*Article* **Efficient Removal of Micropollutants by Novel Carbon Materials Using Nitrogen-Rich Bio-Based Metal-Organic Framework (MOFs) as Precursors**

**Yazi Meng, Xiang Li \* and Bo Wang**

School of Materials, Advanced Research Institute of Multidisciplinary Science, Beijing Institute of Technology, Beijing 100081, China

**\*** Correspondence: xiangli0369@bit.edu.cn

**Abstract:** Eliminating pharmaceuticals with trace concentrations in water is crucial in water purification. Developing an effective adsorbent for removing micropollutants from water has aroused great research interest. In this study, the feasibility of nitrogen-rich bio-based metal–organic framework (MOF)-derived carbon as an effective material to eliminate micropollutants from the water environment is discussed. A mixed ligand approach has been applied to synthesize IISERP-MOF27 successfully via the solvothermal method. Adenine, which is non-toxic, easily obtained, and cheap, was introduced into the structure. The novel heterogeneous porous carbon was produced by pyrolyzation with an extremely high surface area (SBET = 980.5 m2/g), which is 12.8 times higher than that of pristine MOFs. Studies show that the highest surface area and abundant mesoporous structures (Vpore = 0.496 cm3/g) can be obtained when the MOFs are pyrolyzed at 900 ◦C. The saturated adsorption amount for sulfamethylthiazole (SMX) over MOF-derived carbon can reach 350.90 mg/g with a fast initial adsorption rate of 315.29 (mg/g·min). By adding the second linker adenine as the precursor, the adsorption performance for SMX was made extremely better than that of traditional active carbon (AC) and pyrolyzed ZIF-8(ZIF-8-C), one of the most classic Zn-MOFs. The adsorption capacity calculated by the Langmuir model (R<sup>2</sup> = 0.99) for SMX over bio-C-900 was 4.6 and 13.3 times more than those of AC and ZIF-8-C, respectively. The removal percentage of six representative pharmaceuticals can be well correlated to the structural parameter log Kow of each pharmaceutical, indicating the hydrophobic interaction should be one of the major mechanisms for the adsorption in water. This study offers a strategy to develop novel carbon materials to remove pharmaceuticals.

**Keywords:** pharmaceuticals; metal–organic framework; adsorption; performance

### **1. Introduction**

In recent years, regular testing of micropollutants in tap water, river water, and municipal wastewater has received increasing attention [1]. This kind of pollutant includes antibiotics, anti-inflammatory medications, and a variety of chemicals [2]. They have adverse environmental impacts such as mutagenicity, carcinogenicity, aquatic toxicity, and other harmful effects on both humans and the ecosystem [3,4]. For instance, the frequently used antibiotic sulfamethoxazole (SMX) is difficult entirely eliminate in wastewater treatment plants (WWTPs), leading to the transformation of its metabolites and product residues into components of surface water [5]. It has been proven that prolonged exposure to such pollutants may have negative effects on individuals, such as liver failure and genetic damage [5,6]. Therefore, it is of great significance to remove PPCPs from aquatic ecosystems greenly and efficiently.

To remove these contaminants in water, several tertiary treatments have been used, such as photocatalytic degradation [7], the Fenton reaction [8], biodegradation [9], and membrane filtration [10]. Among these, adsorption technology, which is easy to operate and low-cost, has been considered a promising advanced water treatment process for eliminating

**Citation:** Meng, Y.; Li, X.; Wang, B. Efficient Removal of Micropollutants by Novel Carbon Materials Using Nitrogen-Rich Bio-Based Metal-Organic Framework (MOFs) as Precursors. *Water* **2022**, *14*, 3413. https://doi.org/10.3390/w14213413

Academic Editors: Dionysios (Dion) Demetriou Dionysiou, Yujue Wang and Huijiao Wang

Received: 29 September 2022 Accepted: 21 October 2022 Published: 27 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

micropollutants in wastewater. To date, a large number of adsorbents have been investigated to remove micropollutants, including porous carbon [11], cyclodextrin polymer [12], and zeolites [13]. The carbon-based ones among these adsorbents are considered major materials in the water treatment process because of their high water/thermal stability and surface areas [14]. However, according to previous studies, their removal efficiency of SMX still needs to be improved compared to traditional materials (including activated carbon). Normally, about 54.34 ± 2.35% of SMX can be removed by treatment at WWTPs, but SMX content ranging from 0.3 ng L−<sup>1</sup> to 783 ng L−<sup>1</sup> is still detected after removal [15].

In recent years, MOFs with high specific surface areas and high porosities have been considered promising precursors to constructing adsorbents in water [16]. MOFs derived from carbon have been employed for the adsorption of water pollutants [17]. For example, results show that heavy metals, herbicides, and organic dyes can be effectively removed by carbon materials derived from Fe-MOFs [14,18–20]. Due to the especially low boiling point of the Zn atom, derived porous carbon can be created after the evaporation of metal based on Zn-based MOFs, such as ZIF-8, MOF-5, etc. [21–23]. On the other hand, bio-MOFs such as IISERP-MOF27(bio-27) and IISERP-MOF26 [24] with biological linkers should be promising in constructing MOFs because of their non-toxic, easily available, and inexpensive properties [25]. However, to the best of our knowledge, Zn-MOFs with mixed linkers-derived carbon have not been reported yet.

In this study, for the first time, bio-27-derived carbon was synthesized for the removal of micropollutants from water. The major characterization shows that this bio-MOF-derived carbon has obvious heterogeneous pores ranging from 1 nm to 6 nm. The precursors were pyrolyzed at 500~1000 ◦C (bio-C-500~bio-C-1000). The adsorption capacities and initial adsorption rates were evaluated by analyzing the removal performance for SMX over different bio-MOFs. The kinetic data were fitted using the Elovich model. Langmuir and Freundlich's models were applied to compare the capacities. The operational parameters were studied. To further evaluate the adsorption performance for pharmaceuticals with diverse structures, the adsorption behaviors for six pharmaceuticals—ketoprofen (KP), antipyrine (AT), ibuprofen (IBU), chloramphenicol (CAP), paracetamol (PC), and sulfamethoxazole (SMX)—were studied.

### **2. Experimental Section**

### *2.1. Chemical Agents*

Zn(NO3)2·6H2O (CAS, 10196-18-6), dimethylformamide (DMF, 99%), and methanol (MeOH, CAS, 67-56-1) were purchased from Sinopharm (Beijing, China). Adenine (CAS, 73- 24-5) and terephthalic acid (CAS, 100-21-0) were purchased from Energy Chemical (Beijing, China) (>99%). Active carbon was purchased from Energy Chemical Co. Ltd. (CAS, 7440- 44-0). Sulfamethoxazole and other chemicals were all obtained from Sigma–Aldrich with a purity of >99%. Water was obtained from a Milli-Q system with 18.2 MΩ·cm−<sup>1</sup> . The target compounds can be found in Table 1. Kow, which is the ratio of the concentration of molecules in the octanol phase to their concentration in the aqueous phase, is widely used to describe the hydrophobicity of drugs [26]. A drug's greater hydrophobicity is indicated by a larger log Kow value [27]. *Water* **2022**, *14*, x FOR PEER REVIEW 3 of 15 *Water* **2022**, *14*, x FOR PEER REVIEW 3 of 15 **Table 1.** Pharmaceuticals used in this study.

**Full Name Abbra. Molecular Weight Log Kow Chemical Structures**

**Table 1.** Pharmaceuticals used in this study. **Table 1.** Pharmaceuticals used in this study.


ibuprofen IBU 205.1 0.45

chloramphenicol CAP 321.0 1.1

chloramphenicol CAP 321.0 1.1

paracetamol PC 180.1 1.58

paracetamol PC 180.1 1.58


*Water* **2022**, *14*, x FOR PEER REVIEW 3 of 15

*Water* **2022**, *14*, x FOR PEER REVIEW 3 of 15

*Water* **2022**, *14*, x FOR PEER REVIEW 3 of 15

ketoprofen KP 257.3 3.12

ketoprofen KP 257.3 3.12

antipyrine AT 189.1 −1.55

ketoprofen KP 257.3 3.12

antipyrine AT 189.1 −1.55

ibuprofen IBU 205.1 0.45

antipyrine AT 189.1 −1.55

**Full Name Abbra. Molecular Weight Log Kow Chemical Structures** 

**Full Name Abbra. Molecular Weight Log Kow Chemical Structures**

**Full Name Abbra. Molecular Weight Log Kow Chemical Structures**

**Full Name Abbra. Molecular Weight Log Kow Chemical Structures**

**Table 1.** Pharmaceuticals used in this study.

*Water* **2022**, *14*, x FOR PEER REVIEW 3 of 14

ketoprofen KP 257.3 3.12

**Table 1.** Pharmaceuticals used in this study.

**Table 1.** Pharmaceuticals used in this study.

**Table 1.** Pharmaceuticals used in this study.

**Table 1.** *Cont.*

#### paracetamol PC 180.1 1.58 *2.2. Synthesis of Bio-27 and Its Derivatives Bio-C Materials 2.2. Synthesis of Bio-27 and Its Derivatives Bio-C Materials*

*2.2. Synthesis of Bio-27 and Its Derivatives Bio-C Materials* Bio-27 and its derivatives were synthesized via a solvothermal method [24]. Zn(NO3)2·6H2O (260.86 mg, 0.87 mmol), adenine (AD; 117.38 mg, 0.87 mmol), and terephthalic acid (H2BDC; 72.60 mg, 0.435 mmol) were mixed with 10.8 mL DMF, H2O, and MeOH (v/v/v = 1:1:0.6) and dispersed by ultrasonic processing for 30 min. Then, a white dispersion was obtained. The mixed solution was transferred to a 50 mL Teflon-lined autoclave and kept in an oven at 120 °C for 48 h. DMF and methanol were used to wash the sample before it was dried at 60 °C. A specific amount of bio-27 was evenly distributed in the center of the quartz boat and heated by 5 °C min−1 for 3 h in nitrogen after 6 h of vacuum activation at 120 °C. The materials obtained at the different temperatures of T (T = 500~1000 °C) are named bio-C-T. As the furnace chamber cooled to room temperature, bio-C-T compounds were generated. Some basic characterization was conducted to select a suitable material for further application in removing pharmaceuticals from water. The schematic diagram can be found in Figure 1. Bio-27 and its derivatives were synthesized via a solvothermal method [24]. Zn(NO3)2·6H2O (260.86 mg, 0.87 mmol), adenine (AD; 117.38 mg, 0.87 mmol), and terephthalic acid (H2BDC; 72.60 mg, 0.435 mmol) were mixed with 10.8 mL DMF, H2O, and MeOH (v/v/v = 1:1:0.6) and dispersed by ultrasonic processing for 30 min. Then, a white dispersion was obtained. The mixed solution was transferred to a 50 mL Teflon-lined autoclave and kept in an oven at 120 ◦C for 48 h. DMF and methanol were used to wash the sample before it was dried at 60 ◦C. A specific amount of bio-27 was evenly distributed in the center of the quartz boat and heated by 5 ◦C min−<sup>1</sup> for 3 h in nitrogen after 6 h of vacuum activation at 120 ◦C. The materials obtained at the different temperatures of T (T = 500~1000 ◦C) are named bio-C-T. As the furnace chamber cooled to room temperature, bio-C-T compounds were generated. Some basic characterization was conducted to select a suitable material for further application in removing pharmaceuticals from water. The schematic diagram can be found in Figure 1. Bio-27 and its derivatives were synthesized via a solvothermal method [24]. Zn(NO3)2·6H2O (260.86 mg, 0.87 mmol), adenine (AD; 117.38 mg, 0.87 mmol), and terephthalic acid (H2BDC; 72.60 mg, 0.435 mmol) were mixed with 10.8 mL DMF, H2O, and MeOH (v/v/v = 1:1:0.6) and dispersed by ultrasonic processing for 30 min. Then, a white dispersion was obtained. The mixed solution was transferred to a 50 mL Teflon-lined autoclave and kept in an oven at 120 °C for 48 h. DMF and methanol were used to wash the sample before it was dried at 60 °C. A specific amount of bio-27 was evenly distributed in the center of the quartz boat and heated by 5 °C min−1 for 3 h in nitrogen after 6 h of vacuum activation at 120 °C. The materials obtained at the different temperatures of T (T = 500~1000 °C) are named bio-C-T. As the furnace chamber cooled to room temperature, bio-C-T compounds were generated. Some basic characterization was conducted to select a suitable material for further application in removing pharmaceuticals from water. The schematic diagram can be found in Figure 1.

### *2.3. Synthesis of ZIF-8 and Its Derivatized Carbon-Based Materials (ZIF-8-C)*

ZIF-8 was synthesized based on the previous report [28]. Zn(NO3)2·6H2O (743.8 mg, 2.5 mmol) and dimethylimidazole (821.0 mg,10 mmol) were mixed with 50 mL MeOH and put into a 100 mL Shuniu bottle, where they were and continuously and evenly stirred for 24 h at room temperature. After standing for precipitation, the materials were separated and washed three times with fresh MeOH. Then, the materials were dried at 60 ◦C.

### *2.4. Characterization*

The X-ray diffraction pattern was studied by a multipurpose, high-efficiency X-ray diffractometer (PXRD, Rigaku MiniFlEX600) (λ = 0.154 nm). The scan rate was 10◦/min at room temperature. The adsorption–desorption curve was measured by an automatic specific surface and pore analyzer (Quanta chrome ASiQMVH002-5). Materials were activated in vacuum at 120 ◦C for 12 h. The Brunel–Emmett–Teller (BET) equation was used for measuring the specific surface area. The chemical state was obtained by X-ray photoelectron spectroscopy (XPS, PHI QUANTERA-II SXM, ULVAC-PHI, USA), and the X-ray source was Al-Ka (1486.6 eV, line width 0.68 eV). The zeta potential of bio-C was tested using a nanoparticle size and zeta potential analyzer (DLS, Malvern Zetasizer Nano ZS90, Worcestershire, UK).

