Introduction

Fulvic acid (FA), which is one major component of humic substances (HSs)1,2,3, can be found in both natural waters and wastewater, especially in landfill leachate1. In the broadest terms, the structures of HSs can be described as assemblies of covalently linked aromatic and aliphatic residues carrying carboxyl, phenolic and alkoxy groups2. These common functional groups of FA make it possible that FA cause many serious environmental and health problems3. FA has strong complexation ability with heavy metals, leading to the formation of organometallic complexes with increased transportation ability, bioavailability and toxicity4. In the chlorination stage of the water treatment, FA and chlorine may produce a by-product trihalomethane whose strong carcinogenic and mutagenic properties seriously endanger human health5,6,7. Due to the complex structure, FA is frequently identified as refractory organic matter8. Therefore, biological methods are ineffective for FA treatment and many physical, chemical and physico-chemical methods have been employed for FA removal9. The advantages and disadvantages of these methods are summarized in Table 1.

Table 1 The methods of FA removal.
Full size table

Catalytic Wet Hydrogen Peroxide Oxidation (CWPO) is a kind of Advanced Oxidation Processes (AOP)10,11,12,13. The efficiency of CWPO for the degradative removal of organic matters from wastewater is mainly based on high amounts of hydroxyl radicals (·OH) from the reaction of catalyst and H2O2. The generated hydroxyl radicals are highly reactive species and thus can unselectively react with organic matters and convert them into smaller molecular products, and even directly into CO2 and H2O14,15. The advantages of high efficiency and low intermediate pollutants make the CWPO process more attractive for treating complex organic wastewater16,17. The degradation efficiency of CWPO can be improved by enhancing the catalyst activity. Zero valent metals including nanoscale zero valent iron (nZVI)18,19, zero valent aluminum (ZVAl)20, and zero valent copper (ZVC)12,21, could activate molecular oxygen to induce the generation of more powerful reactive oxygen species (ROS), such as superoxide radical anion (·O2−) and hydroxyl radical (·OH). Copper is one of the major redox-active transition metal catalysts, but the application of Cu catalyst in CWPO was severely limited due to the instability of Cu+. However, ZVC has a potential capacity to release Cu+ through oxidation to activate H2O2 to generate ·OH thereby resulting in the degradation of organic contaminants12,20.

Activated carbons have good performance in adsorption and catalysis22. Activated carbons can be produced through either physical or chemical activation. Physical activation involves pyrolysis of a carbonaceous precursor to produce a char follower by some activating agents such as carbon dioxide or steam. In the chemical activation, the precursor was pretreated with chemical activator (e.g. ZnCl2, H3PO4, KOH and NaOH), and then thermal treated via pyrolysis to produce an activated carbon23. Williams24 produced activated carbon via both physical and chemical activation with biomass waste flax fibre as precursor. The results showed that physical activation produced activated carbons with a nodular and pitted surface morphology whereas activated carbons produced through chemical activation had a smooth surface morphology. TEM analysis could identify mesoporous structures in the physically activated carbon and microporous structures in the chemically activated carbons. Sahira et al.25 used different activating agents (KOH, H2SO4, FeCl3, MgCl2 and CaCl2), and found out that the different activating agents have no significant effect on the nature of surface functional groups, with all showing similar oxygenated functional groups in FT-IR such as hydroxyl, carbonyl, carboxyl and lactones. Ahmed et al.26,27 prepared activated carbon via two-stage activation using H3PO4 and KOH and the produced activated carbon possessed high surface area (692.3 and 1368 m2/g) and pore volume (0.44 and 0.92 cm3/g).

