For the preparation of a Chinese herbal medicine, herbal extraction residues were boiled together with all kinds of ingredients. Some of these herbal extraction residues were collected from three pharmacies, and the mixed samples were dried using a freeze dryer, then sieved using a 100-mesh sieve, and ground. The samples and Milli-Q water were mixed at a ratio of 1:10 (w/v) and then shaken horizontally for 24 h at 20°C so as to collect the DOM samples. The DOM extract was centrifuged at 8,000 rpm for 12 min and then filtered to 0.45 microns (Millipore). Dissolved organic carbon concentration (DOC) is a typical DOM concentrate obtained with a total organic carbon analyzer.

To remove the inner filter influence and ensure the comparability of samples, TOC with an absorbance of less than 0.08 at 350 nm was used for the quenching experiments. The temperature of the solution was fixed at 20°C. NaOH and HNO₃ were added to the processor, and the pH value of the samples was within the range of 7.5 ± 0.05. Quenching experiments were carried on by adding Cu(II) ion solutions to the DOM derived from the herbal extraction residues. The ultimate Cu(II) concentrations of DOM solutions were controlled at 0, 20, 40, 60, 80, 100, 120, 150, 180, 210, 240, 270, 300, and 350 μmol L⁻¹. To overcome the influence of concentration, the proportion of added metal solution was kept relatively low (less than 5%). All DOM-HM samples were shaken using a horizontal oscillator for 24 h. The reaction equilibrium was maintained at 20°C. To quantitatively describe the complexation reaction between Cu(II) and DOM, a linear model modified Stern–Volmer (S-V) equation was made available to determine the complexation stability constant for the DOM-HM complexes (Lu and Jaffe, 2001), where it was assumed that metal ions form a 1:1 complex with DOM. F 0 / ( F 0 − F ) = 1 / ( f ⋅ K M ⋅ C M ) + 1 / f (1) where F₀ and F indicate the intensity of the fluorescent component without the metal added and with the concentration of C_M, respectively; f is the initial fluorescence fraction available to the eliminator; and K_M indicates the conditional stability constant (Wu et al., 2004).

The EEM data were collected with an excitation limit of 200–450 nm, the emission extent of 300–550 nm, with a 5-nm gap for both excitation and emission, using a Hitachi F-7000 spectrofluorometer (Hitachi, Japan) at 20°C. To subtract Raman scattering of water, Millie-q water was used for the purpose of a reference for all EEM data. The absorption spectra were applied to correct inner filter effects (Jiang et al., 2017). The acquisition of SFS is constant between the excitation wave and the emission wavelength (Δλ = 60 nm). The solid powder of the DOM sample was collected using vacuum freeze-drying equipment for the measurement of FTIR. The FTIR spectrum was generated from wavelengths of 3,600 to 800 cm⁻¹.

The PARAFAC model is based on the trilinear decomposition theory and an iterative three-dimensional matrix decomposition algorithm with the principle of alternating least squares. The essence of PARAFAC is a decomposition algorithm that uses the core consistency test to obtain the most appropriate group fraction (Bro and Kiers, 2003). The preprocessed data are processed in MATLAB 12.0 software (MathWorks Company, the Natick, Massachusetts of United States). At the same time, split-half analysis and defect analysis were applied to verify the effectiveness of the PARAFAC model and determine the optimal DOM component fraction. A detailed theory was introduced in the study (Stedmon and Bro,2008). And in the current study, 56 EEM data collected from DOM-Cu(II) samples were used with the PARAFAC model. The PARAFAC model showed that a four-substance model could be verified with defect analysis, split-half analysis, and loadings. The maximum fluorescence intensity (Fmax) was obtained from the PARAFAC model for all the DOM-HM samples (Stedmon and Bro, 2008).

2D-COS has been applied as an analytical tool for identifying subtle changes in a spectrum caused by adding HMs (Ozaki et al., 2001). Herein, 2D-FTIR-COS was applied to reveal the binding heterogeneity of herbal extraction residue DOM with Cu(II) ions. “2D Shige” software was used to obtain synchronous and asynchronous maps. The synchronous 2D-COS map was symmetric on the leading line; its peaks belonged to auto-peaks and were induced by an external disturbance condition (different Cu(II) concentrations). The asynchronous 2D-COS map was asymmetric around the main diagonal line; these are just diagonal vertices and no auto-peaks (Ozaki et al., 2001). The transformation of the 2D correlation intensities was further treated by 1/n-th power transformation (n = 9). Origin 2018 was used to draw correlation maps after 1/9-th power transformation.