### *2.5. Adsorption Experiment*

The adsorption experiments for removing pharmaceuticals with two concentrations, 10 mg/L and 400 µg/L, were conducted separately. The solution was prepared with ultrapure water. Amounts equal to 4.5 mg of different adsorbents were added to a 50 mL glass vial containing 50 mL of a simulated solution of PPCPs at pH 7.0. All the batch adsorption experiments were performed on a multi-point stirrer at 500 rpm and 25 ◦C. At each time interval, the solution was sampled and then filtered with a PES filter. The adsorption experiments were conducted for 45 min. The pseudo-second-order kinetic model is shown in Equations (1) and (2), where Q<sup>t</sup> represents the adsorption amount at time t, and µg/g is the unit. Q<sup>e</sup> is the equilibrium adsorption amount, and µg/g is the unit. v<sup>0</sup> represents the initial adsorption rate, and µg/(g·min) is the unit. k is the intraparticle diffusion rate constant.

$$\frac{\text{t}}{\text{Q}\_{\text{t}}} = \frac{1}{\text{v}\_{0}} + \frac{\text{t}}{\text{Q} \text{e}} \tag{1}$$

$$\mathbf{v}\_0 = \mathbf{k} \times \mathbf{Q}\_t^2 \tag{2}$$

### *2.6. Instrumental Analysis for Pharmaceuticals*

Pharmaceuticals with concentrations of 10 mg/L were separated and detected using HPLC-UV with an Agilent 1260 system. The column was an Agilent SB-C18 column (2.7 µm, 4.6 mm × 150 mm). The mobile phase was water containing 0.1% formic acid and 1 mM ammonium acetate ((NH4)2AC) (60%), as well as MeOH (40%). The temperature in the column was 30 ◦C. The UV light's detecting wavelength was 270 nm. Triple quadrupole mass spectrometry (Agilent 1290, 6465 QQQ, Santa Clara, CA, USA) and ultra-performance liquid chromatography were used to measure the drugs with trace concentrations. The column was an Eclipse Plua C18 RRHD (2.1 × 50 mm 1.8 µm). The mobile phases were water with 0.1% formic acid and 2 mM ammonium acetate (A) and acetonitrile (B). The flow rate was set at 0.2 mL/min. The dilute gradient was 0–2 min (10% B), 2–10 min (10~45% B), 10~13 min (45~90% B), 13~14 min (90% B), and 14.1 min (10% B). The injection volume was 20 µL. The flow rate was 0.2 mL/min. The peak areas of each pharmaceutical were quantitively analyzed by the Quant-My-Way software, developed by Agilent.

### **3. Results and Discussion**

### *3.1. Characterizations of Bio-C-T*

### 3.1.1. Structural and Porous Properties

As shown in Figure 2a, the PXRD patterns of bio-27 were consistent with the simulation, with the main characteristic peaks at 5◦ , 9◦ , and 15◦ [24], indicating the metal–organic framework (MOF) called bio-27 was synthesized successfully. The PXRD pattern of bio-C-T can be found in Figure 2b, indicating that all the synthesized carbons are in an amorphous state. It is to be noted that bio-C-800, 900, and 1000 showed two broad peaks at 24◦ and 43◦ , corresponding to the diffraction peaks of the (002) crystal plane and (100) crystal plane, respectively [29]. The observed peaks move slightly to smaller angles with increased activation temperatures, indicating the distance of the interlayer could gradually increase. According to previous studies, this shift could be caused by defects during carbonization [30]. However, bio-C-500 may not be carbonized completely based on the positions of characteristic peaks in the PXRD spectrum. On the other hand, the colors of bio-C-800, bio-C-900, and bio-C-1000 are black, consistent with the color of carbon black. This could also confirm the completeness of the carbonization of the four materials. The nitrogen adsorption isotherm in Figure 2c reveals a mixture of type I and IV isotherms. Obviously, the adsorption curve of bio-C-900 has a hysteresis loop, indicating that the material has a hierarchical porous structure and a relatively more mesoporous structure. creased activation temperatures, indicating the distance of the interlayer could gradually increase. According to previous studies, this shift could be caused by defects during carbonization [30]. However, bio-C-500 may not be carbonized completely based on the positions of characteristic peaks in the PXRD spectrum. On the other hand, the colors of bio-C-800, bio-C-900, and bio-C-1000 are black, consistent with the color of carbon black. This could also confirm the completeness of the carbonization of the four materials. The nitrogen adsorption isotherm in Figure 2c reveals a mixture of type I and IV isotherms. Obviously, the adsorption curve of bio-C-900 has a hysteresis loop, indicating that the material has a hierarchical porous structure and a relatively more mesoporous structure.

As shown in Figure 2a, the PXRD patterns of bio-27 were consistent with the simulation, with the main characteristic peaks at 5°, 9°, and 15° [24], indicating the metal–organic framework (MOF) called bio-27 was synthesized successfully. The PXRD pattern of bio-C-T can be found in Figure 2b, indicating that all the synthesized carbons are in an amorphous state. It is to be noted that bio-C-800, 900, and 1000 showed two broad peaks at 24° and 43°, corresponding to the diffraction peaks of the (002) crystal plane and (100) crystal plane, respectively [29]. The observed peaks move slightly to smaller angles with in-

*Water* **2022**, *14*, x FOR PEER REVIEW 5 of 14

**3. Results and Discussion**  *3.1. Characterizations of Bio-C-T* 

3.1.1. Structural and Porous Properties

**Figure 2.** (**a**) PXRD patterns of IISERP−MOF−27 and its (**b**) derived carbon materials, which are pyrolyzed at temperatures ranging from 500 °C to 1000 °C. (**c**) The nitrogen adsorption and desorption isotherms of bio−27 and bio−C−T. (**d**) The pore size distribution of the as-prepared samples of bio−27 and bio−C−T (T = 500~1000 °C). **Figure 2.** (**a**) PXRD patterns of IISERP−MOF−27 and its (**b**) derived carbon materials, which are pyrolyzed at temperatures ranging from 500 ◦C to 1000 ◦C. (**c**) The nitrogen adsorption and desorption isotherms of bio−27 and bio−C−T. (**d**) The pore size distribution of the as-prepared samples of bio−27 and bio−C−T (T = 500~1000 ◦C).

In comparison, the hysteresis loops of bio-C-800 and bio-C-1000 are not obvious, and the adsorption curves are more in line with the type I adsorption isotherm. The surface area and pore volume show an upward trend with increased carbonization temperature. Still, the specific surface area of bio-C-1000 was lower than that of bio-C-900, which may In comparison, the hysteresis loops of bio-C-800 and bio-C-1000 are not obvious, and the adsorption curves are more in line with the type I adsorption isotherm. The surface area and pore volume show an upward trend with increased carbonization temperature. Still, the specific surface area of bio-C-1000 was lower than that of bio-C-900, which may be due to the partial collapse of the mesoporous structure at high temperature due to the complete evaporation of Zn atoms (Table 2).

**Table 2.** Calculated surface area (SBET) and porous structures.


The pore size distribution curve calculated in the microporous region (Figure 2d) shows that all catalysts are dominated by micropores with a pore size of 0.8 nm. Although the microporous area of bio-C-500 accounts for a small proportion, the pore volume of bio-C-500 is higher than that of bio-C-800 due to the wider distribution of the mesoporous area. The surface area of the pure MOF (bio-27) is 76.4 m2/g, as shown in Table 2. However, bio-27-C has much higher surface areas ranging from 204.2 m2/g to 980.4 m2/g. shows that all catalysts are dominated by micropores with a pore size of 0.8 nm. Although the microporous area of bio-C-500 accounts for a small proportion, the pore volume of bio-C-500 is higher than that of bio-C-800 due to the wider distribution of the mesoporous area. The surface area of the pure MOF (bio-27) is 76.4 m2/g, as shown in Table 2. However, bio-27-C has much higher surface areas ranging from 204.2 m2/g to 980.4 m2/g. On the other hand, the pore volume of bio-27-C-900 is the greatest, with a value of

The pore size distribution curve calculated in the microporous region (Figure 2d)

be due to the partial collapse of the mesoporous structure at high temperature due to the

**Materials SBET (m2/g) Vtotal (cm3/g) Pore Width (nm)**  Bio-27 76.5 0.054 2.890 Bio-27-500 90.2 0.160 3.698 Bio-27-800 204.2 0.108 0.783 Bio-27-900 980.4 0.496 0.852 Bio-27-1000 605.3 0.294 0.783

*Water* **2022**, *14*, x FOR PEER REVIEW 6 of 14

**Table 2.** Calculated surface area (SBET) and porous structures.

complete evaporation of Zn atoms (Table 2).

On the other hand, the pore volume of bio-27-C-900 is the greatest, with a value of 0.496 cm3/g. The pore width of the bio-C-900 is the greatest, with a value of 0.852 nm, among all the materials. The well-developed pore structure in the adsorbent may provide more adsorption sites for SMX, which should be beneficial for reducing diffusion resistance. 0.496 cm3/g. The pore width of the bio-C-900 is the greatest, with a value of 0.852 nm, among all the materials. The well-developed pore structure in the adsorbent may provide more adsorption sites for SMX, which should be beneficial for reducing diffusion resistance.

#### 3.1.2. Morphology of the Samples 3.1.2. Morphology of the Samples

In Figure 3a,b, the precursor bio-27 has an irregular two-dimensional sheet structure, but the thickness is too large. With increasing pyrolysis temperatures, the edge morphology of bio-C-500 becomes irregular (Figure 3c), but it can be observed that it exists in the form of sheet-like stacking, and the scale becomes thinner. While bio-C-800, bio-C-900, and bio-C-1000 (Figure 3d–f) exhibit changes in longitudinal scale, the edges are more regular, and the ordered sheet-like stacking structure is favorable for the mass transfer process [31]. The jagged edges of bio-C-1000 may be due to the collapse of the internal structure. HRTEM was used to further characterize the morphology and microstructure of the catalyst. In the bright field image (Figure 3g), it can be observed that the surface of the catalyst with sheet-like morphology has uniformly distributed mesopores and only short-range ordered ones. N-doped carbon fringes do not exhibit any long-range ordered lattice fringes, indicating no crystalline species in the catalyst, which is mutually confirmed by the results of PXRD, as discussed above. The results show that the prepared bio-C-900 material has abundant pore structures. In Figure 3a,b, the precursor bio-27 has an irregular two-dimensional sheet structure, but the thickness is too large. With increasing pyrolysis temperatures, the edge morphology of bio-C-500 becomes irregular (Figure 3c), but it can be observed that it exists in the form of sheet-like stacking, and the scale becomes thinner. While bio-C-800, bio-C-900, and bio-C-1000 (Figure 3d–f) exhibit changes in longitudinal scale, the edges are more regular, and the ordered sheet-like stacking structure is favorable for the mass transfer process [31]. The jagged edges of bio-C-1000 may be due to the collapse of the internal structure. HRTEM was used to further characterize the morphology and microstructure of the catalyst. In the bright field image (Figure 3g), it can be observed that the surface of the catalyst with sheet-like morphology has uniformly distributed mesopores and only short-range ordered ones. N-doped carbon fringes do not exhibit any long-range ordered lattice fringes, indicating no crystalline species in the catalyst, which is mutually confirmed by the results of PXRD, as discussed above. The results show that the prepared bio-C-900 material has abundant pore structures.

**Figure 3.** (**a**–**f**) SEM images of bio−27 and bio−C−T. (**g**,**h**) HRTEM of bio−C−900. **Figure 3.** (**a**–**f**) SEM images of bio−27 and bio−C−T. (**g**,**h**) HRTEM of bio−C−900.

3.1.3. Fourier Transform Infrared (FTIR) Spectrum of Bio-27-Derived Carbon

Analysis via FTIR spectroscopy was carried out in order to investigate the various surface functional groups. According to Figure 4, the stretching vibrations of O-H and N-H can be found at around 3400 cm−<sup>1</sup> [32]. The band at 2100 cm−<sup>1</sup> is ascribed to the C=C=C stretching vibration [33]. The C=O stretching vibration is responsible for the band at 1650 cm−<sup>1</sup> [32], while the N-H in-plane and out-of-plane bending vibrations are responsible for the bands at 1580 cm−<sup>1</sup> and 850 cm−<sup>1</sup> , respectively [32]. Furthermore, the bands at 1380 cm−<sup>1</sup> and 1070 cm−<sup>1</sup> are attributed to C-O stretching vibrations and C-N stretching vibrations [32,34], respectively, indicating the existence of nitrogen and oxygen functional groups on bio-C.

3.1.3. Fourier Transform Infrared (FTIR) Spectrum of Bio-27-Derived Carbon

Analysis via FTIR spectroscopy was carried out in order to investigate the various surface functional groups. According to Figure 4 , the stretching vibrations of O-H and N-H can be found at around 3400 cm−1 [32]. The band at 2100 cm−1 is ascribed to the C=C=C stretching vibration [33]. The C=O stretching vibration is responsible for the band at 1650 cm−1 [32], while the N-H in-plane and out-of-plane bending vibrations are responsible for the bands at 1580 cm−1 and 850 cm−1, respectively [32]. Furthermore, the bands at 1380 cm−<sup>1</sup> and 1070 cm−1 are attributed to C-O stretching vibrations and C-N stretching vibrations [32,34], respectively, indicating the existence of nitrogen and oxygen functional groups on

**Figure 4.** Fourier transform infrared (FTIR) spectrum of IISERP−MOF−27−derived carbon which is pyrolyzed at temperatures ranging from 500 °C to 1000 °C. **Figure 4.** Fourier transform infrared (FTIR) spectrum of IISERP−MOF−27−derived carbon which is pyrolyzed at temperatures ranging from 500 ◦C to 1000 ◦C.

#### *3.2. Adsorption Experiments 3.2. Adsorption Experiments*

bio-C.