Chitosan (CS), the second most abundant biopolymer after cellulose in nature and derived from chitin of crab shells and fungus cell walls28, has gradually attracted the attention of researchers28,29,30,31,32 due to its non-toxicity and biocompatible, cost-effective nitrogen precursor, and environmentally friendly features. Hitherto, CS and its derivatives were used in different areas including flocculating agents, adsorbents, catalyst, thickeners, food preservation, and many others33,34. Recently, a variety of carbon materials has been developed from CS and used in different applications, thereby highlighting its great impact on chemistry and material science35. He et al. prepared activated carbon by surface modification with CS as nitrogen source during KOH activation process, displaying a remarkable CO2 uptake achievement of 0.00583 mol/g36. In addition, CS contains abundant hydroxy and amine functional groups, which are responsible for its easy incorporation with the other materials31. CS can serve as a chelating agent for 3d metal ions such as Fe3+, Cu2+ and Ni2+ because of its flexible structure and high chemical reactivity32. The CS-derived metal–carbon nanocomposites can serve as catalysts due to their textural properties, the presence of nitrogen species, and the uniform dispersion of metal nanoparticles35. Guo et al.37 prepared the chitosan-supported iron (III) tetraphenylporphyrin. In their study, under reaction conditions of 418 K and 0.8 MPa, the cyclohexane oxidation catalyzed by chitosan-supported iron (III) tetraphenylporphyrin had 10.48% cyclohexane conversion and 79.20% cyclohexanone and cyclohexanol selectivity. Wang et al.38 prepared the Cu–M@CS–SiO2 degrading 1,1-dimethyl hydrazine (UDMH) wastewater. The COD removal of UDMH wastewater was 87.38% in half an hour.

Traditionally, the CWPO experimental conditions have been optimized by several single-factor or orthogonal experiments, in which the optimal conditions can be found only within the chosen experiment points. Additionally, these traditional methods also ignore the interaction among some parameters37. Plackett–Burman (PB)39,40 design is a statistical method for screening the significant influence factors in multi-factors experiments. The statistical experiment design of the response surface method (RSM) can optimize the values of all the influence parameters including the interaction factors and provide an optimal prediction model for the experiments. RSM can even find the optimal target response point outside the set condition interval41.

In this study, chitosan was used to produce stable chelates with copper ions and then the chitosan chelates was carbonated to provide well-distributed ZVC active sites42. The chitosan-activated carbon was used as both an adsorbent for adsorbing organic compounds to the reaction site and more importantly a carrier to support active metals of the catalyst. Additionally, carbon was used to reduce Cu2+ to Cu0 during the pyrolysis, which makes the catalyst perform better in the CWPO. The purpose of this study was to evaluate the efficiency of FA removal in the CWPO process with ZVC/CTS-ACB as catalyst. The RSM coupled with PB was used to optimize the parameters of the CWPO experiment. The catalyst was prepared and the catalytic performance in CWPO was investigated in comparison with other degradation processes in removal of FA, TOC and colour number (CN).

Materials and methods

Materials

FA was purchased from Cool Chemical Technology Corporation (Beijing). The appearance color of the FA solution varies from dark yellow to light yellow depending on the concentration. Chitosan was supplied by Jinan Haidebei Marine Bioengineering, whose Deacetylation Degree ≥ 85%, Particle ≥ 40 Mesh and viscosity = 200 mpas. All other agents were of analytical grade and purchased from Beijing Chemical Plant. The pH of the FA solution was adjusted by adding H2SO4 or NaOH. All the water used in the experiment was deionized water, from a Millipore-Q system with a resistance of 18.2 MΩ.

Preparation and characteristics of ZVC/CTS-ACB

The preparation of ZVC/CTS-ACB was carried out according to the following steps: (1) Preparation of chitosan gel: 7.25 g of CuNO4·3H2O was weighed and dissolved in 960 ml of deionized water, at the same time 0.1 g of citric acid was added, and then 30 g of chitosan was weighed and dispersed in a copper nitrate solution to form a suspension. 40 ml of acetic acid was added and stirred quickly and uniformly until a gel was formed. The gel was allowed to stand overnight to discharge air bubbles. (2) The gel was added dropwise to a 3.75 to 5 wt% sodium hydroxide solution using a syringe, then allowed to stand for 4–6 h. (3) After washing to neutral and drying at 60–100 °C for one night, the precursor was kept at 800–850 °C for 2–5 h in an inert gas tube furnace to complete its carbonization and become an activated carbon ball. (4) The activated carbon ball was rinsed several times in ethanol and deionized water, and then dried at 80 °C to remove moisture.

