SRNOM and MRNOM were purchased from the International Humic Substances Society (IHSS). Samples were obtained in South Korea on October 2021, in which the Nakdong River NOM (NDNOM) was extracted from the Nakdong River near Bonpo Waterfront Park (35.373295N, 128.646821W), and the effluent organic matter (EfOM) was obtained from the Jinhae municipal wastewater treatment plant (35.1436N, 128.69W), South Korea. pH and water conductivity were measured using a multi-meter (Orion 4 star, Thermo, Cleveland, OH, USA). Following sample collection, the samples were filtrated using a 0.45-μm cellulose acetate membrane filter (Whatman Cat No: 7001 0004, D-47mm, Japan) and stored at 4 °C for further analysis. DOC was measured using the total organic carbon (TOC) analyzer (Sievers M9, SUEZ, France). An ultraviolet (UV)-visible spectrometer (UV-1601, Shimadzu, Japan) was used to measure the UV absorbance of the samples at 254 nm (UVA₂₅₄). The specific UV absorbance (SUVA) value was calculated from the UVA₂₅₄ and DOC values. The ion concentration was measured using ion chromatography (ICS-90, Dionex, CA, USA). An AS14 column (250 × 4 mm, Dionex, CA, USA) was then used to measure the cation and anion concentrations. A fluorescence spectrophotometer (RF-6000, Shimadzu, Japan) was used to analyze the qualitative oragnic properties of the selected NOMs. The emission spectra range used was between 250–600 nm, in which the excitation spectra ranged from 230–550 nm. The wavelength step of 2 nm was used to draw the excitation-emission matrix (EEM) in the form of three-dimensional (3D) contour plots. The detected intensities of each sample were divided into three regions for characterization of DOM [19–21] protein-like fluorescence (PLF) at λ_ex/λ_em=270–290/350–365 nm, fulvic-like fluorescence (FLF) at λ_ex/λ_em=320–340/410–430 nm, and humic-like fluorescence (HLF) at λ_ex/λ_em=370–390/460–480 nm.

Fluorescence indices, such as humification index (HIX), fluorescence index (FI), and biological index (BIX), were estimated using the following equations, respectively: [22]

HIX indicates the humification degree (Eq. (1)), FI estimates the degradation degree of DOM (Eq. (2)), and BIX shows the degree of the influence by microbial activity (Eq. (3)).

The selected samples were analyzed in triplicates using the pyrolysis gas chromatography-mass spectrometry (Py-GC-MS) system. The samples were concentrated using a rotary evaporator and were then freeze-dried as a pre-treatment. The system was coupled with a curie-point pyrolyzer (JCI-22, JAI, Japan) and GCMS-QP2010 (Shimadzu, Japan). Approximately 0.1 mg of the sample was prepared for pyrolysis in a ferromagnetic foil (Profoil, JAI). The pyrolysis temperature was 590°C, and helium was used as the carrier gas. The initial gas chromatography (GC) oven temperature was heated from 40°C to 300°C. Temperature increase was set at a rate of 7 °C/min. The holding time at the final temperature was 10 min. The DB-5MS (30 m, i.d. 0.25 mm, film thickness 0.50 μm, Agilent, USA) was used to separate volatilized organic compounds (VOCs), using helium as the carrier gas. The ion source was set to a temperature of 210°C. The compounds were ionized at 70 eV and mass analyzed over a range of 30–500 amu. The pyrolyzed fragments were identified using the NIST 14 library as the reference database. Based on the work of Bruchet et al. [23], each sample result was categorized into six groups: polysaccharide (PS), amino sugar (AS), protein, polyhydroxy aromatic compounds (PHA), lignin, and lipid.

As a pre-treatment step for the orbitrap mass spectroscopy analysis, the solid phase extraction (SPE) was conducted to eliminate inorganic salts. The SPE method and orbitrap mass spectroscopy operating conditions followed those of the method described by Baek et al., [24]. NOM was absorbed using the manual packed SPE containing HLB (Oasis, waters, USA), ENV+ (International sorbent technology, UK), Strata X-AW, and X-CW (Phenomenx, UK). Cartridge conditioning was performed using 5 ml and 10 ml of methanol and deionized water, respectively, in which 1 L of the sample was loaded. Thereafter, the absorbed NOM was extracted using 6 ml ethyl acetate/methanol (50:50 v/v) with 0.5 % ammonia and 3 ml of ethyl acetate/methanol (50:50 v/v) with 1.7 % formic acid. This method demonstrated a high recovery rate of the substrate, exceeding 75% [25]. The pretreated sample was analyzed using the Ultimate 300 UPLC system (Thermo Fisher Scientific, San Jose, CA, USA) coupled with the exactive and orbitrap mass spectrometry (Thermo Fisher Scientific, San Jose, CA, USA) with a heated electrospray interface (HESI). HESI was operated under the following conditions: 45 L/min sheath gas flow, 320 capillary temperature, 3800V/3000V spray voltage, 10 arbitrary units auxiliary gas pressure, and 2 arbitrary units ion sweep gas. The injection volume was 200 μl with the methanol mobile phase. Mass spectra were recorded between 100 and 2000 m/z. The molecular formulas were assigned using a compound identification algorithm on MATLAB, as introduced by EB Kujawinski and MD Behn [26].