### 3.2.1. Pseudo-Second-Order Adsorption Kinetics of SMX

3.2.1. Pseudo-Second-Order Adsorption Kinetics of SMX Within a contact time of 45 min, the impact of bio-C-T materials at various pyrolysis temperatures on SMX adsorption was examined. The results are displayed in Figure 5a. To fit the experimental data, a pseudo-second-order kinetic model was applied, and the relevant fit plots were produced. The model compound's adsorption of the adsorbent is indicated by the variables Qt and Qe in the equation at reaction time t (min)—when the reaction achieves saturation equilibrium. Qe and v0 represent the adsorption amount and the initial adsorption rate (mg/g·min), respectively. According to Table 3, bio-C-900 has the highest Qe (100.85 mg/g) and the best adsorption capacity of SMX among the series of bio-C-T materials. In addition, its v0 (315.29 mg/g·min) is the highest among all the prepared materials. The mass transfer resistance of bio-C-900 to SMX is substantially lower than that of the manufactured bio-C at other temperatures, and its adsorption sites are more readily accessible to SMX. Additionally, its v0 is more than 30 times that of bio-C-1000. We therefore hypothesize that surface area has a close relationship with the adsorp-Within a contact time of 45 min, the impact of bio-C-T materials at various pyrolysis temperatures on SMX adsorption was examined. The results are displayed in Figure 5a. To fit the experimental data, a pseudo-second-order kinetic model was applied, and the relevant fit plots were produced. The model compound's adsorption of the adsorbent is indicated by the variables Q<sup>t</sup> and Q<sup>e</sup> in the equation at reaction time t (min)—when the reaction achieves saturation equilibrium. Q<sup>e</sup> and v<sup>0</sup> represent the adsorption amount and the initial adsorption rate (mg/g·min), respectively. According to Table 3, bio-C-900 has the highest Q<sup>e</sup> (100.85 mg/g) and the best adsorption capacity of SMX among the series of bio-C-T materials. In addition, its v<sup>0</sup> (315.29 mg/g·min) is the highest among all the prepared materials. The mass transfer resistance of bio-C-900 to SMX is substantially lower than that of the manufactured bio-C at other temperatures, and its adsorption sites are more readily accessible to SMX. Additionally, its v<sup>0</sup> is more than 30 times that of bio-C-1000. We therefore hypothesize that surface area has a close relationship with the adsorption amount of SMX in this study. A higher surface area should favor a higher adsorption amount.


tion amount of SMX in this study. A higher surface area should favor a higher adsorption

amount. **Table 3.** Pseudo-second-order adsorption kinetics fitting values.

On the other hand, large pore sizes could enhance the adsorption kinetics [35]. Similar results can be found in the adsorption process by other porous material [36]. For example, Guo found that the introduction of mesopores above 2 nm can greatly increase the pore volume. The graphene oxide/carbon composite nanofibers with abundant mesopores prepared in this study achieved improved adsorption capacity for volatile organic compounds (VOCs), and the highest adsorption capacity values for benzene and butanone reached 83.2 cm<sup>3</sup> g <sup>−</sup><sup>1</sup> and 130.5 cm<sup>3</sup> g −1 , respectively. According to Table 2, we can know that the

pore volume of bio-C-T-series materials was also greatly improved after the appearance of mesopores, which is consistent with the experimental results. *Water* **2022**, *14*, x FOR PEER REVIEW 8 of 14

**Figure 5.** (**a**) Adsorption kinetics curves of bio−C−T. (**b**) The corresponding fitted pseudo-secondorder adsorption kinetics. Reaction conditions: SMX = 10 mg/L, reaction volume = 50 mL, catalyst dosage = 180 mg/L, initial solution pH = 7, temperature = 25 °C, reaction time = 45 min). **Figure 5.** (**a**) Adsorption kinetics curves of bio−C−T. (**b**) The corresponding fitted pseudo-secondorder adsorption kinetics. Reaction conditions: SMX = 10 mg/L, reaction volume = 50 mL, catalyst dosage = 180 mg/L, initial solution pH = 7, temperature = 25 ◦C, reaction time = 45 min).

#### **Table 3.** Pseudo-second-order adsorption kinetics fitting values. 3.2.2. Effect of Reaction Time

**Materials k (g/mg·min) Qe (mg/g) v0 (mg/g·min) R2** Bio-27-500 —— 1.65 —— 0.99 Bio-27-800 0.0059 27.27 4.38 0.98 Bio-27-900 0.031 100.85 315.29 0.93 Bio-27-1000 0.0021 70.79 10.52 0.98 On the other hand, large pore sizes could enhance the adsorption kinetics [35]. Similar results can be found in the adsorption process by other porous material [36]. For example, Guo found that the introduction of mesopores above 2 nm can greatly increase the The elimination of SMX was compared over bio-C-900, AC, and ZIF-8 (Zn as the metal center). The ZIF-8-C obtained after carbonization at the same temperature was applied in an adsorption experiment for removing SMX with an initial concentration of 10 ppm for 60 min. As shown in Figure 6a, bio-C-900 showed ultra-fast adsorption behavior, with 55% removal within only 15 s. A higher removal rate of 90% could be achieved within only 10 min. After reaching adsorption saturation, the removal rate of SMX by bio-C-900 can reach 98.7%. In comparison, the removal rates for SMX over AC and ZIF-8-C were 36.5% and 9.7%, respectively. Two models, the Elovich model and the second-order kinetic model, were applied in Table 4. The Elovich kinetic fitting formula (Equation (3)) is as follows:

$$\mathbf{Q}\_t = \mathbf{A} + \mathbf{b} \times \ln \mathbf{t} \tag{3}$$

pounds (VOCs), and the highest adsorption capacity values for benzene and butanone reached 83.2 cm3 g−1 and 130.5 cm3 g−1, respectively. According to Table 2, we can know that the pore volume of bio-C-T-series materials was also greatly improved after the apwhere A and b are both Elovich constants. The fitting degree of the two models is very high, indicating that the adsorption process of the adsorbent includes both chemisorption with electron transfer and electron transfer during the pseudo-second-order kinetic process. *Water* **2022**, *14*, x FOR PEER REVIEW 9 of 14

pearance of mesopores, which is consistent with the experimental results.

Q୲ = A + b ൈ ln t (3) where A and b are both Elovich constants. The fitting degree of the two models is very high, indicating that the adsorption process of the adsorbent includes both chemisorption with electron transfer and electron transfer during the pseudo-second-order kinetic pro-**Figure 6.** (**a**) Adsorption kinetics curves of all materials for SMX in this study; (**b**) The corresponding pseudo-second-order adsorption kinetic model and Elovich kinetic model were fitted. Reaction conditions: initial concentration of SMX = 10 mg/L, reaction volume = 50 mL, catalyst dosage = 4.5 mg, initial solution pH = 7, temperature = 25 °C, reaction time = 45 min. **Figure 6.** (**a**) Adsorption kinetics curves of all materials for SMX in this study; (**b**) The corresponding pseudo-second-order adsorption kinetic model and Elovich kinetic model were fitted. Reaction conditions: initial concentration of SMX = 10 mg/L, reaction volume = 50 mL, catalyst dosage = 4.5 mg, initial solution pH = 7, temperature = 25 ◦C, reaction time = 45 min.

The amount of adsorption increases with increasing initial concentration and reaches saturation for bio-C-900, AC, and ZIF-8-C, as well as for adsorbents over the studied concentration (Figure 7). At initial concentration of 20 mg/L, 86.6% of SMX could be removed by the bio-C-900 sorbent. When the initial concentrations were 40, 50, 60, and 80 mg/L, SMX was partially removed with efficiencies of 58.3%, 53.1%, 44.8%, and 32.7%, respectively. Such a phenomenon can be explained by the limited adsorption sites of the same adsorbent. The adsorption capacity increases with a higher concentration of adsorbate. This demonstrates that the initial concentration should limit the adsorption amount because the adsorption sites are sufficient [37]. In addition, the data were fitted using the

Qୣ = Q ൈ

Qୣ = KCୣ

ైେ

భ

ଵାైେ (4)

(5)

cess.

3.2.3. Effect of Initial Concentration

Langmuir and Freundlich equations.

**Table 4.** Langmuir and Freundlich adsorption isotherm model.

**Sample Qmax (mg/g) KL (L/mg) R2 KF (mg g−1(L mg−1)1/n) n R2** Bio-C-900 350.90 0.061 0.99 53.23 2.47 0.96 AC 76.96 0.025 0.98 29.79 4.03 0.82 ZIF-8-C 26.44 0.027 0.99 10.88 4.32 0.88


**Figure 6.** (**a**) Adsorption kinetics curves of all materials for SMX in this study; (**b**) The corresponding pseudo-second-order adsorption kinetic model and Elovich kinetic model were fitted. Reaction conditions: initial concentration of SMX = 10 mg/L, reaction volume = 50 mL, catalyst dosage = 4.5 mg,

**Table 4.** Langmuir and Freundlich adsorption isotherm model. **Table 4.** Langmuir and Freundlich adsorption isotherm model.

initial solution pH = 7, temperature = 25 °C, reaction time = 45 min.

*Water* **2022**, *14*, x FOR PEER REVIEW 9 of 14

3.2.3. Effect of Initial Concentration 3.2.3. Effect of Initial Concentration

The amount of adsorption increases with increasing initial concentration and reaches saturation for bio-C-900, AC, and ZIF-8-C, as well as for adsorbents over the studied concentration (Figure 7). At initial concentration of 20 mg/L, 86.6% of SMX could be removed by the bio-C-900 sorbent. When the initial concentrations were 40, 50, 60, and 80 mg/L, SMX was partially removed with efficiencies of 58.3%, 53.1%, 44.8%, and 32.7%, respectively. Such a phenomenon can be explained by the limited adsorption sites of the same adsorbent. The adsorption capacity increases with a higher concentration of adsorbate. This demonstrates that the initial concentration should limit the adsorption amount because the adsorption sites are sufficient [37]. In addition, the data were fitted using the Langmuir and Freundlich equations. The amount of adsorption increases with increasing initial concentration and reaches saturation for bio-C-900, AC, and ZIF-8-C, as well as for adsorbents over the studied concentration (Figure 7). At initial concentration of 20 mg/L, 86.6% of SMX could be removed by the bio-C-900 sorbent. When the initial concentrations were 40, 50, 60, and 80 mg/L, SMX was partially removed with efficiencies of 58.3%, 53.1%, 44.8%, and 32.7%, respectively. Such a phenomenon can be explained by the limited adsorption sites of the same adsorbent. The adsorption capacity increases with a higher concentration of adsorbate. This demonstrates that the initial concentration should limit the adsorption amount because the adsorption sites are sufficient [37]. In addition, the data were fitted using the Langmuir and Freundlich equations.

$$\mathbf{Q\_e} = \mathbf{Q\_{max}} \times \frac{\mathbf{K\_L C\_e}}{1 + \mathbf{K\_L C\_e}} \tag{4}$$

$$\mathbf{Q\_e = K\_F C\_e^{\frac{1}{n}}} \tag{5}$$

**Figure 7.** Adsorption isotherm for the removal of SMX over the three materials: active carbon (AC), ZIF−8−C, and bio−C−900. Reaction conditions: initial concentration of SMX = 5~80 mg/L, initial solution pH = 7, temperature = 25 ◦C, reaction time = 600 min.

The adsorption capacity of SMX over bio-C-900 was 4.6 and 13.3 times higher than those of AC and ZIF-8-C, respectively. The K<sup>L</sup> value indicates that SMX is more easily removed by bio-C-900, confirming that the prepared bio-C-900 could have relatively better adsorption properties than those of conventional materials. The adsorption of SMX over bio-C-900 tends to be homogeneous monolayer adsorption [38] because the Langmuir model linear fitting result (R<sup>2</sup> = 0.99) is slightly better than that in the Freundlich model (R<sup>2</sup> = 0.96). Therefore, the material tends towards homogeneous monolayer adsorption. In addition, the maximum adsorption capacity of SMX over bio-C-900 (350.90 mg/g) was 3.6 times higher than that over AC (76.96 mg/g), which further proved that bio-C-900 is conducive to the transport of substrates on its surface and pores. The heterogeneous multilayer adsorption effectively improves the adsorption capacity and adsorption kinetics. The specific fitting data are shown in Table 4 below.

### *3.3. Exploring the Adsorption Mechanism for Micropollutants* evaluated in Figure 8a. As seen in Figure 8, the removal percentages of SMX varied sig-

ics. The specific fitting data are shown in Table 4 below.

*3.3. Exploring the Adsorption Mechanism for Micropollutants* 

*Water* **2022**, *14*, x FOR PEER REVIEW 10 of 14

lution pH = 7, temperature = 25 °C, reaction time = 600 min.