Experimental studies

The CWPO reaction was carried out in a reactor whose temperature can be controlled. The pH value of the water sample was adjusted by 0.5 M H2SO4 and 1 M NaOH solutions, and an appropriate amount of ZVC/CTS-ACB was added to the water sample. When the reaction temperature reached the set temperature, 30% of H2O2 was dropped into the reactor through the transfer pipe to make hydrogen oxide fully contact the catalyst to produce strong oxidizing free radicals43, such as ·OH, ·O2, etc. These free radicals can oxidize organics into small molecular organics and even H2O and CO2. Samples were taken every ten minutes from the reaction effluent under different conditions. MnO2 was added to the sample to prevent the influence of excess H2O2 on the detection results, and then the sample was filtered through a 0.45 μm membrane for the subsequent results testing.

Analytical methods

The surface and profile morphology of the catalyst and the distribution of the supported metal crystals were observed using SEM (FEI Quanta 200). The active crystal phase composition of the catalyst was identified by XRD (D8 Advance type). The element composition was analyzed by the XPS. The chemical group was detected by the FT-IR (IRPrestige-21). Total copper ion in solution was quantified by the ICP. The pH value was measured using a pH meter (pHs-3C type). The total organic carbon (TOC) of the influent and effluent was analyzed using a total organic carbon analyzer (TOC-5000A, Shimadzu, Japan), and the mineralization rate (α, %) was calculated by Eq. (1).

$$\upalpha = \frac{{{\text{TOC}}_{0} - {\text{TOC}}}}{{{\text{TOC}}_{0} }} \times 100\% ,$$
(1)

where TOC0 is the initial concentration and TOC is the effluent concentration.

The absorbance of FA was measured at a wavelength of 254 nm using an Ultraviolet–Visible spectrophotometer (Hitachi U-3900). The removal rate (β, %) of the FA is calculated by Eq. (2).

$${{\upbeta }} = \frac{{{\text{UV}}_{{254\left( 0 \right)}} - {\text{UV}}_{{254}} }}{{{\text{UV}}_{{254\left( 0 \right)}} }} \times 100\% .$$
(2)

The color number of FA can be calculated and analyzed by UV–Vis44. Since the visible region of the leachate spectrum showed no limited absorption maxima, the colour number (CN) defined by Eq. (3) was used to characterize the colour. CN relies on the measurement of the spectral absorption coefficient (SAC) in the visible range at wavelengths of 436, 525 and 620 nm. SAC is determined by the absorption value (Abs) divided by a cell of thickness x, which is shown in Eq. (4).

$$CN = \frac{{SAC_{{436}}^{2} + SAC_{{525}}^{2} + SAC_{{620}}^{2} }}{{SAC_{{436}} + SAC_{{525}} + SAC_{{620}} }},$$
(3)
$$SAC_{i} = \frac{{Abs_{i} }}{x}.$$
(4)

The CN removal rate (γ, %) of FA is calculated by Eq. (5).

$$\gamma = \frac{{CN_{i} - CN_{0} }}{{CN_{i} }} \times 100\% .$$
(5)

Plackett–Burman and response surface methodology

The Design Expert Software (version 8.0) was used for experiment design and data analysis. Plackett–Burman (PB) coupled with central composite design response surface methodology (CCD-RSM) were used to evaluate and optimize the impact factors. Taking the removal rate of FA as the response target, two steps were used to optimize the experimental factors: Firstly, the PB experimental design was used to select the factors that significantly influenced the response, then CCD of RSM is used to find the optimum experimental conditions. During the experiment period, the CWPO was performed to degrade FA according to the designed conditions, the absorbance values of the FA samples before and after the reaction were measured at a wavelength of 254 nm, and the removal rate of the target fulvic acid was calculated by Eq. (2). After the experiment, the second-order response surface model equation was obtained and the optimal experimental conditions were determined by PB and CCD-RSM, and finally the optimal conditions were verified by conducting three repetitive tests.