Figure 1 reveals the fluorescence components obtained by PARAFAC analysis. Four fluorescence components can be identified from the herbal extraction residue DOM: one humic-like (Component 4) and three protein-like components (Components 1, 2, and 3). Component 1 (C1) included two excitation maxima at 220 and 275 nm, with an emission maximum at 330 nm and a shoulder at 310 nm. It was related to a tryptophan-like fluorophore in conjunction with a tyrosine-like peak, and the tryptophan-like fluorophore was an analog to the traditional tryptophan-like peak B. The presence of Component 1 suggested that the tyrosine-like fluorophore was converted into a tryptophan-like fluorophore. Component 2 (C2) was a typical protein-like material such as a tyrosine-like material and exhibited a main peak at 275 nm and a second peak at 230 nm, with the same maximum excitation wavelength at 310 nm. Component 3 (C3) consisted of three peaks, at 265, 285, and 330 nm, with an emission maximum at 345 nm. This substance was also related to tryptophan-like substances and can be considered a biologically labile substance or an amino acid (Yao et al., 2011; Guo et al., 2020). Similarly, three fluorescent peaks at 240, 290, and 320 nm with a maximum emission wavelength of 435 nm could be found in Component 4 (C4). Component 4 was due to the substance of fluorescent fulvic-like substances.

The maximum fluorescence intensity (Fmax) shows the relative prevalence of fluorescent components in the herbal extraction residue DOM. It can be observed from Figure 2 that the relative abundance of the four substances followed the order of C1>C2>C3>C4, accounting for 54.79, 25.90, 15.95, and 3.36%, respectively. It should be noted that fulvic-like C4 presented an extremely low Fmax value. These results also indicate which protein-like fluorescent components were mainly fluorophores in the herbal extraction residue DOM after long decoction of Chinese medicine. A strong correlation between biochemical oxygen demand (BOD) and protein-like fluorescent components has also been explained in the literature (Hudson et al., 2008). Fluorescent components showed a strong correlation with BOD, chemical total nitrogen (TN), and oxygen demand (COD) concentrations (Hurand Cho, 2012), suggesting that the release of herbal extraction residue DOM can lead to a risk of water quality deterioration. Therefore, it is very important to prevent the discarding of herbal extraction residues to protect the water environment.

Two positive peaks could be identified in the synchronous map of DOM-Cu(II) at 275 and 383 nm, respectively (Figure 3). The peak at 275 nm has a bearing on protein-like materials, such as tryptophan- or tyrosine-like, and the peak at 383 nm was indicative of humic-like constituents (Guoet al., 2022). In view of Noda’s rule (Noda and Ozaki, 2004), the protein-like constituents at 275 nm exhibited a high sensitivity of Cu(II) compared to the humic-like constituents at 383 nm. The asynchronous map of DOM-Cu(II) presented one positive and two negative peaks at 286, 301, and 383 nm, separately. The presence of peaks at 286 and 301 nm belonged to the presence of protein-like materials, specifically tryptophan-like materials (Barker et al., 2009; Tan et al., 2021). The protein-like materials identified at a shorter wavelength (286 nm) in Figure 3 indicated the preferential binding of Cu(II) to the herbal extraction residue DOM. In addition, the priority binding with Cu(II) followed the subsequence as 286→ 383→301 nm.

Figure 4 shows the asynchronous and synchronous maps of 2D-COS of DOM-Cu(II) obtained by the 1/9-th power transformation. The synchronous map displayed two cross- and two auto-peaks at 383/275 nm, 327/275 nm, 275, and 383 nm, separately. The analysis outcome of the synchronous map was consistent with the result of the 2D-COS that protein-like fluorophores at 275 nm exhibited a high sensitivity to Cu(II) compared to the humic-like components at 383 nm. However, the asynchronous map indicated that eight fluorescent signal regions may be found at 286 nm, 301 nm, 327 nm, 383/275 nm, 383/327 nm, 409 nm, 433 nm, and 463 nm, suggesting that the 1/9-th power transformation was feasible for identifying small features obscured by strong peaks. A peak at 327 nm can be found, which was assigned to fulvic-like components. The peaks at 409 nm, 433 nm, and 463 nm were also concerned with humic-like fluorophores (Guoet al., 2022). According to the results of 2D-SFS-COS obtained by the 1/9-th power transformation, the priority binding of Cu(II) followed the sequence as protein-like (286 nm) → fulvic-like (327 nm) → humic-like (409 nm) → humic-like (433 nm) → humic-like (463 nm) → humic-like (383 nm) → protein-like (301 nm). The results also indicated that the 2D-SFS-COS obtained by the 1/9-th power transformation strengthened shielded long-wavelength and low-intensity fluorescence signals and can obtain more information such as ligand binding characteristics.