**Figure 7.** Adsorption isotherm for the removal of SMX over the three materials: active carbon (AC), ZIF−8−C, and bio−C−900. Reaction conditions: initial concentration of SMX = 5~80 mg/L, initial so-

The adsorption capacity of SMX over bio-C-900 was 4.6 and 13.3 times higher than those of AC and ZIF-8-C, respectively. The KL value indicates that SMX is more easily removed by bio-C-900, confirming that the prepared bio-C-900 could have relatively better adsorption properties than those of conventional materials. The adsorption of SMX over bio-C-900 tends to be homogeneous monolayer adsorption [38] because the Langmuir model linear fitting result (R2 = 0.99) is slightly better than that in the Freundlich model (R2 = 0.96). Therefore, the material tends towards homogeneous monolayer adsorption. In addition, the maximum adsorption capacity of SMX over bio-C-900 (350.90 mg/g) was 3.6 times higher than that over AC (76.96 mg/g), which further proved that bio-C-900 is conducive to the transport of substrates on its surface and pores. The heterogeneous multi-layer adsorption effectively improves the adsorption capacity and adsorption kinet-

To gain more insight into the reaction mechanism, the effect of solution pH (3~9) was

To gain more insight into the reaction mechanism, the effect of solution pH (3~9) was evaluated in Figure 8a. As seen in Figure 8, the removal percentages of SMX varied significantly with initial pH values. SMX is an amphiphilic molecule with two pK<sup>a</sup> values (pKa1 = 1.8 and pKa2 = 5.6) [39]. When the solution pH is higher, SMX is prone to exist in its deprotonated form with a negative charge. The zeta potential can be found in Figure 8b. As a result, the adsorption amount decreased obviously, probably because of electrostatic expulsion. However, in an acid environment, SMX is usually protonated, with a positive charge, which could lead to a decrease in electrostatic interaction. However, the adsorption performance for the removal of SMX is higher at pH values ranging from 3 to 5. Thus, several major mechanisms could co-exist to contribute to the high adsorption amount. For example, the solubility of SMX is lower in an acid environment, and it is very slightly soluble in water but is soluble in alkali hydroxides, as previously reported [40,41]. The properties of hydrophilic interaction should be considered. nificantly with initial pH values. SMX is an amphiphilic molecule with two pKa values (pKa1 = 1.8 and pKa2 = 5.6) [39]. When the solution pH is higher, SMX is prone to exist in its deprotonated form with a negative charge. The zeta potential can be found in Figure 8b. As a result, the adsorption amount decreased obviously, probably because of electrostatic expulsion. However, in an acid environment, SMX is usually protonated, with a positive charge, which could lead to a decrease in electrostatic interaction. However, the adsorption performance for the removal of SMX is higher at pH values ranging from 3 to 5. Thus, several major mechanisms could co-exist to contribute to the high adsorption amount. For example, the solubility of SMX is lower in an acid environment, and it is very slightly soluble in water but is soluble in alkali hydroxides, as previously reported [40,41]. The properties of hydrophilic interaction should be considered.

**Figure 8.** (**a**) Effect of pH on SMX removal. (**b**) Zeta-potential of bio−C−900. **Figure 8.** (**a**) Effect of pH on SMX removal. (**b**) Zeta-potential of bio−C−900.

To further evaluate the adsorption performance for pharmaceuticals with diverse structures, the adsorption behavior for six pharmaceuticals—ketoprofen (KP), antipyrine (AT), ibuprofen (IBU), chloramphenicol (CAP), paracetamol (PC), and sulfamethoxazole (SMX)—were studied. Obviously, we found that the adsorption performance (removal percentages within 45 min reaction time) correlated well (R2 = 0.97) with the physical– chemical parameter log Kow of each pharmaceutical. The log Kow values of ketoprofen (KP), antipyrine (AT), ibuprofen (IBU), chloramphenicol (CAP), paracetamol (PC), and To further evaluate the adsorption performance for pharmaceuticals with diverse structures, the adsorption behavior for six pharmaceuticals—ketoprofen (KP), antipyrine (AT), ibuprofen (IBU), chloramphenicol (CAP), paracetamol (PC), and sulfamethoxazole (SMX)—were studied. Obviously, we found that the adsorption performance (removal percentages within 45 min reaction time) correlated well (R<sup>2</sup> = 0.97) with the physical– chemical parameter log Kow of each pharmaceutical. The log Kow values of ketoprofen (KP), antipyrine (AT), ibuprofen (IBU), chloramphenicol (CAP), paracetamol (PC), and sulfamethoxazole (SMX) were 3.12, −1.55, 0.45, 1.1, 1.58, and 0.89 [42–45], respectively. Tung Xuan Bui et al. found a linear relationship between log Kow and adsorption amount at a wide range of pH values for 12 drugs, indicating that the adsorption of drugs on TMS-SBA-15 is mainly driven by hydrophobic interactions [46]. Taku Matsushita [47] et al. evaluated the adsorption capacity of nine activated carbons for geosmin and 2 methylisoborneol (MIB) and showed that hydrophobic materials enhance the adsorption. Based on the above discussion, hydrophobic interactions are dominant in this adsorption mechanism. Based on the discussion above, the proposed mechanism is shown in Figure 9b.

**Figure 9.** (**a**) Effect of Kow on diverse micropollutants. (**b**) Proposed reaction mechanism. **Figure 9.** (**a**) Effect of Kow on diverse micropollutants. (**b**) Proposed reaction mechanism.

### **4. Conclusions 4. Conclusions**

In summary, a novel MOF-derived carbon has been designed and synthesized for the first time. A mixed ligand approach has been applied to successfully synthesize bio-27 via the solvothermal method. By pyrolyzation at different temperatures, the optimal condition at 900 °C was selected to generate a novel heterogeneous porous carbon with the highest surface area (SBET = 980.5 m2/g) and a large pore volume (0.496 cm3/g). The maximum saturated adsorption capacity for sulfamethylthiazole (SMX) over MOF-derived carbon can reach 351.3 mg/g with a fast initial adsorption rate of 315.29 (mg/g·min). The adsorption capacity calculated by the Langmuir model (R2 = 0.99) for SMX over bio-C-900 was 4.6 and 13.3 times more than the values for AC and ZIF-8-C, respectively. Hydrophobic interaction should be one of the major mechanisms for the material's adsorption in In summary, a novel MOF-derived carbon has been designed and synthesized for the first time. A mixed ligand approach has been applied to successfully synthesize bio-27 via the solvothermal method. By pyrolyzation at different temperatures, the optimal condition at 900 ◦C was selected to generate a novel heterogeneous porous carbon with the highest surface area (SBET = 980.5 m2/g) and a large pore volume (0.496 cm3/g). The maximum saturated adsorption capacity for sulfamethylthiazole (SMX) over MOF-derived carbon can reach 351.3 mg/g with a fast initial adsorption rate of 315.29 (mg/g·min). The adsorption capacity calculated by the Langmuir model (R<sup>2</sup> = 0.99) for SMX over bio-C-900 was 4.6 and 13.3 times more than the values for AC and ZIF-8-C, respectively. Hydrophobic interaction should be one of the major mechanisms for the material's adsorption in water. This study offers a strategy to develop novel carbon materials to remove pharmaceuticals from water.

sulfamethoxazole (SMX) were 3.12, −1.55, 0.45, 1.1, 1.58, and 0.89 [42–45], respectively. Tung Xuan Bui et al. found a linear relationship between log Kow and adsorption amount at a wide range of pH values for 12 drugs, indicating that the adsorption of drugs on TMS-SBA-15 is mainly driven by hydrophobic interactions [46]. Taku Matsushita [47] et al. evaluated the adsorption capacity of nine activated carbons for geosmin and 2-methylisoborneol (MIB) and showed that hydrophobic materials enhance the adsorption. Based on the above discussion, hydrophobic interactions are dominant in this adsorption mechanism.

Based on the discussion above, the proposed mechanism is shown in Figure 9b.

ceuticals from water. **Author Contributions:** Conceptualization, X.L. and Y.M.; methodology, X.L. and Y.M.; software, Y.M.; validation, X.L. and Y.M.; formal analysis, X.L. and Y.M.; investigation, X.L. and Y.M.; resources, B.W.; data curation, X.L. and Y.M. writing—original draft preparation, Y.M.; writing—re-**Author Contributions:** Conceptualization, X.L. and Y.M.; methodology, X.L. and Y.M.; software, Y.M.; validation, X.L. and Y.M.; formal analysis, X.L. and Y.M.; investigation, X.L. and Y.M.; resources, B.W.; data curation, X.L. and Y.M. writing—original draft preparation, Y.M.; writing—review and editing, X.L; visualization, X.L; supervision, B.W.; project administration, X.L. and B.W.; funding acquisition, X.L. and B.W.; All authors have read and agreed to the published version of the manuscript.

water. This study offers a strategy to develop novel carbon materials to remove pharma-

view and editing, X.L; visualization, X.L; supervision, B.W.; project administration, X.L. and B.W.; funding acquisition, X.L. and B.W.; All authors have read and agreed to the published version of the manuscript. **Funding:** This work was financially supported by the National Natural Science Foundation of China **Funding:** This work was financially supported by the National Natural Science Foundation of China (Grants 21625102, 21971017, and 21906007), China's National Key Research and Development Program (Grant 2020YFB1506300), and the Beijing Institute of Technology Research Fund Program. We gratefully acknowledge the Analysis and Testing Center of the Beijing Institute of Technology.

gratefully acknowledge the Analysis and Testing Center of the Beijing Institute of Technology.

(Grants 21625102, 21971017, and 21906007), China's National Key Research and Development Program (Grant 2020YFB1506300), and the Beijing Institute of Technology Research Fund Program. We **Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Gayathri Anil 1,2, Jaimy Scaria <sup>2</sup> and Puthiya Veetil Nidheesh 2,\***


**Abstract:** In the present study, heterogeneous electro-Fenton (HEF) process using MnFe2O<sup>4</sup> -GO catalyst is employed for the successful removal of dye from aqueous solution. Pt coated over titanium and graphite felt were used as the electrodes. The study focuses on the efficiency of the electrodes and catalyst used for the successful removal of Rhodamine B (RhB) from aqueous solution and the application of the same in real textile wastewater. The effect of various operational parameters like pH, applied voltage, catalyst concentration, initial pollutant concentration and effect of ions were investigated. The optimized condition of the electrolytic system was found as pH 3, applied voltage of 3 V, and catalyst concentration of 20 mg L−<sup>1</sup> for the removal of 10 ppm RhB. At the optimized condition, 97.51% ± 0.0002 RhB removal was obtained after an electrolysis time of 60 min. The role of individual systems of Fe, Mn, GO and MnFe2O<sup>4</sup> without support were compared with that of catalyst composite. On examining the practical viability in real textile effluent, a significant colour reduction was observed (reduced by 61.24% ± 0.0261 in 60 min). Along with this, the biodegradability enhancement (BOD/COD ratio from 0.07 to 0.21) after treatment was also observed.

**Keywords:** heterogeneous electro-Fenton; composite catalyst; Rhodamine B dye; textile wastewater; biodegradability; advanced oxidation processes

### **1. Introduction**

Textile effluent, defined by a solid colour, soluble dyes, and organics, makes up half of all industrial wastewater in the world [1]. India stands second in the world in the production of textiles and fabrics [2]. Additives used in the textile industry for various softening and polishing purposes include colouring pigments, such as dyes and surfactants [3]. Synthetic dyes in the textile effluent maintain their colour and structural integrity under extreme weather situations and exhibit a strong resistance to microbial degradation [4,5]. They take a long time to degrade when exposed to the environment. These toxic organic pollutants generated from textile industries can affect living beings and can cause serious threats to humans and the environment. Synthetic dyes cause a wide range of health effects on living beings. They impart colour to the aquatic environment, even in a small concentration [6]. Synthetic dyes possess carcinogenic, mutagenic and genotoxic properties, prolonged exposure to which can cause human health effects, including skin/eye irritation, neurotoxicity, and endocrine disruption [7,8]. Based on the types of dyes and accompanying chemicals employed, the behavior of textile discharge changes [9]. Textile wastewater usually has a low BOD to COD ratio (less than 0.1) making it unfit for biodegradation [10]. The effluent has to be treated effectively before discarding into the environment [11].

Electrochemical advanced oxidation processes (EAOPs) are one of the most environmentally benign processes for wastewater treatment, because the reagent utilized is the electron, and the oxidizing agents, such as hydroxyl radical, are in-situ generated [12]. The process is considered to be very effective since it is ecologically friendly and can create good amount of hydroxyl radicals by controlling the applied current [13]. It is considered as an

**Citation:** Anil, G.; Scaria, J.; Nidheesh, P.V. Removal of Synthetic Dye from Aqueous Solution Using MnFe2O4-GO Catalyzed Heterogeneous Electro-Fenton Process. *Water* **2022**, *14*, 3350. https://doi.org/10.3390/w14203350

Academic Editors: Dionysios (Dion) Demetriou Dionysiou, Yujue Wang and Huijiao Wang

Received: 19 September 2022 Accepted: 19 October 2022 Published: 21 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

effective method for the treatment of dyes, as the dye compounds are readily converted into carbon dioxide, water and other inorganic ions [5].

The electro-Fenton (EF) process is the most accepted EAOP, due to the benefits such as a broad application range, excellent anti-interference ability, ease of operation, and rapid pollutant removal and mineralization efficiency [14–16]. For the past few years, EF has been the most employed EAOP method and is considered most effective due to its ability to eliminate refractory compounds [17,18]. The efficacy of the EF procedure as a tool for removing dyes from water medium has been established [19].

The EF system is supplied continuously with air or oxygen and H2O<sup>2</sup> is in-situ formed by the two-electron reduction of oxygen (Equation (1)) at the cathode. The •OH are generated in the system by the Fenton reaction (Equation (2)), by the electrogenerated H2O<sup>2</sup> at the cathode and Fe2+ catalyst added to the system [20]. The Fe2+ that is used by the system is then regenerated by the cathodic reduction of Fe3+ (Equation (3)) [21,22]. The formation of OH takes place, preferably in acidic media [23,24].

$$\text{O}\_2 + 2\text{H}^+ + 2\text{e}^- \rightarrow \text{H}\_2\text{O}\_2\tag{1}$$

$$\mathrm{Fe^{2+}} + \mathrm{H\_2O\_2} \rightarrow \mathrm{Fe^{3+}} + \mathrm{^{\bullet}OH} + \mathrm{OH^-} \tag{2}$$

$$\text{Fe}^{3+} + \text{e}^- \rightarrow \text{Fe}^{2+} \tag{3}$$

The EF process can be classified into two types, based on the catalyst used. The process can be said to be a homogenous electro-Fenton process, if the externally added catalyst is soluble in the system. If the catalyst is not soluble or is barely soluble in the system, the process is known as heterogeneous electro-Fenton (HEF). In a homogeneous system, the catalytic process takes place in the bulk liquid phase, whereas the catalytic process occurs on the surface of the catalyst in a heterogeneous system [25,26]. The mechanism of both the process is more or less similar. In HEF, the solid iron or its minerals helps in the generation of •OH [27].