Plackett–Burman experiment

The Plackett–Burman experimental design was proposed by Plackett and Burman in 1946. It was based on the principle of incompletely balanced plates. At most (N − 1) variables (N was generally a multiple of 4) could be studied by N experiments45. During the experiment, dummy variables are usually reserved as error analysis. Each variable has two levels, high and low, marked as (+) and (−), respectively.

The Plackett–Burman design of the CWPO experiment was shown in Table 2. The seven main factors of the FA degradation experiment were screened, plus four dummy variables. Each variable was determined at two levels (+) and (−), and a total of 12 experiments were conducted to determine the impact factors.

Table 2 Factors of PB design experiment.
Full size table

Response surface methodology

The Central Composite Design (CCD) developed by Box and Wilson is a commonly used response surface design method, with which an optimal fitted model can be obtained with minimum numbers of experiments. The second-order empirical model is generally used to characterize the response behavior of variables.

$$Y = \beta _{0} + \sum\limits_{{i = 1}}^{k} {\beta _{{\text{i}}} X_{i} } + \sum\limits_{{i = 1}}^{{j - 1}} {\sum\limits_{{j = 1}}^{k} {\beta _{{{\text{ij}}}} } } X_{i} X_{j} + \sum\limits_{{i = 1}}^{k} {\beta _{{{\text{ii}}}} } X_{i}^{2} ,$$

where: Y represents the system response; β0, βi, βii are the offset term, linear offset and second-order offset coefficient, respectively; βij is the interaction coefficient; Xi is the horizontal value of each factor.

Comparative experiment of catalytic performance

In order to better understand the role of the catalyst (ZVC/CTS-ACB) in the CWPO process, the UV254, TOC and CN removal efficiency of FA was investigated in various CWPO systems.

Results and discussion

Characterization of ZVC/CTS-ACB

The composition of the ZVC/CTS-ACB was characterized by X-ray diffraction (XRD) and X-ray photoelectron spectrometry (XPS). The diffraction peaks at 43.46°, 50.56° and 74.31° match well with the standard pattern of zero-valent copper (JCPDS 85-1326) (Fig. 1a)46. The crystalline grain size of Cu was calculated as 7.76 nm by the Williamson–Hall method. The Cu2p XPS spectrum (Fig. 1b) showed the peak of CuO at 943.3 eV as well as peaks of Cu0 at 931.7 eV and 951.6 eV46. All results showed that zero-valent copper existed in the catalyst. The absorption bands for ZVC/CTS-ACB (Fig. 1c) are 925.8, 1099.4 and 3437.1 cm−1, which correspond to C–O–C stretching, C=O bending and O–H stretching. All of these indicate the carboxyl groups on ZVC/CTS-ACB47, which may facilitate electron transfer48. Furthermore, the scan electron microscopy (SEM) morphologies (Fig. 2) showed that the ZVC/CTS-ACB have porous network structure on the surface, and the cross-sectional view showed uniform distribution of copper microcrystalline particles in the interior. The BET results presented that the specific surface area, pore volume and pore size were 42.4 m2/g, 0.98 cm3/g and 1.9 nm, respectively.

Figure 1
figure 1

(a) XRD pattern of ZVC/CTS-ACB. (b) XPS of ZVC/CTS-ACB. (c) FT-IR of ZVC/CTS-ACB.

Full size image
Figure 2
figure 2

(a) SEM of outer surface is magnified 2000 times of ZVC/CTS-ACB. (b) SEM of outer surface is magnified 5000 times of ZVC/CTS-ACB. (c) SEM of inner cut surface is magnified 2000 times of ZVC/CTS-ACB. (d) SEM of inner cut surface is magnified 8000 times of ZVC/CTS-ACB.

Full size image

Plackett–Burman experimental design results and analysis

The results of 12 runs of the FA removal experiments are shown in Table 3. The analysis of variance (ANOVA) results are shown in Table 4 with a list of significant differences of each factor impact.