Table 1 summarizes the results of the water quality analysis and compares the difference among the dissolved NOM samples. The pH values of the dissolved SRNOM (D-SRNOM) and MRNOM (D-MRNOM) were approximately 4.3, and those of the NDNOM and dissolved EfOM were approximately 7. The low pH of the D-SRNOM and D-MRNOM is attributed to the cation-exchange resin that was used for desalting the NOMs. According to Green et al. [27], IHSS NOM is concentrated using the reverse-osmosis (RO) membrane system. Then, desalting of the concentrated NOM results from the reduction of the high concentration of salts by the ion-exchange resin. With the release of hydrogen ions by the cation exchange resin during cation adsorption from the sample, the functional group of NOM is then protonated [28]. Hence, the low conductivity and low pH of the D-SRNOM and D-MRNOM can be attributed to the desalting process.

The difference in the hydrophobicity of each NOM is represented by the specific UV absorbance (SUVA) value, as high SUVA values correspond to higher hydrophobicity. The D-SRNOM had the highest hydrophobic NOM, followed by D-MRNOM, NDNOM, and EfOM. The ion concentration analysis also showed a similar pattern of conductivity, in which D-SRNOM and D-MRNOM had low concentrations of anions and cations. However, the conductivity of NDNOM and EfOM was higher than that of D-SRNOM and D-MRNOM. This may be attributed to the sample source, as NDNOM originated from natural river water, whereas EfOM was sampled from wastewater treatment effluent. As EfOM also underwent a post-chlorination process, high concentrations of chloride were observed. Hence, the comparison of the water quality of dissolved NOM confirmed that the changing water characteristics should be considered when using a commercial NOM to simulate feedwater conditions.

Fig. 1 shows the 3D EEM spectra of DOMs, in which the fluorescence spectra provide organic fingerprinting information of a heterogeneous sample. As the 3D EEM spectra can be interpreted using the peaks or regions at different excitation and emission wavelengths related to the chemical composition, it is considered as a non-destructive method. Both D-SRNOM and D-MRNOM showed significant humic-like and fulvic-like fluorescence components based on the obtained HLF (D-SRNOM: 31.0, D-MRNOM: 29.7) and FLF (D-SRNOM: 50.0, D-MRNOM: 45.7) values in Table 2. These indicate that approximately 80% of the detected organic matters are classified as allochthonous, whereas only 20% of the detected organic matters are classified as protein-like substances. NDNOM and EfOM showed higher portions of PLF values than D-SRNOM and D-MRNOM, suggesting that these NOMs are not highly humified and contain a relatively high biodegradable portion in their molecular structures.

Various fluorescence indices can be used to examine differences in humification and precursor organic matter [29, 30]. Higher HIX values indicate a higher degree of humification. Humification refers to the process by which complex organic compounds are transformed into humic substances, which are more stable and resistant to degradation. Higher FI values classify the sample as being of terrestrial origin, as they represent the relative contribution of terrestrial and microbial sources to the DOM. Higher BIX values correspond to recently produced DOM of autochthonous origin. The HIX of D-SRNOM and D-MRNOM was 14.22 and 11.44, respectively, indicating high humidification. This result was consistent with the high SUVA values of D-SRNOM and D-MRNOM in Table 1 (D-SRNOM: 3.71, D-MRNOM: 3.48), including the HLF and FLF values. Additionally, their FI values were also relatively lower than those of EfOM, where the latter showed similar values with NDNOM. EfOM showed relatively higher BIX values than D-SRNOM and D-MRNOM, which indicates that it contains higher fractions of organic matters that originate from microbial activity [31, 32], such as those found in biological wastewater treatment plants, owing to the microbial byproduct-like material that contributes to the high fraction of the organic substances. The high PLF result of EfOM also indicates that EfOM is highly influenced by biological processes.

The analysis of the biomolecular structure of the selected NOMs using the Py-GCMS showed that SRNOM and MRNOM exhibited similar biomolecular compositions, in which they consisted of polysaccharides (approximately 50%) and lignin (approximately 35%), which accounted for the majority of their structure (Fig. 2). Meanwhile, the portions of protein and lipid accounted for approximately 5% and 7%, respectively (Fig. 2). According to the IHSS, the SRNOM that was extracted from blackwater River containing peat and decomposing vegetation, in which lignin and polysaccharide, which were the most dominant components, derived polyphenols or cellulose. Although the detailed characteristics of the MRNOM is not provided by the IHSS, the result implies that the molecular characteristics of MRNOM and SRMON are similar in this study. This was also consistent with the results of previous studies [33, 34] that reported aromaticity ranging between 20–25% and partition coefficients of 1.29 and 1.49 for SRNOM and MRNOM, respectively.