The main advantage of HEF process over conventional EF process is that, the heterogeneous catalyst can be recovered and reused after the Fenton process. This lowers the cost of operation and reduces catalyst wastage. The HEF process could work in an extended range of pH. This way adjusting the pH of the solution in the conventional Fenton process can be avoided. All the reactions associated with the process occur at the catalyst surface. The ferric or ferrous ion which is present in the catalyst do not form hydroxyl complexes as they are mostly in the stable form. The production of iron sludge is also reduced by the HEF process [28].

HEF employing iron-based clays, zero-valent iron, iron oxide minerals and ironcontaining materials, has generated a lot of interest. The material must have robust catalytic activity that is unaffected by experimental conditions, particularly pH, as well as high stability to ensure reusability across several runs [17]. The most extensively used catalysts are hematite, goethite, and magnetite because of their specific properties [29]. Iron oxides are considered as environmentally acceptable Fenton catalyst, as they are abundant in nature and are easy to synthesize, making them low cost [23]. Recently, various transition metals such as cobalt, copper, nickel, vanadium and manganese have been utilized as solid catalysts with iron/iron oxides [17]. Adding transition metals such as Cu, Mn and Ni in Fe3O<sup>4</sup> is reported to be effective for the increased catalytic activity [30]. Since graphene has been discovered to considerably enhance catalyst activity, researchers have hybridized metal oxides with graphene for its immense supporting power [31]. In the present study, an approach has been made to combine the iron and transition metals with a support of graphene oxide (GO) to develop a heterogeneous catalyst and to examine its efficiency for the removal of RhB dye. For the remediation of various pollutants, such as pharmaceuticals, etc., the combination of these has shown to be successful [32,33]. When manganese was present in a magnetite structure, the rate of H2O<sup>2</sup> decomposition for the breakdown of organic molecules increased noticeably [34,35].

A number of studies have been conducted on the removal of RhB from aqueous solution using HEF. Jinisha et al. [36] used iron-doped SBA-15 mesoporous silica as a heterogeneous catalyst for the effective removal of RhB from aqueous solution. The authors used graphite felt as both anode and cathode. Tian et al. [37] used sponge iron as the catalyst along with gas diffusion electrodes (GDE) for RhB removal. Zheng et al. [38] studied the removal of RhB from aqueous solution using NiFe2O4/Fe2O<sup>3</sup> (a charcoal shaped catalyst) as the heterogeneous catalyst and graphite as the electrodes. The importance of Mn based ferrites for HEF is found nowhere in the literature.

This study aims to develop MnFe2O4-GO as the heterogeneous EF catalyst for the effective removal of RhB dye from aqueous solution. Pt coated over titanium was used as the anode and graphite felt was used as the cathode for the in-situ generation of H2O2. The effects of various operational parameters including pH, applied voltage, catalyst concentration, pollutant concentration, and the effect of ions were studied and the process was optimized. The optimized conditions were applied to the real textile wastewater and the rate of removal was investigated.

### **2. Materials and Methods**

### *2.1. Chemicals and Reagents*

Analytical grade chemicals including RhB (C28H31ClN2O3) supplied from Loba Chemie (Mumbai, India) were used for the experiments. Ferric chloride anhydrous (FeCl3, ≥98%) purchased from Merck (Mumbai, India), manganese chloride (MnCl2·4H2O, ≥ 95%) purchased from Sisco Research Laboratory (Mumbai, India), sodium acetate (CH3·COONa·3H2O, 99.0–101.0%) purchased from Qualigens were used for the synthesis of MnFe2O<sup>4</sup> catalyst. Graphite powder (99.5%) for the synthesis of graphene oxide (GO), was purchased from S D Fine Chemicals (Chennai, India).

Other chemicals, including potassium dichromate (K2Cr2O7, ≥99.0%), potassium dihydrogen phosphate (KH2PO4, ≥99.5%), di-potassium hydrogen phosphate (K2HPO4, ≥99.0%), ammonium chloride (NH4Cl, ≥99.0%), sodium thio sulphate (Na2S2O3, ≥99.5%), t-butanol((CH3)3COH, ≥99%), sodium hydroxide pellets (NaOH, ≥ 97%), sodium azide (NaN3, ≥99%), Benzoquinone (≥98%), Phenanthroline (C12H8N2·H2O, ≥99.5%), copper sulphate (HgSO4, ≥99.5%), silver nitrate (AgNO3, ≥99.5%), silver sulphate (AgSO4, ≥98%), magnesium sulphate (MgSO4, ≥98%), sodium sulphate (Na2SO4, ≥99.0%), potassium permanganate (KMnO4, ≥99.0%), and manganese sulphate (MnSO4·H2O, ≥98%), were procured from Merck. Sodium chloride (NaCl, ≥99.5%), sodium bicarbonate (Na2CO3, ≥99.9%), di sodium hydrogen phosphate (Na2HPO4·7H2O, ≥99.9%), magnesium sulphate (MgSO4·7H2O, ≥99.9%), calcium chloride (CaCl2, ≥98%), ferric chloride (FeCl3·6H2O, ≥98%), ferrous ammonium sulphate ((NH4)2(SO4)2·6H2O, ≥99%), were procured from Fisher Scientific. Ammonium acetate (CH3COONH4, ≥98%), hydroxyl amine (NH2OH, ≥96%), and potassium iodide (KI, ≥99%) were purchased from Qualigens.

Solvents including ethylene glycol (HOCH2- CH2OH, ≥99.0%), glycerol and ethyl alcohol (C2H5OH, ≥95%) from Merck, and poly-ethylene glycol (99.9%) from Qualigens, were used for different analytical experiments. Sulphuric acid (H2SO4, 95.0–98%) and ferroin indicator from Merck, and phenolphthalein supplied from Rankem, were used for the experiments.

### *2.2. Electrodes*

Pt coated over titanium plates procured from Titanium Tantalum Products Limited, Chennai, India, was used as an anode and graphite felt was used as a cathode.

### *2.3. Preparation of Heterogeneous Catalyst*

### 2.3.1. Synthesis of Graphene Oxide (GO)

For the synthesis of graphene oxide (GO) from natural graphite powder, a modified version of Hummers' method combined with laboratory optimizations, as mentioned in our earlier work [39], was used. In a water bath, 1 g of the graphite powder was mixed with 68 mL of conc. H2SO<sup>4</sup> and mixed thoroughly for 1 h. While keeping the ice bath conditions (to keep the temperature below 20 degrees), 4 g KMnO<sup>4</sup> was added gradually (1 g at a time). The resultant mixture was agitated at room temperature for 6 h, until the solution changed colour from green to dark brown–black. The oxidation was then stopped by drop-wise addition of 140 mL of double-distilled water (DDW), which followed by 5 mL H2O<sup>2</sup> addition. Finally, the metal ions were removed by washing the mixture three times with 10% HCl.

Washing with DDW was repeated (by centrifugation at 7500 rpm) until the pH of the supernatant liquid approached neutral condition. The addition of silver nitrate confirmed the removal of chloride from the supernatant. After drying at 40 ◦C in a hot air oven, the resultant solid was pulverized [39].

### 2.3.2. Synthesis of Manganese Ferrite Supported on GO

To synthesize MnFe2O4–GO composite, the facile solvothermal reduction was used. To a 250 mL volumetric flask containing 40 mL ethylene glycol (EG), 0.0387 g of GO was added. The mixture was kept in a bath sonicator for 1 h for the effective dispersion of GO. 0.811 g ferric chloride and 0.3010 g manganese chloride was added. The mixture was kept for magnetic stirring between 650–700 rpm until the FeCl<sup>3</sup> and MnCl<sup>2</sup> added were completely dissolved. A total of 3.6 g sodium acetate (CH3.COONa.3H2O) followed by 1 mL polyethylene glycol (PEG) were added to the above mixture and agitated at 550 rpm for 1 h. The agitated mixture was then transferred to a Teflon-lined stainless steel autoclave and kept for 6 h at 200 ◦C. After cooling, the precipitate was then separated using a magnet. The dark brown precipitate was washed thrice with ethanol and DDW. The washed precipitate was then vacuum dried for 6 h. After cooling, it was powdered and used for experiments [39]. EG used in the preparation acted as both solvent and reducing agent [39].

To find out the role of Fe, Mn, GO and spinel structure in the catalyst, the above method was followed. Catalysts were prepared by adding Fe only (MnCl<sup>2</sup> was not added), Mn only (FeCl<sup>3</sup> was not added), GO only (both FeCl<sup>3</sup> and MnCl<sup>2</sup> was not added) and MnFe2O<sup>4</sup> without support (GO was not added).

### *2.4. Catalyst Characterization*

The structural characterization of the laboratory-synthesized catalyst was done using X-ray powder diffraction (XRD, Bruker D8 Advance). The functional groups present on the surface of the catalyst were detected by Fourier transform infrared (FTIR) spectroscopy (FTIR, Bruker, Germany). The size of the prepared catalyst was determined by transmission electron microscope (TEM, Tecnai G2 20 S-TWIN).

### *2.5. Experimental Procedure*

Electrolytic experiments were carried out in a reactor of 1 L capacity. A 1000 mL solution containing of 10 ppm RhB was prepared at the beginning of each experiment. Initial pH was maintained at 3 for all the experiments, except those in which the effect of different pH was examined. pH was adjusted using 0.1 N H2SO<sup>4</sup> and 0.1 N NaOH. pH was monitored using a Laqua pH meter (Horiba scientific). Two anodes and two cathodes of size 5 × 8 cm with a 1 cm inner electrode spacing were used for the experiments. Using a commercially available fish aerator, a continuous supply of air was fed into the system until the end of the experiments. After an aeration time of 5 min, catalyst was added. The electrodes were connected to a DC power supply (make: Crown). The experimental setup is shown in Figure 1. Samples were collected at 15 min intervals and stored in the refrigerator in brown vials to avoid further degradation or reaction. The residual RhB present in each sample was measured using a UV/Vis spectrophotometer (UV-1900I, Shimadzu, Japan) at 555 nm (which is the peak wavelength of RhB). All the experiments were conducted at room temperature. The effect of different operational parameters such as pH, voltage, catalyst concentration, and pollutant concentration were also studied.

**Figure 1.** Schematic diagram of the experimental setup.

### *2.6. Real Textile Wastewater: Collection and Analysis*

Real textile wastewater was collected from an industry situated in Mumbai, Maharashtra, India. A 10 L sample was collected, sealed tightly and stored in the refrigerator. Parameters including total dissolved solids (TDS), total solids (TS), chloride, sulphate, biological oxygen demand (BOD), chemical oxygen demand (COD) were analyzed as per the standard methods [40]. pH and electrical conductivity of the wastewater at room temperature were recorded using a Laqua pH meter (Horiba scientific). The absorbance of the wastewater was recorded using a UV/Vis spectrophotometer (UV 1900 I, Shimadzu, Japan). Using a total organic carbon (TOC) analyser (Shimadzu, Japan, Model: TOC Veph), the TOC content of the real textile wastewater was determined.

### *2.7. Energy Consumption*

The specific energy consumption with different applied voltages was found using the equation mentioned in [41].

$$\mathbf{Q} = \frac{\text{VIt}}{\text{m}}\tag{4}$$

where Q = specific energy consumption (KWh−<sup>1</sup> ), V = applied voltage (V), I = current (A), t = electrolysis time (min), m = change in concentration.

### *2.8. Leaching of Catalyst*

The quantity of Fe leached was quantified using a 1.10-phenanthroline method by spectrophotometric quantification. The quantity of Mn leached from the catalyst was quantified using ICP-OES (iCAP-7400, Thermo Fisher Scientific, Waltham, MA, USA).

### **3. Results and Discussions**

### *3.1. Characterization of the Catalyst*

XRD, TEM and FTIR were used to characterize the MnFe2O4–GO catalyst. XRD patterns provide the crystal structure and phase information of the catalyst [42]. Figure 2a illustrates the XRD pattern of MnFe2O4–GO. The obtained diffraction peaks are similar to that of JCPDS 38-0430. The peaks observed at 2 θ values of 16.92, 29.59, 35.75, 36.32, 42.92, 52.59, 56.01, 63.96, 76.64, which can be indexed as (111), (202), (311), (222), (004), (422), (333), (440), and (533) planes, respectively. Similar patterns of MnFe2O<sup>4</sup> were observed earlier [42,43]. The peak observed at the 2 θ value of 7.69 corresponds to the (001) characteristic plane of GO [44]. The presence of a GO peak indicates the ethylene glycol is insufficient to convert GO to rGO when more than one metal oxide is present during the solvothermal process, whereas this conversion is viable when only one metal oxide is present, as reported in our previous study [39]. This may be because the amount of solvent and time taken for the synthesis may not be enough for the reduction of GO [45].

**Figure 2.** XRD pattern of MnFe2O4–GO (**a**) TEM images of MnFe2O4–GO (**b**), (**c**), FTIR spectrum of MnFe2O4–GO (**d**).

The shape, size and distribution of particles can be found using a TEM [46]. Figure 2b,c shows the TEM images of the catalyst composite synthesized. From Figure 2b,c, the irregular rough surface MnFe2O<sup>4</sup> can be observed, indicating the small MnFe2O<sup>4</sup> particles are clustered together to form a spherical aggregate, with a diameter of 132 ± 92 nm. The presence of GO sheets is evident in Figure 2c. On further examination of Figure 2b,c, the different stages of spherical structure development indicate that aggregation and particle growth are not completed and are at different phases.

The FTIR spectrum of MnFe2O<sup>4</sup> -GO is shown in Figure 2d. As illustrated in the figure, bands of functional groups such as –OH, –C=O, C– H, C–O–C/C–O–H, –C=C, and –C–O were obtained at around 3403.9 cm−<sup>1</sup> , 2917.8 cm−<sup>1</sup> , 1581.6 cm−<sup>1</sup> , 1125.5/1437.3 cm−<sup>1</sup> , and 1081.3 cm−<sup>1</sup> , respectively. The bands observed at 580 cm−<sup>1</sup> and 443 cm−<sup>1</sup> can be attributed to the bonds corresponding to Fe-O and Mn-O, respectively [47,48]. The presence of oxygen functionalities implies the presence of GO [45].