Table 3 PB design and response values.
Full size table
Table 4 ANOVA results.
Full size table

As shown in Table 4, the Model’s F-value of 51.46 proves the model is significant. There is only a very small chance (< 0.004) that the F-value of the model is the result of noise. In this case, A, G are significant model terms. Values greater than 0.1000 indicates the model terms are not significant. It could be concluded that only the temperature and acidity factors have significant effects on the target values among the seven factors considered. It has been reported that the higher the reaction temperature, the faster the reaction49, and that pH can affect the chemical reaction rate and the production of free radicals50. Therefore, temperature and acidity were selected for the following central combination design.

CCD optimization design results and response surface analysis

The CCD was conducted for the Temperature and acidity selected by Plackett-Buiman with other non-critical factors fixed: Initial volume = 250 ml, Initial concentration = 100 mg/l, Time = 60 min, H2O2 = 20 mmol, (ZVC/CTS-ACB) = 3 g/l. The CWPO degradation FA experiment was performed according to the designed conditions, and the target response value was calculated.

Model fitting and analysis of variance

Table 5 shows the conditions of 13 runs and the value of FA removal rates. Evaluation of data in this table provides a second-order polynomial to express the relationship between FA removal efficiency and the experimental parameters.

$$Y = 80.25 + 15.98A + 20.28B - 6.64A^{2} - 25.70B^{2} ,$$

where: Y is the target response value, i.e. the removal of the FA. A and B represent temperature and acidity, respectively.

Table 5 CCD experimental design table.
Full size table

Table 6 shows the results of the RSM model fitting in the form of an analysis of variance (ANOVA). According to the table, the high F-value (F-value = 29.21) and the very low probability values (P value < 0.0001) indicates the model obtained is highly significant. At the same time, the F-value of the missing term is 4.66, indicating that the missing term has no significant effect, so the established model can be referenced51. From the corresponding P-value, it was shown that among the tested variables, the B-acidity and A-temperature value had the greatest influence on the removal efficiency of FA, and the surface effect of factor B2 on the FA removal effect is significant.

Table 6 ANOVA for the regression quadratic model of CCD design.
Full size table

The correlation system R2 is an important reference for the degree of fit. When R2 tends to be unified, the fit of the empirical model of the actual data is better. Joglekar and Ma52 suggested that for a good fit of the model, R2 should be at least 0.80. The R2 value of this model was 0.9359, indicating that the regression model fits well with the experiment.

RSM analysis

Figure 3 is a response surface graph and its contour plot from the multiple regression Eq. (3). It can be seen that the 3D response surface graph presents an inverted “U” shape, and a “red peak” appears at the center of the right side, meaning that the target optimal response value is obtained near here53. In the case of a fixed temperature, the target response value first increases and then tends to be gradual, and finally decreases slowly. On the other hand, if the acidity value is fixed, the target response value increases slowly and slightly with the increasing temperature. This also demonstrated that the degree of influence of acidity on the target FA removal was greater than temperature. The shape of the contour plot indicates the nature and extent of the interaction. The regular elliptical nature of the contour plot shows significant interactions, while the near-circular nature of the contour plots shows less prominent or negligible interactions53. It can be seen from the contour plot in Fig. 3 that there is no strong interaction between the influencing factors A and B here, which is consistent with the results obtained by the ANOVA above (Table 6). From the contour plot, the optimal acid of the model could be obtained in the acid pH range of 3 to 5, whereas the optimal temperature could not be found, it might be over 90 °C, so the model did need further optimization.

Figure 3
figure 3

3D response surface graph and contour plot.

Full size image

Optimization of influencing factors

The main purpose of the optimization was to determine the optimal parameter values for maximizing the FA removal. Therefore, the maximum removal of FA was selected as the target value. Since the optimal temperature may emerge over 90 °C, the temperature range was enlarged from 60–90 to 60–130 °C. Then the optimum values were obtained, as shown in Fig. 4, whereby two optimization schemes were also obtained in Table 7. Since the conditions of the two schemes were very close, and the expected target values are not different from each other, the temperature and acidity were finally selected as 94 °C and pH 3.8 respectively for the convenience of the experimental conditions setting.

