The initial TRO and FRO concentrations in treated water (before neutralization) at ballasting (day 0) were 7.53–9.85 mg/L and 5.42–9.68 mg/L, respectively. After 5 days (after neutralization), all deballasting water satisfied the maximum allowable discharge concentration of <0.2 mg/L TRO as Cl₂ under all salinity conditions. Chlorine produced in the electrolysis process is converted into HOCl, OCl⁻, OBr⁻, and HOBr⁻, which act as free oxidants. These species react with ammonia or organic nitrogen present in the water to form chloramines, such as monochloramine, dichloramine, and trichloramine, also known as combined chlorine (Mazhar et al., 2020). Further, as active substances, TRO and FRO react with organic matter in water (Lu et al., 2009; Zhang et al., 2009; Lee et al., 2017). Both TRO and FRO appeared to decrease during the holding period because of the organic matter added to the challenge water according to Regulation D-2 (Ballast Water Performance Standard) of the Convention (Supplementary Table S2).

In contrast to the concentrations of TROs and FROs, concentration of DBPs (THM₄, HAA₁₀, and HANs) in the treated water following electrolysis increased from day 0–five under all salinity conditions (Figure 3A). The treated samples before neutralization in seawater had the highest levels (day 5: 163–759 μg/L), followed by brackish (day 5: 136–501 μg/L) and fresh water (day 5: 170–332 μg/L); the concentrations were low after neutralization (deballasting). We collected and analyzed challenge water and untreated water as a control, but did not obtain any specific results regarding DBPs.

The concentrations of THMs were significantly higher than those of HAA₁₀ and HAN_9. On day 5, the THM₄ concentration was highest in seawater (123–584 μg/L), followed by that in brackish (93.3–382 μg/L) and fresh water (117–263 μg/L). Among the THMs, TMB had a 100% detection frequency and was mainly detected in seawater (119–570 μg/L) and brackish water (83.3–361 μg/L). Both dibromochloromethane (DBCM) (seawater: 11.9 μg/L; brackish water: 15.4 μg/L) and dichlorobromomethane (DCBM) (seawater: 1.68 μg/L; brackish water: 6.32 μg/L) were present at relatively low concentrations. In contrast, the trichloromethane (TCM) concentration was highest (21.1–153 μg/L) in fresh water. Before neutralizing the treated water (day 5), tribromomethane (TBM) had the highest concentration, followed by DBCM and DCBM, while TCM was not detected in seawater (Figure 3B). The composition of THMs in brackish water was similar to that in seawater. However, the distribution of THMs in freshwater showed the opposite pattern, with TCMs and DCBMs accounting for 65% and 20%, respectively. HAAs and HANs did not show a clear, substance-specific pattern like THMs did, but were classified as belonging to chlorinated and brominated DBPs.

HAA₁₀ occurred in a pattern similar to that of THM₄ under each salinity condition, with 39.6–360 μg/L in seawater, 41.1–144 μg/L in brackish water, and 51.9–67.1 μg/L in fresh water on day 5. Among the HAA₁₀, dibromoacetic acid (DBAA) was detected under all salinity conditions. THM₄ (64%–74%) and HAA₁₀ (20%–34%) were the dominant groups, with higher contributions among DBPs in our experiments. Studies have reported that the reaction time positively influences the formation of THMs and HAAs (Pourmoghaddas and Stevens, 1995; Nikolaou et al., 2004; Water Resources Research Institute, 2008; Kali et al., 2021). Consistent with our observations, THMs and HAAs as major byproducts were shown to account for at least 80% of total DBPs formed in chlorinated water (Liu et al., 2011; Delacroix et al., 2013).

By contrast, HAN₉ showed a decreasing trend during the holding period. N-DBPs, including HAN₉ and CP, were present at lower levels than THM₄ and HAA₁₀. On day 5, the N-DBP levels were 1.12–13.5 μg/L in seawater, 1.78–22.8 μg/L in brackish water, and 0.90–1.08 μg/L in fresh water. In treated water, BCAN and DBAN (Dibromoacetonitrile) were detected with a 100% detection frequency under all salinity conditions. N-DBP concentrations did not increase continuously in treated water, and the levels were lower than those of other groups such as THM₄ and HAA₁₀.