### *3.2. Rhodamine B Degradation in MnFe2O4–GO-Based Electro-Fenton System*

### 3.2.1. Comparison between Different Processes

During HEF treatment of dye, a combination of various processes such as adsorption, anodic oxidation (AO), and AO+H2O<sup>2</sup> may be contributed. In order to determine the contribution of these processes, degradation of RhB in different individual processes was carried out. Adsorption of dye on catalyst was analyzed in the absence of current supply

and aeration. The effect of AO was investigated by applying electricity to the electrochemical system without providing catalyst and aeration. AO + H2O<sup>2</sup> was carried out in the electrochemical system with electrodes, and external aeration without the addition of the catalyst. The other experimental conditions were pH 3, applied voltage of 3 V and initial pollutant concentration of 10 ppm.

Figure 3a depicts the degradation rate achieved with different processes. The different degradation percentage obtained by the individual process is as follows: 12.36% ± 0.0028 by adsorption, 64.91% ± 0.0439 by AO, 90.42% ± 0.0395 by AO + H2O<sup>2</sup> and 97.52% ± 0.0002 by HEF process. On examining the kinetics, all the above individual process follows firstorder kinetics. The first-order rate constant of each process are as follows: adsorption (0.0024 min−<sup>1</sup> ), anodic oxidation (0.019 min−<sup>1</sup> ), anodic oxidation combined with aeration (0.04 min−<sup>1</sup> ) and HEF (0.079 min−<sup>1</sup> ). Even though AO + H2O<sup>2</sup> and HEF showed comparable dye removal within 60 min, the reaction rate of HEF is found to be double than that of AO combined with aeration. Thus, the addition of catalyst can significantly enhance the pollutant removal in EF.

**Figure 3.** (**a**) Degradation of RhB with different process with rate of reaction in inset (**b**) Effect of catalyst dosage (**c**) Effect of pH (**d**) Effect of Initial dye concentration.

Similar results were obtained in HEF treatment of tannery wastewater [49]. The author used Fe3O4/Mn3O4/ZnO–rGO hybrid quaternary nano-catalyst. Graphite was used as the cathode and Ti/IrO2/RuO<sup>2</sup> as the anode. The authors compared different processes of HEF, AO + H2O2, AO and adsorption. Higher degradation was obtained by HEF with 97.08% removal. A study by Jinisha et al. [36] compared electro-sorption, adsorption and HEF process. Higher degradation was shown by HEF (97.7%) process. A study conducted by Bedolla-Guzman and co-workers [50] compared the efficiency of AO + H2O<sup>2</sup> and EF process in reactive yellow 160 dye with boron-doped diamond (BDD) as the anode and carbon polytetrafluorethylene (PTFE) as the cathode with Fe2+ as catalyst. The authors achieved higher efficiency with the EF process (91%). The study on the degradation of sunset yellow FCF azo dye [51] revealed that on comparison of AO−H2O<sup>2</sup> and EF, the authors achieved higher degradation with EF. Only 88% dye was removed by AO−H2O<sup>2</sup> in 360 min, whereas in EF around 50% dye was removed within 5 min, which fully decolorized in 45 min.

### 3.2.2. Effect of Catalyst Dosage

The catalyst dosage has a crucial role in the HEF process [52]. In order to optimize the catalyst dosage, HEF treatment was carried out by varying the catalyst dosage (5 mg L−<sup>1</sup> , 10 mg L−<sup>1</sup> , 20 mg L−<sup>1</sup> and 50 mg L−<sup>1</sup> ). As illustrated in Figure 3b, on increasing the catalyst dosage from 5 mg L−<sup>1</sup> to 10 mg L−<sup>1</sup> , RhB removal slightly improved from 93.09% ± 0.0081 to 94.02% <sup>±</sup> 0.0123. As, Fe2+ generation is less at lower catalyst concentrations, the production of hydroxyl radical is also less, which explains the lesser efficiency in degradation at 5 mg L−<sup>1</sup> catalyst concentration [23,41,53]. When increasing catalyst dosage to 20 mg L−<sup>1</sup> , a sharp increase in pollutant removal was evident, as RhB removal reached 97.51% ± 0.0002 within 60 min. Whereas, further increase in catalyst dosage lowered the RhB removal (reduced to 95.81% ± 0.0066).When excessive catalyst concentration is present in the system, the generated hydroxyl radicals could be scavenged and could lead to hydroperoxyl radical formation which has less oxidation potential than •OH (Equations (5)–(7)) [23,53,54]. Thus, the optimized catalyst concentration is taken as 20 mg L−<sup>1</sup> for all further experiments.

$$\text{Fe}^{2+} + \text{HO}^{\bullet} \rightarrow \text{Fe}^{3+} + \text{OH}^{-} \tag{5}$$

$$\text{Fe}^{2+} + \text{H}\_2\text{O}\_2 \rightarrow \text{Fe}-\text{OOH}^{2+} + \text{H}^+ \tag{6}$$

$$\text{Fe} - \text{OOH}^{2+} \rightarrow \text{HO}\_2^{\bullet} + \text{Fe}^{2+} \tag{7}$$

### 3.2.3. Effect of pH

The solution pH is a decisive parameter of the Fenton process [55]. An acidic pH, preferably pH 3, results in more radical generation, thus more activity. To determine the optimum pH for the HEF process, experiments were carried out under different pH of 2, 3, 7, 9, and 11. As Figure 3c indicates, higher RhB degradation is evident at acidic conditions (2 and 3). At pH 2, RhB removal of 97.72% ± 0.0021 and at pH 3, removal of 97.52% ± 0.0002, were observed. Whereas at neutral pH, only 18.08% ± 0.0765 of RhB was removed. At pH 9, 18.37% ± 0.0082 and at pH 11, 25.06% ± 0.0077 removal, were obtained.

Many studies reported the higher degradation of compounds at pH 3. Zheng et al. [38] observed a high degradation in RhB at pH of 3 on using NiFe2O4/Fe2O<sup>3</sup> as the heterogeneous catalyst with graphite felt as the electrodes. Nidheesh et al. [54] studied the degradation of RhB using magnetite as the catalyst and graphite as the electrode. The authors also obtained higher efficiency at the pH of 3. Fayazi and Ghanei-Motlagh, [56] obtained higher degradation efficiency of methylene blue at pH 3 when sepiolite/pyrite composite was used as the catalyst, and graphite and platinum sheet was used as the electrodes. On the degradation of Ponceau SS dye using heterogeneous electro-Fenton process, dos Santos et al. [57] observed higher degradation at pH 3, when vermiculite was used as the catalyst.

For pH above 5, precipitation of ferric oxyhydroxide (FeOOH2+) and ferric hydroxide (Fe(OH)3) by ferric ions takes place terminating the Fenton reaction [58,59]. Formation of ferric hydroxides of iron species occurs at higher at higher pH values leading to the lowering of Fe2+/Fe3+ ratio [23,60,61]. For the formation of H2O2, H<sup>+</sup> is needed (Equation (1)). H<sup>+</sup> is available only in acidic conditions. So, pH 3 is taken as optimum pH for further experiments.

### 3.2.4. Effect of Voltage

For achieving better process efficiency and for the formation of Fenton reagent, applied voltage or current density is a crucial parameter [12]. The effect of different applied voltages ranging from 1 V to 5 V were scrutinized to find out the optimum voltage. As the

applied voltage was increased, the degradation also increased and after a certain range the degradation rate decreased. When the applied voltage was 1 V, the removal percentage was 22.15% ± 0.0414; on increasing the applied voltage to 2 V, degradation enhanced to 37.32% ± 0.0856. An upsurge degradation was achieved when the applied voltage was 3 V i.e., 97.52% ± 0.0002. On applying a voltage of 4 V, a similar removal percentage of 3 V was obtained (97.44% ± 0.0009) and was further decreased on applying 5 V (68.21% ± 0.0672). The RhB removal attained with different applied voltages is illustrated in Figure 4a.

**Figure 4.** (**a**) Effect of voltage on RhB degradation. (**b**) Energy consumption with varying voltage.

The enhancement in degradation with increasing voltage is due to the increase in the production of •OH [8,54]. Reduction in oxidation at higher voltages after a certain level is due to the hydrogen gas evolution, decomposition of H2O<sup>2</sup> (Equation (8)) and formation of H2O.

$$\rm 2H\_2O\_2 \to 4H^+ + O\_2 + 4e^- \tag{8}$$

The energy consumption for the RhB removal via HEF is calculated for varying applied voltages (Figure 4b). On comparing different voltages, 3 V showed significant reduction in RhB concentration by utilizing the lower energy. In both higher and lower voltage conditions, energy efficiency was less. Higher energy consumption will increase the cost of the process and thus a voltage of 3 V was selected for further experiments [54].

### 3.2.5. Effect of Initial Dye Concentration and Electrolysis Time

The effect of RhB concentration in HEF activity was evaluated by changing the initial pollutant concentration of 5 ppm, 10 ppm, 20 ppm and 50 ppm. The different removal rates with different concentrations of initial dye are explained in Figure 3d. For 5 ppm dye, 88.57% ± 0.0699 removal was observed which enhanced to 97.51% ± 0.0002 when the dye concentration was increased to 10 ppm. Further increase in concentration affected the efficiency drastically. When the initial dye concentration was 20 ppm, 50.36% ± 0.0472 removal was obtained, which lowered to 22.82% ± 0.0335 when the initial dye concentration was raised to 50 ppm.

Rate of removal of dye decreases with an increase in dye concentration [62]. The decrease in removal of dye with an increase in concentration can be ascribed to the lesser number of available •OH for the oxidation [25]. The lower pollutant removal in case of initial dye concentration below 10 ppm is due to the insufficient radical generation. This results in lower dye removal as collision between the particles is limited [54].

RhB degradation readily increases with electrolysis time. From Figure 3d, it is clear that a good amount of RhB particles have been degraded within the first 15 min of the electrolysis. As the time increased, around 97.51% ± 0.0002 was degraded from the system at the end of 60 min. Nidheesh et al. [54,63] reported the increasing RhB degradation with an increase in time. The decolorization rate was higher at the initial time intervals of electrolysis because of the higher formation of •OH. The degradation rate decreased as the time increased, because of the decrease in collision of RhB molecules with •OH [54,63].

### 3.2.6. Effect of Ions

The effect of different anions such as chlorides, sulphates and carbonates on RhB solution at optimized conditions was studied. Figure 5 illustrates the effect of various ions on the degradation of RhB. 50 mg L−<sup>1</sup> and 100 mg L−<sup>1</sup> of each NaCl, Na2SO<sup>4</sup> and Na2CO<sup>3</sup> were added to the RhB solution.

**Figure 5.** (**a**) Effect of NaCl on the degradation of RhB. (**b**) Effect of Na2SO<sup>4</sup> on the degradation of RhB. (**c**) Effect of Na2CO<sup>3</sup> on the degradation of RhB Experimental conditions: pH-3, Voltage–3, Catalyst dosage- 20 mg L−<sup>1</sup> , Initial dye concentration-10 ppm.

When NaCl was added into the RhB solution, the removal percentage was 68.21% ± 0.0189 for 50 mg L−<sup>1</sup> of NaCl and 93.1% <sup>±</sup> 0.0184 for 100 mg L−<sup>1</sup> of NaCl. When 50 mg L−<sup>1</sup> of Na2SO<sup>4</sup> was added to the RhB solution the degradation percentage obtained was 59.35% <sup>±</sup> 0.0265, and on addition of 100 mg L−<sup>1</sup> the degradation obtained was 86.1% <sup>±</sup> 0.0782. When 50 mg L−<sup>1</sup> of Na2CO<sup>3</sup> was added, the degradation percentage obtained was 38.63% <sup>±</sup> 0.0437, and 22.63% <sup>±</sup> 0.0039 when 100 mg L−<sup>1</sup> of Na2CO<sup>3</sup> was added. Among the different anions added, the presence of carbonates has shown a strong influence in HEF activity. Whereas other ions such as chloride and sulphate inhibit HEF activity at lower concentration, and on higher dosage, can result in supporting the HEF activity via additional radical generation.

The presence of Cl− in wastewaters have a suppressing effect on AOPs as they are said to have scavenging effect on •OH [64]. However, in this case, addition of chlorides as well as sulphates does not diminish the degradation but on the contrary increased the degradation with the increase of concentration. The upgrade in degradation with the increase of sulphate concentration was reported by Zhou et al. [65]. The authors explained the increase in degradation as the enhancement of current density with sulphate ions. The comparative higher degradation in chloride media may be due to the formation of active chlorine species [12]. The decrease in degradation in the presence of anions is due to the scavenging effect of hydroxyl radicals by sulphate (Equation (9)) and chloride (Equations (10) and (11) forming hydroxyl ions [54]. In the presence of carbonates, the degradation of RhB is suppressed by the formation of FeCO3. Jinisha et al. [36] concluded that the effect of anions like sulphates and carbonates does not influence the production of H2O2.

$$\mathrm{HO}^{\bullet} + \mathrm{SO}\_{4}^{2-} \rightarrow \mathrm{OH}^{-} + \mathrm{SO}\_{4}^{-} \tag{9}$$

$$\mathrm{HO}^{\bullet} + \mathrm{Cl}^{-} \leftrightarrow \mathrm{HOCl}^{-} \tag{10}$$

$$\mathrm{HOCl}^{-\bullet} + \mathrm{Cl}^{-} \rightarrow \mathrm{Cl}\_{2}^{-} + \mathrm{OH}^{-} \tag{11}$$

3.2.7. Radical Scavenging Tests

To evaluate the contribution of various radicals such as hydroxyl radical on the surface ( •OHsurf) and on bulk (•OHbulk), super oxides (O<sup>2</sup> −•) and singlet oxygen (1O2) scavengers such as t-butanol, KI, benzoquinone and sodium azide (NaN3) were utilized. The scavenger t-butanol is known to have a good ability to scavenge •OH [12]. Benzoquinone was used as superoxide radical scavenger [61]. The surface •OH present on the catalyst surface are scavenged by KI [66]. The potential generation of singlet oxygen can be evaluated by sodium azide [67].