Figure 4
figure 4

Desirable slope for numerical optimization of the CWPO conditions.

Full size image
Table 7 RSM system optimized solution.
Full size table

Figure 5 is the optimized 3D surface graph and contour plots. Compared with Fig. 3, a mountain shape 3D surface graph was obtained and it was obvious that the maximum removal of FA (93.8633%) was at the peak of the mountain.

Figure 5
figure 5

Optimized 3D surface graph and contour plots.

Full size image

Results verification

To confirm whether the model is sufficient to predict the maximum FA removal efficiency, three repetitive tests were performed under the optimal operating condition as shown in Table 8. The average maximum FA removal efficiency produced by three replicate experiments was 93.02%, while the RSM optimal target response was 93.86%. The good agreement between the predicted results and the experimental results verifies the validity of the model and confirms that CCD-RSM is a powerful tool for optimizing the experimental factors.

Table 8 Optimum value of the process parameter for maximum efficiency.
Full size table

Removal efficiency under different systems and the stability of ZVC/CTS-ACB

The removal rates of FA, TOC, and CN expressed by α, β, and γ respectively were shown in Fig. 6. It can be seen from the figure that the addition of oxidant H2O2 is crucial. The experimental groups with H2O2 added had better removal rates of FA, TOC and CN compared to those without H2O2. Among the latter three groups, the groups with catalyst achieved a higher removal rate than the group without catalyst. The removal rates of the three indicators had increased, especially the removal rate of TOC had improved significantly, from 13.2 to 81.9%, which indicates that the addition of ZVC could greatly increase the mineralization rate of FA in the CWPO reaction. Under the same condition, the mineralization rate was nearly 2.5 times higher than that of the experimental group without ZVC. ZVC also has obvious effects on FA degradation and color removal. All the results show that the ZVC/CTS-ACB catalyst prepared in this research has high catalytic activity in FA removal and organic compound mineralization.

Figure 6
figure 6

Catalyst and oxidant controlled trials (FA = 100 mg/l, temperature = 94 °C, time = 90 min, pH 3.8, CTS-ACB and ZVC/CTS-ACB = 4 g/l, H2O2 = 20 mmol).

Full size image

The possible degradation mechanism might be: the reaction starts with the Cu0 and H2O2, producing Cu+ and OH-, Cu+ reacts with H2O2 producing ·OH and Cu2+. The generated ·OH could mineralize FA into small molecules, even to CO2 and H2O.

The stability of ZVC/CTS-ACB was investigated through FA degradation repeating experiments in which the catalyst was repeated five times. The FA removal decreased from 94.78 to 90.05% in five runs. The ICP analysis of the effluent in the first run and fifth run showed that, the leaching concentrations of Cu ion were 0 mg/l and 0.08 mg/l respectively, which demonstrate the recyclability and stability of ZVC/CTS-ACB.

Conclusion

This study demonstrates the applicability of the prepared ZVC/CTS-ACB to the degradation of FA in CWPO. The statistical tools Plackett–Buiman and central composite design coupled with the response surface model were used to analyze the experimental data and predict the optimal target response value. Under the optimal conditions given by the system, the experiment was repeated three times. The average removal rate of FA was 93.02%, which was very close to the predicted target response value of 93.86%, indicating that the model was established accurately. The experiment results confirm that the PB experiment and the CCD-RSM are suitable for optimizing the operating conditions in a multi-factor operating environment to obtain the maximum FA degradation rate. The comparison experimental results showed that the catalyst and oxidant were essential factors in the CWPO reaction. The ZVC/CTS-ACB catalyst could greatly increase the mineralization rate of FA. The TOC removal rate of 81.9% indicated that most FA were directly mineralized into CO2 and H2O during the CWPO process, in which ZVC/CTS-ACB showed the high catalytic activity. The low leaching rate of copper ion also showed that the catalyst had good stability. Further study will be conducted on the catalytic mechanism of ZVC/CTS-ACS to FA removal and application of the catalyst in leachate treatment with CWPO.