2D-COS in conjunction with FTIR spectra has been extensively used to explore information on potential binding sites (Xu et al., 2013; Guo et al., 2020). Figure 5 depicts the asynchronous and synchronous maps of 2D-FTIR-COS of DOM-Cu(II). The synchronous map of DOM-Cu(II) suggests that a large auto-peak and three cross-peaks appeared at about 2,300 and 2,920 cm⁻¹, separately. The peaks at about 2,300 cm⁻¹ were considered to be the C≡C stretching vibration of alkyne, while the band at 2,920 cm⁻¹ was considered to be the C−O vibration of carboxylic acid or the C−H stretching vibration of alkane (Yang et al., 2019). These results reveal a preferential response in the functional group of alkyne due to the increase in Cu(II) concentration.

The asynchronous map of DOM-Cu(II) exhibits three positive bands (3,370, 2,920, and 1,600 cm⁻¹) and four negative bands (2,270, 1,490, 1,330, and 1,180 cm⁻¹). These bands contain about seventeen binding sites, suggesting that the herbal extraction residue DOM provided a large number of binding sites for Cu(II). According to previous literature (Yu et al., 2012; Huang et al., 2018; Guo et al., 2020), the peaks at 3,370 and 1,600 cm⁻¹ are related to the N−H stretching of amine and the aromatic C=C stretching or N−H bending of primary amide (band II), separately. The two peaks at 1,490 and 1,330 cm⁻¹ were down to the OH deformation of the phenolic hydroxyl group, and the C−O vibration of esters or the C−N stretching of aliphatic amine were identified according to the band at 1,180 cm⁻¹. As a result, the N−H of amine functionalities could preferentially integrate with Cu(II), and the Cu(II) binding with the herbal extraction residue DOM followed the order 3370→2920 →1600→2270→1490→1180→1330 cm⁻¹.

Supplementary Figure S1 exhibits the quenching curves of Cu(II) with the herbal extraction residue DOM for the fluorescence substance determined by the PARAFAC model and the 2D-COS analysis. The fluorescence intensity of protein fluorophores declined with the increase in the Cu (II) ion content. Protein-like materials presented better quenching effects in the DOM-Cu(II). 2D-FTIR-COS also revealed that the N−H of amine preferentially combined with Cu(II), suggesting that the N−H of amine mainly originated from the protein-like fluorophore. Fulvic-like fluorophores showed poor quenching effects due to the presence of limited fluorophore which could bind with Cu(II) ion. Clear quenching curves for these peaks with low intensities can be found in Supplementary Figure S1.

Table 1 lists the parameters of DOM-Cu(II) calculated with the revised Stern–Volmer model. The log K_M value was in the range of 3.96–4.24 for components 1–4, and 4.14–4.81 for C₂₇₅, C₂₈₆, C₂₉₉, C₃₂₇, and C₄₃₃, respectively. These results suggested that the log K_M value of different components with Cu(II) increased with an increase in the wavelength of these components. However, the log K_M values were generally lower than those of DOM-Cu(II) from other sources, such as digestate (Guoet al., 2020), lake water (Craven et al., 2012), and biochar (Guoet al., 2021; Guo et al., 2022), indicating a relatively chemical stability of DOM-Cu(II). The f% value decreased with the increased wavelength of fluorescent components. However, Component 1 revealed the largest f% value, suggesting that a combination of tyrosine- and tryptophan-like fluorophores can increase the concentration of the total ligand. Humic-like substances showed the largest log K_M value, followed by fulvic-like substances, compared with protein-like materials. Although fulvic- and humic-like substances with Cu(II) have high log K_M values, their contents were dramatically low in the herbal extraction residue DOM. Therefore, protein-like fluorophores dominate due to their high fluorophore content.

Traditional Chinese medicine, including herbal medicine, is still used extensively in China. As mentioned earlier, the haphazard stacking and discarding of herbal extraction residue put tremendous pressure on the environment, especially in rural areas. Protein-like substances dominated in the herbal extraction residue DOM used in this study, and this residue can easily enter rivers via rainwater leaching, resulting in the deterioration of the water environment. The herbal extraction residue DOM and HMs can affect each other, affecting their transformation, biotoxicity, and other properties. With the increase in HM pollution in the soil, herbal extraction residue DOM increases the risk of HMs entering the groundwater from the soil because the complexes formed by DOM-HM have low stability. Therefore, the treatment and utilization of the herbal extraction residues have become an urgent problem, and solving this problem can reduce the environmental deterioration caused by the random stacking and discarding of herbal extraction residues. Biomass energy, papermaking, composites, and re-extraction of bioactive components are promising solutions for herbal extraction residues (Lu and Li, 2021). For example, herbal extraction residues have been used to explore the feasibility of producing biogas for industry (Cheng and Liu, 2010; Li et al., 2011). Herbal extraction residues can also be used to prepare biochar materials for other applications (Wang et al., 2010). In addition, the improvement of environmental protection awareness and environmental infrastructure is also an important way to solve the environmental pollution of the herbal extraction residue in rural areas. The implementation of garbage classification can greatly reduce the environmental risk of the herbal extraction residues.