As shown in Figure 6a, the surface •OH activity is insignificant, whereas the bulk •OH radical strongly contributed to HEF activity. From Figure 6a, the non-radical pathway via singlet oxygen is the predominant pathway of HEF activity, and followed the trend, <sup>1</sup>O<sup>2</sup> > •OHbulk > O<sup>2</sup> −• .

**Figure 6.** (**a**) Radical scavenging tests with rate of reactions; (**b**) sodium azide quenching experiments at different conditions.

To confirm the contribution of singlet oxygen in the system, quenching using NaN<sup>3</sup> was conducted in the presence and absence of the MnFe2O4–GO catalyst. As illustrated in Figure 6b, a strong presence of a non-radical pathway is evident in the AO + H2O<sup>2</sup> system, where the NaN<sup>3</sup> lowered the RhB degradation rate to 0.0125 min−<sup>1</sup> , comparable to the HEF with NaN<sup>3</sup> (0.0105 min−<sup>1</sup> ). This observation indicates the formation of <sup>1</sup>O<sup>2</sup> is not solely from the catalyst.

Similarly, Lu et al. [68] reported the formation of singlet oxygen in the AO+H2O<sup>2</sup> system in the presence of chloride (Cl−). Herein, the chloride presence is evident, and the reasons might be (1) from the tap water used for making RhB-simulated water (37.7 mg L−<sup>1</sup> of chloride), (2) Cl− release from the RhB structure due to degradation. So, the singlet oxygen formation can be attributed to the singlet oxygen formation in chloride medium.

### 3.2.8. Role of Fe, Mn, GO and MnFe2O<sup>4</sup>

The individual contribution of Fe, Mn, GO and MnFe2O<sup>4</sup> spinel structure in providing HEF activity were investigated using catalysts prepared under respective conditions by excluding other precursors involved. Because of the availability of numerous oxygen functional groups, large surface area, electrical conductivity and mechanical stability, graphene and GO are very promising support materials [69,70], and thereby enhance catalytic activity [71]. The enhancement in catalytic activity of the L–GO–ZnO composite on the photocatalytic activity of RhB was reported by Yaqoob et al. [10]. There are a few studies in the literature which report the catalytic activity of GO [72]. Thus, the role of GO alone is also examined to investigate the possibility of GO as a catalyst for HEF activity. Figure 7a depicts the different degradation rates of RhB obtained under each condition. From Figure 7a, it is clear that the catalyst works efficiently when it is in the form of spinel structure with GO support with a rate of RhB removal of 0.079 min−<sup>1</sup> . In the case of GO, Mn,

Fe and MnFe2O<sup>4</sup> without GO, the rates of RhB removal were 0.0449 min−<sup>1</sup> , 0.0601 min−<sup>1</sup> , 0.0610 min−<sup>1</sup> , 0.0640 min−<sup>1</sup> , respectively (inset of Figure 7a). The GO solely does not directly contribute to the HEF activity, as the RhB degradation in that case is similar to that of AO + H2O<sup>2</sup> (0.04 min−<sup>1</sup> ). Similar results were obtained by Yao et al. [71]. In addition, this study observed an additional benefit of GO in the catalyst composite development, as the GO support contributed to lowering the solvothermal synthesis duration. In the absence of GO, MnFe2O<sup>4</sup> development required a minimum of 9 h, whereas in the presence of GO, proper catalyst development was observed within 6 h of solvothermal treatment.

**Figure 7.** (**a**) Role of Fe, Mn and GO in catalyst with rate of reaction. (**b**) Leaching of Fe and Mn.

### 3.2.9. Leaching Studies

The stability of the as-synthesized catalyst was evaluated by analyzing the Fe and Mn leaching using ICP-OES and 1,10- phenanthroline method [39]. The amount of leached iron and Mn from the catalyst is illustrated in the Figure 7b. The amount of Fe2+ and total iron in the initial 30 min showed no significant leaching. After 30 min, there was a gradual increase in concentration, indicating the leaching of Fe2+ and total iron into the solution. The amount of Mn leached out showed fluctuating results which indicated that no gradual Mn release occurred.

### 3.2.10. Mechanism of Dye Removal

The mechanism involved in the HEF oxidation of RhB is given in Figure 8. The Fe2+ from the catalyst decomposes in-situ generated H2O<sup>2</sup> to form hydroxyl radicals (Equation (5)) which degrade the RhB molecules into CO2, H2O and other byproducts. Mn metal present in MnFe2O<sup>4</sup> strongly accelerated the H2O<sup>2</sup> decomposition, as in Equations (12) and (13) [34,35]. The Fe2+ formation from Fe3+ occurred by electron loss (which is explained by Equations (1)–(3)). All the possible reductions of Fe3+ to Fe2+ and degradation occur at the cathode, as RhB is a cationic dye [54].

$$\text{Mn}^{2+} + \text{H}\_2\text{O}\_2 \rightarrow \text{Mn}^{3+} + \text{HO}^\bullet + \text{OH}^- \tag{12}$$

$$\text{Mn}^{3+} + \text{H}\_2\text{O}\_2 \rightarrow \text{Mn}^{2+} + \text{HO}\_2^{\bullet} + \text{H}^+ \tag{13}$$

(RhB) degradation by reactive radicals is possible by ring-opening reaction and N-deethylation pathways [63]. As explained in Section 3.2.1, the slow decolorization of the RhB in AO + H2O<sup>2</sup> system in the absence of catalyst may be facilitated by singlet oxygen. Singlet oxygen can be possibly formed by deprotonization, followed by electron rearrangement of oxygen molecules and disproportionation of H2O<sup>2</sup> [73–75].

The results from this study compared with the efficiency of various catalysts in the removal of various dyes using HEF process, and given in Table 1. From Table 1, it is clear that the use of MnFe2O4–GO composite catalyst in the present study has significantly reduced the total electrolysis time, while keeping higher dye removal efficiency. The efficiency of the process is higher when compared with the conventional EF process [13].

**Figure 8.** Mechanism of dye removal.



### *3.3. HEF Activity in Real Textile Wastewater*

The wastewater collected was slightly acidic in nature (pH of 5.34) and had a conductivity of 13.2 mS cm−<sup>1</sup> . The optimized experimental conditions (pH-3, applied voltage of 3 V, catalyst concentration of 20 mg L−<sup>1</sup> ) were applied to the textile wastewater for its effective treatment. The colour change was examined at a wavelength of 555 nm. As illustrated in Figure 9, colour reduction of 61.4% ± 0.061 was obtained after HEF treatment. The decolorization is caused by the breakdown of the dyes belonging to the chromogen group into its by-products [77]. Similarly, Nidheesh et al. [63] found 97.5% colour removal in textile wastewater by the HEF process. Colour removal in real textile water was investigated in batch and continuous modes by Nidheesh and Gandhimathi [77]. In the batch study, there was 83% colour removal and in the continuous study, 68% colour removal was obtained.

**Figure 9.** Colour removal and COD (mg L−<sup>1</sup> ) removal of real textile wastewater.

The COD were determined at different intervals to understand the degree of pollutant mineralization. The COD removal within 1 h is shown in Figure 9. Fluctuating COD percentage removal was observed. The abnormal result may be due to the mineralization of the compounds and intermediates present in the wastewater [78]. The degradation of complex organic pollutants, (non-detectable in COD analysis) to simpler organic pollutants detectable in COD analysis may result in rising COD values during the HEF process [77]. In general, the dyes present in textile wastewater are aromatic compounds with multiple aromatic rings and are not degradable by dichromate ions (used for COD analysis). However, these compounds are degradable by hydroxyl radicals, as they have higher oxidation potential than dichromate ions. However, the intermediate products and byproducts formed during the HEF process may not be as stable as that of dye and can be degraded by dichromate ions (thus, COD values will be provided by these compounds).

The TOC of the textile wastewater at different time intervals was determined, and the values before and after treatment are shown in Table 2. The initial TOC of 1551 mg L−<sup>1</sup> reduced to 1478 mg L−<sup>1</sup> after an electrolysis time of 60 min, with an increase in BOD from 631.58 to 1894.74 mg L−<sup>1</sup> (Table 2). The conversion of complex organic structures to simpler biodegradable organic compounds, instead of total mineralization, could end up in lower TOC removal with improvement in BOD [79,80]. Generally, carboxylic acids such as oxalic acids are formed as intermediates in the degradation of dyes [81]. These compounds are unable to be degraded with •OH due to their inefficiency [82], while microorganisms are found to be effective for the degradation of organic compounds [83]. This might have resulted in the increase in BOD.

Textile effluent characteristics before and after treatment were examined (Table 2). A slight increase in chloride and sulphate values was observed after the treatment. The increase in chloride and sulphate values may be due to the mineralization of chloride and sulphate containing complex aromatics present in the wastewater. Textile wastewaters are characterized by a high content of chloride in the range of 1600–2100 mg L−<sup>1</sup> , as reported by the earlier studies [3]. Brillas et al. [5] suggested the increasing chances of chlorinated compounds in the wastewater after electrochemical processes. Studies have reported the increase in concentration of inorganics after electrochemical treatment [84]. For the treatment of textile effluents, hybrid systems are considered, as they will resolve the various limitations associated with the treatment using single systems. The increase in chloride and sulphate after HEF treatment may affect subsequent treatment. For example, AOPs followed by biological treatments are preferred because of the enhancement in biodegradability after the treatment [85]. However, the increase in chloride and sulphate can hamper the biological treatments [84]. Sodium and magnesium have been significantly decreased after treatment. The BOD/COD ratio shot up from 0.07 to 0.21, indicating the increase in biodegradability of the wastewater after treatment. Thus, the HEF process can be effectively used as a pre-treatment process for wastewater.

**Table 2.** Physicochemical characteristics of textile wastewater.


### **4. Conclusions and Future Perspectives**

Solvothermally synthesized MnFe2O<sup>4</sup> supported on GO was used as the catalyst for the effective removal of RhB from aqueous solution by HEF process. The XRD, FTIR and TEM confirmed the formation of MnFe2O<sup>4</sup> over GO. The optimized experimental conditions were found as pH 3, applied voltage of 3 V, catalyst concentration of 20 mg L−<sup>1</sup> and initial pollutant concentration of 10 ppm. After an electrolysis time of 60 min, a removal efficiency of 97.51% could be achieved with these optimized conditions. Bulk •OH and super-oxides were found out to be the superior oxidants controlling the process. The role of GO in catalyst was significant, as the time for the catalyst formation could be reduced. Mn leaching from the catalyst was found to be insignificant. However, the leaching of iron could be seen after an electrolysis time of 30 min. The optimized experimental conditions were applied to the real textile wastewater and an efficient colour removal of 61.23% at 555 nm could be observed. A TOC removal of 4.71% was obtained. Even though the COD results were fluctuating, an increase in BOD/COD ratio from 0.07 to 0.21 was observed after treatment, which indicated that the biodegradability of the wastewater was significantly enhanced. Thus, the HEF process could be used as an effective pre-treatment method for wastewaters.

The main challenge faced during the HEF process was the need for frequent replacement of the cathode, especially for the treatment of real wastewater. The efficiency of the dye removal greatly depends on the cathode, and passivation of the cathode during treatment will negatively affect its efficiency. Another challenge faced is associated with the variation in the behavioural properties of the catalyst during multiple synthesis.

The need for pilot-scale research in textile wastewater has to be mentioned. Further studies require an increase in the efficiency of textile wastewater in terms of colour and COD, as well as TOC. The degradation of textile wastewater can also be found as a factor of time, i.e., whether the optimized conditions will increase efficiency in extending the electrolysis time. Biological studies can be conducted to confirm the biodegradability of the textile wastewater after treatment.

**Author Contributions:** Conceptualization, P.V.N.; methodology, P.V.N. and J.S.; investigation, G.A. and J.S.; writing—original draft preparation, G.A.; writing—review and editing, P.V.N. and G.A.; supervision, P.V.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article.

**Acknowledgments:** The authors are thankful to the Director, CSIR NEERI, for their support. The authors would also like to thank the Sophisticated Analytical Instrument Facility (SAIF), Indian Institute of Technology-Bombay, India; Indian Science Technology and Engineering Facilities Map (I-STEM), the Centre for Nano Science and Engineering, Indian Institute of Science, Bangalore; and the Institute Instrumentation Centre (IIC), the Indian Institute of Technology, Roorkee, India, for their assistance with various analyses.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


## *Article* **Optimized Removal of Azo Dyes from Simulated Wastewater through Advanced Plasma Technique with Novel Reactor**

**Yang Liu 1,\* , Jia-Wei Song <sup>1</sup> , Jia Bao 1,\* , Xin-Jun Shen <sup>1</sup> , Cheng-Long Li <sup>1</sup> , Xin Wang <sup>1</sup> and Li-Xin Shao <sup>2</sup>**


**\*** Correspondence: liuyang@sut.edu.cn (Y.L.); baojia@sut.edu.cn (J.B.)

**Abstract:** Increasing attention has been paid to removal of aqueous contaminations resulting from azo dyes in water by plasma technology. However, the influence factors and removal mechanism of plasma technology were still obscure, moreover, energy consumption and oxidized degradation efficiency of plasma reactor were also inferior. In the present study, a comparative analysis was performed using 100 mg/L of Methyl Orange (MO) in the simulated wastewater with a novel plasma reactor to achieve the ideal parameters involving voltage, discharge gap, and discharge needle numbers. Therefore, the optimal removal rate for MO could be up to 95.1% and the energy consumption was only 0.26 kWh/g after the plasma treatment for 60 min, when the voltage was set as 15 kV, the discharge gap was 20 mm, and the discharge needle numbers was 5. Based upon the response surface methodology (RSM), the removal rate of MO was predicted as 99.3% by massive optimization values in software, and the optimum conditions were confirmed with the plasma treatment period of 60 min, the voltage of 14.8 kV, the discharge gap of 20 mm, and the discharge needles of 5. Plasma associated with catalysts systems including plasma, plasma/Fe2+, plasma/PS, and plasma/PS/Fe2+ were further investigated, and the best removal rate for MO reached 99.2% at 60 min under the plasma/PS/Fe2+ system due to simultaneously synergistic reactions of HO• and SO<sup>4</sup> •−. Moreover, it was also revealed that –N=N– bond was attacked and broken by active species like HO• , and the oxidized by-products of benzenesulfonic acid and phenolsulfonic acid might be generated, via the analysis of the variation in the absorbances through UV-Vis spectrophotometry during the plasma treatment. As a result, the advanced plasma technique in this study presented excellent efficacy for MO removal from simulated wastewater with low energy consumption.