According to research on the formation and pathway of N-DBPs, such as haloacetonitrile, halonitromethanes, and haloacetamides, they are rapidly generated during chlorine or chloramine disinfection via decarboxylation and undergo hydrolysis reactions (Reckhow et al., 2001; Lui et al., 2011; Shah and Mitch, 2012; Huang et al., 2013). They are produced with the influence of specific nitrogen-containing organic matter such as proteins and amino precursors (Huang et al., 2013). Since the concentration of nitrogenous compounds is typically less than 5% of the DOC level in water, nitrogen-containing DBPs occur at concentrations lower than those of carbonaceous DBPs (THMs and HAAs) and may have different occurrence patterns (Reckhow et al., 2001; Lui et al., 2012; Shah and Mitch, 2012). Kim et al. (2002) reported that HANs decreased with contact time after 48 h as DCAN, as reaction intermediates of DBP formation, such as TCAN, DCAA, or TCAA, disappeared. In addition, HANs are converted to haloacetamides after hydrolysis (Nihemaiti et al., 2017; Vu et al., 2018; Kali et al., 2021; Shao et al., 2022).

In a recent study, 30 DBPs classified as relevant chemicals were detected in treated waters (deballasting) generated from a BWTS (42 in total) using active substances (David et al., 2018). The frequency of occurrence of the four bromine DBPs was over 70%, including TBM (average: 247.14 μg/L), DBAA (average: 48.66 μg/L), DBCM (average: 22.11 μg/L), and monobromoacetic acid (average: 15.48 μg/L). Moreover, analysis of the DBPs (HAA, HAN, and THMs) in ballast water from seven ports in the Adriatic Sea within each medium (seawater, sediment, and transplanted mussels) showed that the highest TBM concentrations were present in seawater. The pattern and distribution of these compounds were similar to those detected in the present study, although the concentrations were slightly lower than those in this study.

Bromide ions (Br⁻) play a major role as a catalyst of brominated compounds (Allonier et al., 1999; Taylor, 2006). Consequently, DBP species had different composition characteristics of chlorinated and brominated compounds in treated water during the holding period (Figure 4).

In the case of THM₄, the brominated fraction (TBM) was greater than the other fractions in seawater (95.0%–100%) and brackish water (39.1%–94.6%). Although the brominated fraction was dominant in brackish water, the composition patterns somewhat differed from those in seawater because of the differences in Br⁻ levels. Compared to seawater, brackish water had higher contents of chlorinated (0.0% compared to 0.0%–47.0%, respectively) and bromochlorinated (0.0%–4.96% compared to 5.4%–29.1%, respectively) species. In particular, the chlorinated compound TCM was not detected in seawater during ballast water electrolysis. The chlorinated fraction (56.5%–69.9%) was the highest in fresh water, followed by the bromochlorinated (30.1%–41.8%) and brominated fractions (0.0%–1.78%); further, relatively higher fractions of chlorinated compounds were present in freshwater than in seawater and brackish water.

The composition of HAA₁₀ showed similar tendencies to that of THM₄. In seawater, brominated species (39.9%–100%), such as TBAA and DBAA, were dominant, followed by the bromochlorinated (0.0%–60.1%) and chlorinated fractions (0.0%–16.1%). In brackish water, the brominated fraction (25.1%–67.6%) was found to be lower than in seawater, while chlorinated (12.6%–51.1%) and bromochlorinated species (0.0%–55.0%) showed the opposite trend. In fresh water, the fractions of chlorinated species (60.6%–68.5%), including TCAA and DCAA, were higher than those of both the brominated (6.4%–10.5%) and bromochlorinated species (24.4%–30.2%).

In seawater, brominated N-DBPs (20.4%–87.3%) were dominant, followed by bromochlorinated (6.0%–37.5%) and chlorinated species (2.1%–43.4%). The N-DBP compositions of brominated (29.3%–87.8%), chlorinated (3.9%–50.4%), and bromochlorinated species (4.9%–27.7%) in brackish water were similar to those in seawater, unlike THM₄ and HAA₁₀. Although the brominated fraction (21.8%–65.6%) in fresh water was lower than that in seawater and brackish water, brominated N-DBPs were more dominant than bromochlorinated (32.5%–73.0%) and chlorinated species (0.0%–31.9%). N-DBPs had somewhat different tendencies from those of THM₄ and HAA₁₀, suggesting that limiting factors such as nitrogen also affected N-DBP formation.