SUVA was detected at 254 nm when the SRNOM and MRNOM was 4.25 and 4.0, respectively. The FI was also relatively similar to each other, with 1.304 and 1.567 for SRNOM and MRNOM, respectively [35]. These results imply that both NOM are influenced by plant decomposition processes, arising from high lignin and polysaccharide portions and low protein and lipid portions. This can be attributed to the protein and lipid-like biopolymers that derive microbial debris.

Wastewater treatment effluents or enriched microbial environmental samples normally contain high portions of protein [3, 36]. In this study, EfOM showed a relatively high portion of protein percentage (31%). Consistent with previous studies, the result indicates that the major source of EfOM differs from SRNOM and MRNOM. As wastewater treatment plants treat highly biodegradable organic matters such as human feces and sewage, they can be easily decomposed using activated sludge processes. Here, the biopolymer composition of EfOM contained a large portion of microbe, which comes from biopolymers such as proteins and lipids. This finding was also similar to that of a previous study by Park et al. [3] that analyzed the DOM of water from a constructed wetland with a wastewater plant effluent as the influent, in which the NDNOM was composed of 34% polysaccharide, 23% protein, 2% PHA, 30% lignin, and 11% lipid. These portions indicate that the organic characteristics of NDNOM were mixed with previous analytical results from SRNOM, MRNOM, and EfOM, as SRNOM and MRNOM are normally characterized by high percentages of lignin and low percentages of proteins. The EfOM showed the highest portion of protein and lipid among the samples. Hence, the observed proportion of the biopolymers of NDNOM is inferred to be between these normal values, as NDNOM originates from a natural ecosystem that contains various organisms.

Orbitrap mass spectrometry is one of the most powerful tools that can aid in understanding the complexity of NOMs, as it interprets thousands of molecular compositions in a sample and provides estimated molecular formulas for each composition. The relative number of significant elements, such as carbon, hydrogen, and nitrogen, can then be calculated based on the results. Estimation of the hydrophobicity or classification of the biopolymeric character is also possible, as the typical atomic H/C or O/C ratio of each individual biopolymeric compound can be easily expressed. Several studies have suggested that specific regions of some major biomolecular components can be obtained from the van Krevelen diagram [37, 38]. Since NOM is composed of various organic compounds, the results of mass spectrometry reveal the diversity of molecules present. The number of peaks or spots in the diagram indicates the variety of molecules within a sample. The scattering patterns can vary depending on the origin of the NOM. Therefore, the differences observed among samples provide valuable information for characterizing each individual sample.

Laszakovits and MacKay [39] indicated that each study plotted the van Krevelen diagram, differently. Here, data-based chemical class regions were suggested to be used for van Krevelen diagrams, in which major biomolecules can be divided into six groups according to the H/C and O/C ratio: 1) amino sugar-like substance (1.62 < H/C < 2.35, 0.56 < O/C < 0.95), 2) carbohydrate-like substance (1.53 < H/C < 2.20, 0.56 < O/C < 1.23), 3) lignin-like substance (0.86 < H/C < 1.34, 0.21 < O/C < 0.44), 4) lipid-like substance (1.34 < H/C < 2.18, 0.01 < O/C < 0.35), 5) peptide-like substance (1.33 < H/C < 1.84, 0.17 < O/C < 0.48), 6) tannin-like substance (0.70 < H/C < 1.01, 0.16 < O/C < 0.84). In this study, these values were used to interpret the different NOM samples. The assigned MS sample peaks are described in the van Krevelen diagrams based on the CHON formulae in Fig. 3. Both D-SRNOM and D-MRNOM showed similar graphical patterns, in which the H/C ratio from the 0.8–1.2 range and the O/C ratio from the 0.3–0.6 range are highly concentrated. In these regions, lignin-like and tannin-like substances can be particularly found, which is consistent with the result of a previous study analyzing humic substances [37]. The diagrams of the NDNOM and EfOM also indicate that the lignin-like and peptide-like substances are the major biomolecule classes. However, amino sugar-like substances were also found, particularly in the EfOM samples, where they had the highest content among all the samples, indicating that EfOM contains organic matter from microorganisms, since lipid and amino sugar are included in the major compositions of microbial extracellular polymeric substances [40].

Table 3 summarizes the comparison of the number of biomolecules among the samples. D-SRNOM and D-NRNOM contained the largest portion of tannin-like substances (~40%), indicating that the samples are derived from a terrestrial source, as tannin can exist in leaves [41]. Amino sugar, lipid, and peptide-like substances also had low percentages, suggesting that they are highly humified, owing to the high biodegradable components. EfOM contained the highest percentage of amino sugar-like substances (14.7%) among the samples. The portion of lipid and peptide-like substances related to microbial activities were also relatively higher in EfOM than in the other study samples. Meanwhile, NDNOM showed mixed characteristics between D-SRNOM/D-MRNOM and EfOM, owing to the influence of both microbial and terrestrial sources. The sum of the percentages of the lignin and tannin-like substances is indicative of the influence of a terrestrial source, whereas that of amino sugar, lipid, and peptide-like substances indicates the influence of a bacterial source. The influence of the terrestrial and microbial sources on NDNOM was 46.1% and 40.9%, respectively.