Overall, brominated DBPs increased with increasing Br⁻ levels in ballast water, while chlorinated and bromochlorinated species showed the opposite trend. DBPs such as THMs and HAAs increase as Br⁻ increases (Pourmoghaddas and Stevens, 1995). During ballast water electrolysis, the chlorine produced converts some of the Br⁻ into hypobromous acid (HOBr⁻). This is because chlorine is a stronger oxidizer than Br⁻, and can oxidize these ions to form HOBr⁻. The generated HOBr⁻ then reacts with the dissolved organic matter (DOM) in the water. If there are enough functional groups on the DOM for HOBr⁻ to react with, both electrophilic substitution and electron transfer reactions will actively occur, resulting in the formation of a large amount of organic bromine compounds. Some of the formed organic bromine compounds can be further chlorinated by chlorine to form chlorinated organic compounds. During this process, Br⁻ is re-oxidized by chlorine to form more HOBr⁻; if sufficient Cl⁻ is present for reactions with organic matter, this cycle will continue (Langsa et al., 2017). In this study, brominated DBPs were dominant in both seawater and fresh water. The types and formation of DBPs influenced by Br⁻ differed from those in freshwater, which is similar to results reported by previous studies (Pourmoghaddas and Stevens, 1995; Taylor, 2006; Sun et al., 2009). Differences in Br⁻ (seawater: 51.3–65.1 mg/L; brackish water: 19.0–37.9 mg/L; and freshwater: 1.49–2.27 mg/L) and Cl⁻ (seawater: 15,200–19400 mg/L; brackish water: 6,230–10900 mg/L; freshwater: 75.4–89.1 mg/L) concentrations due to salinity gradients can cause differences in the distribution of DBPs. Additionally, DBP concentrations during the holding period continuously increased due to chemical reactions between Br⁻ and residual chlorine in the treated water.

The DOC and POC levels in challenge water were 2.79 ± 0.96 and 2.57 ± 0.61 mg/L (mean ± standard deviation) in seawater, 8.03 ± 0.33 and 13.5 ± 4.70 mg/L in brackish water, and 10.5 ± 0.42 and 5.31 ± 2.77 mg/L in fresh water, respectively (Figure 5).

The SUVA was calculated to investigate the properties of the organic material in the water (Figure 5). In all cases, the values prior to electrolysis were <1 (0.13–0.62). In the treated water after electrolysis on day 0, except for system A, the SUVA values increased by more than 4 (maximum: 7.36), and subsequently decreased over time. Generally, a SUVA value of >4 indicates relatively hydrophobic DOM composed of aromatic, high molecular weight, organic substances (primarily aquatic humic substances). In contrast, a SUVA value of <3 indicates DOM composed of low molecular weight, organic, hydrophilic, non-humic substances. According to previous studies, an increase in SUVA values is indicative of a higher decrease in the DOC concentration and depends on the ratio of glucose to humic substances during microbial cultivation (Hur et al., 2009; Hur et al., 2013; Liu et al., 2014). In this study, the increase in the SUVA value on day 0 after electrolysis may be attributed to a considerable amount of DOM, produced by the decomposition of dead microbes into polymeric and refractory substances (Barker and Stuckey, 1999; Jarusutthirak and Amy, 2007). Furthermore, organic additives such as glucose and fructose added to the water could be converted into organic compounds with double bonds or lead to the enhancement of molecular aromatic substances after electrolysis. The higher the content of unsaturated bonds and aromatic substances in the water, the higher the UV₂₅₄ value. These substances are highly reactive with oxidants such as ozone and chlorine, and thus the SUVA value appears to decrease over time (Liu et al., 2011).

Figure 6 shows the SUVA and DBP values normalized relative to the DOC concentrations. At SUVA values >3, the DBP formation pattern in water followed the order THM₄ > HAA₁₀ > N-DBPs under all salinity conditions. THMs can be generated from highly hydrophobic substances such as humic acids (Malliarou et al., 2005). However, DBP formation may also be affected by other water conditions (e.g., pH), as well as the properties of specific organic materials. For example, THMs are produced in greater amounts than HAAs at a pH of approximately 8 (Pourmoghaddas and Stevens, 1995; Lu et al., 2009). As the pH values of our test waters ranged from 7.6 to 9.2, the pH likely affected the compositions of THM₄ and HAA₁₀. Meanwhile, N-DBPs did not have any specific effects on the SUVA values because of their dependency on nitrogen (Liu and Li, 2010).