**Keywords:** azo dyes; Methyl Orange; novel plasma reactor; oxidized degradation; plasma technology; response surface methodology (RSM)

### **1. Introduction**

In 1862, the inartificial dyes were initially obtained from natural sources, including leaves, roots, and branches, etc. [1], which usually limited the range of colors [2]. Thereafter, various kinds of artificial dyes were explored to satisfy daily demands. According to their chemical structures and chromophores, dyes are classified as azo (mono-azo, di-azo, tri-azo, poly-azo), anthraquinone, phthalocyanine, diarylmethane, triarylmethane, indigo, azine, oxazine, thiazine, xanthene, nitro, nitroso, methine, thiazole, indamine, indophenol, lactone, amino ketone, hydroxyl ketone stibene, and sulfur dyes [3]. So far, there are over two thousand types of dyes available for commercial purposes in the market with quantities exceeding 7 <sup>×</sup> <sup>10</sup><sup>5</sup> tons worldwide annually [4,5], of which over 50% were azo dyes [4]. Azo dyes are a class of organic compounds with azo linkages (–N=N–) connecting aryl groups, which were widely used in textile, printing, paper and other fields [1,4].

Azo dyes possessed strong physical and chemical stability, extraordinary resistance to high temperature, photolysis, biological and oxidative degradation [6]. Therefore,

**Citation:** Liu, Y.; Song, J.-W.; Bao, J.; Shen, X.-J.; Li, C.-L.; Wang, X.; Shao, L.-X. Optimized Removal of Azo Dyes from Simulated Wastewater through Advanced Plasma Technique with Novel Reactor. *Water* **2022**, *14*, 3152. https://doi.org/10.3390/ w14193152

Academic Editor: Zacharias Frontistis

Received: 14 September 2022 Accepted: 4 October 2022 Published: 6 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

great quantities utilizing and inappropriate discharge of azo dyes have resulted in water pollutions and accumulations in the water cycle, threatening the safety of drinking water. Textile-processing units discharged extremely colored water effluents with dye content in the range of 10–200 mg/L [3]. For instance, the outflow concentration of the dye Acid Orange 10 from the final clarifier of a textile factory in India was 45 mg/L [7]. The outflow concentrations of azo dye from 14 Ramadan textile industries in Iraq ranged between 20 mg/L and 50 mg/L [7]. Human exposure to these azo dyes and their intermediates would cause mutagenic, carcinogenic, nervous system disorders, and dermatitis [8], as well as severe and long-lasting impact on the liver, kidney, brain, and reproduction system [9]. Therefore, increasing efforts would be warranted to implement the elimination of azo dyes from waters.

Azo dyes usually contain a strong stable benzene ring structure and azo bonds that prevent mineralization, and even dissolved functional groups of sulfonate, hydroxyl, or carboxyl, leading to difficult water treatment [10–13]. Extensive research has been done on utilizing physical and chemical techniques for the removal of azo dyes, including adsorption, chemical oxidation, biological techniques, and plasma technology [14–16], etc. The adsorption method could adsorb azo dyes effectively, but only by transferring the contaminants from water to the solid phase, without achieving the degradation effect [17], and the adsorbent regeneration was an intractable problem [18]. The chemical oxidation process could decolorize azo dyes, but the mineralization rates were usually inferior, and the addition of chemical agents was inclined to cause secondary pollution [16,19]. The biological method was used for azo dyes degradation due to their high activity and strong adaptability [20], but high levels of azo dyes could inhibit the activity of bacteria [16]. Therefore, an environmentally friendly and high efficiency degradation technique for azo dyes was required for developing to overcome the above-mentioned problems.

Plasma technology is considered as the advanced oxidation process (AOP); the plasma discharge could break molecular structure efficiently by its generation of strong active species (HO• , H•O• , HO<sup>2</sup> • , NO<sup>2</sup> • , etc.) and other physiochemical effects (UV irradiation, shockwaves, local high temperature, etc.) in situ [17,21,22]. The water molecule of solution in the discharge process leads to produce HO• and H• by its dissociation, ionization, and vibrational/rotational excitation, as shown in Equations (1)–(4) [15]. H• and O• could be generated by vibrationally/rotationally excited water molecules that are released into a lower energetic state, as shown in Equations (5)–(7) [15]. Subsequently, the enduring H2O<sup>2</sup> could be formed by dimerization of HO• (Equation (8)) [23]. In addition, plasma containing water has UV light emission as a result of exited species relaxation to lower energetic states, which are generated from the collisions between electrons and neutral molecules [15]. In the plasma treatment process, the organic molecules (M) absorb the radiation and transfer into an excited state (M\*) by ultraviolet radiation, whereafter the excited M\* could be decomposed into a new product in the transversion of M\* goes back to the ground state immediately due to short lifetime (10−9–10−<sup>8</sup> s), as shown in the Equation (9) [15]. Thus, this technique was suitable for the removal of degradation-resistant organic compounds [16]. Ma et al. have used Dielectric Barrier Discharge (DBD) plasma technology to degrade 44.36 g/L of methylene blue (MB), the degradation rate could reach up to 98.3% after 3 min [24]. Sarangapani et al., have used Box-Benkhen Design (BBD) model and Response Surface Methodology (RSM) to optimize the experimental parameters of the DBD plasma technique for Methyl Orange (MO) degradation, which determined the optimal experimental conditions with voltage at 70 kV, treatment time as 120 s, and MO concentration as 100 mg/L [25].

$$\text{Dissociation} : \text{H}\_2\text{O} + \text{e} \to \text{HO}^\bullet + \text{H}^\bullet + \text{e} \tag{1}$$

$$\text{Ionization}: \text{ H}\_2\text{O} + \text{e} \to 2\text{e} + \text{H}\_2\text{O}^+\tag{2}$$

$$\text{H}\_2\text{O}^+ + \text{H}\_2\text{O} \rightarrow \text{HO}^\bullet + \text{H}\_3\text{O}^+ \tag{3}$$

$$\text{Vibrational/rotational excitation}: \text{ H}\_2\text{O} + \text{e} \rightarrow \text{H}\_2\text{O}^\* + \text{e} \tag{4}$$

$$\mathrm{H}\_{2}\mathrm{O}^{\*} + \mathrm{H}\_{2}\mathrm{O} \rightarrow \mathrm{H}\_{2}\mathrm{O} + \mathrm{H}^{\bullet} + \mathrm{HO}^{\bullet} \tag{5}$$

$$\rm H\_2O^\* + H\_2O \to H\_2 + O^\bullet + H\_2O \tag{6}$$

$$\rm H\_2O^\* + H\_2O \to 2H^\* + O^\* + H\_2O \tag{7}$$

$$\rm HO^{\bullet} + \rm HO^{\bullet} \to \rm H\_2O\_2 \tag{8}$$

$$\mathbf{M} + \mathbf{h}\mathbf{v} \to \mathbf{M}^\* \to \mathbf{product} \tag{9}$$

However, the influence factors and degradation mechanism of plasma technology on azo dyes at the molecular level were still obscure, as well as usually followed with high energy consumption and inferior degradation efficiency due to low mass transfer efficiency between gas and liquid phase in plasma reactor [26,27]. In general, active species including HO• , H• , O• , etc. generated from plasma could penetrate into liquid, but with shallow penetration depth and attain some dozens of microns [28].

Therefore, the establishment of the interaction between active species and organic contaminants would enhance the removal efficacy. In this study, a novel reactor with an improved plasma area was designed through regulating the grounded plate and the plasma discharge area, which could increase the removal efficiency by the intensive reaction between active species and MO contaminants on the interface of gas-liquid. Moreover, the addition of typical catalysts such as persulfate (PS) and Fe2+ would present excellent degradation efficacy due to the high oxidizability of generated active species [29–32].

In order to solve the above-said problems, the plasma technique for the azo dyes was implemented in the present study to achieve the objectives as follows: (1) confirm the optimal influence parameters by comparing the plasma removal performance on the target azo dye of MO under different conditions, involving the reaction time, voltage, discharge gap, and discharge needle numbers; (2) improve MO removal efficiency and reduce energy consumption using the novel plasma reactor through the optimal response surface methodology (RSM); (3) investigate the removal efficiency of plasma associated with catalyst system including PS/Fe2+; and (4) determine the removal mechanism via the UV-Vis analysis of MO during the plasma process.

### **2. Materials and Methods**

### *2.1. Chemicals and Reagents*

Methyl Orange (MO, 99.9%) was acquired from Fluka (Steinheim, Germany). Sodium hydroxide (NaOH, 99.9%), ferrous sulfate (FeSO4, 99.9%), and sodium persulphate (Na2S2O8, 99.9%) were purchased from Acros Organics (Geel, Belgium). All solutions were prepared using ultrapure water with a conductance of 18.2 MΩ/cm (Millipore, Bedford, MA, USA).

### *2.2. Experimental Setup and Procedure*

A novel reactor was designed as a cylinder with well appropriate between the plasma discharge area and the grounded electrode in the bottom of the reactor. The novel reactor could increase the effective discharge area for active species formation during the plasma process. The fully propagated plasma occupied a roughly circular region on the liquid surface with visually uniform leader distribution [33]. As shown in Figure 1, the novel plasma reactor was an organic glass reactor with a 50 mL effective volume for MO removal from simulated wastewater. In detail, the bottom radius was 50 mm, the height was 70 mm, and the wall thickness was 5 mm. Two air vents with a radius of 4 mm were set on both sides, which were 20 mm away from the top of the reactor. A sample collection hole with a 4 mm diameter was installed at 12 mm away from the bottom of reactor. The mass transfer efficiency of active species in liquid could be improved by enhancing the air tightness of the reactor, all vents were blocked with an airtight polymer cap during the operation. The circular grounded metal electrode plate with a 45 mm radius and 2 mm thickness was immersed in the MO solution, the same size as a perforated metal plate fixed with needle-like electrodes was used as high voltage discharge electrode plate. Needle-like electrodes were stainless steels which widely used due to their mechanical and

anti-corrosion properties [34]. The nylon columns were used to support the high voltage electrode and the grounded electrode, and the high voltage electrode could be adjusted by the distance above the solution. dle‐like electrodes were stainless steels which widely used due to their mechanical and anti‐corrosion properties [34]. The nylon columns were used to support the high voltage electrode and the grounded electrode, and the high voltage electrode could be adjusted by the distance above the solution.

70 mm, and the wall thickness was 5 mm. Two air vents with a radius of 4 mm were set on both sides, which were 20 mm away from the top of the reactor. A sample collection hole with a 4 mm diameter was installed at 12 mm away from the bottom of reactor. The mass transfer efficiency of active species in liquid could be improved by enhancing the air tightness of the reactor, all vents were blocked with an airtight polymer cap during the operation. The circular grounded metal electrode plate with a 45 mm radius and 2 mm thickness was immersed in the MO solution, the same size as a perforated metal plate fixed with needle‐like electrodes was used as high voltage discharge electrode plate. Nee‐

*Water* **2022**, *14*, 3152 4 of 16

**Figure 1.** The sketch map of a novel plasma reactor. (1. Air vent 2. Sample collection hole 3. Grounded electrode plate 4. Discharge electrode plate 5. Discharge needle 6. Nylon column 7. Power lead 8. Ground lead). **Figure 1.** The sketch map of a novel plasma reactor. (1. Air vent 2. Sample collection hole 3. Grounded electrode plate 4. Discharge electrode plate 5. Discharge needle 6. Nylon column 7. Power lead 8. Ground lead).

The high voltage direct current (DC) power supply (LYZGF, Zhi‐Cheng Company, China) was used to generate plasma, characterized with the range of 0–60 kV for discharge voltage and 0–5 mA for output current. Based upon the important influence parameters, the discharge voltages were in the range between 11 kV and 15 kV, the discharge gap was between 19 mm and 21 mm, and discharge needle numbers were 3, 4, and 5, together with the initial MO concentration of 100 mg/L in simulated solution [35–37]. The added cata‐ lysts of Fe2+ and PS were set as 0.02 mmol/L under the optimal experimental conditions after the experimental factors were optimized by the RSM in the experiment of MO re‐ moval with plasma. The high voltage direct current (DC) power supply (LYZGF, Zhi-Cheng Company, China) was used to generate plasma, characterized with the range of 0–60 kV for discharge voltage and 0–5 mA for output current. Based upon the important influence parameters, the discharge voltages were in the range between 11 kV and 15 kV, the discharge gap was between 19 mm and 21 mm, and discharge needle numbers were 3, 4, and 5, together with the initial MO concentration of 100 mg/L in simulated solution [35–37]. The added catalysts of Fe2+ and PS were set as 0.02 mmol/L under the optimal experimental conditions after the experimental factors were optimized by the RSM in the experiment of MO removal with plasma.

#### *2.3. Sample Preparation and Analysis* The experimental duration of each plasma process was 60 min. Samples were taken *2.3. Sample Preparation and Analysis*

at 0 min, 10 min, 20 min, 30 min, and 60 min, the reactor effluent from MO wastewater treatment was analyzed after filtration of glass fiber membrane (0.45 μm). The samples were scanned using an ultraviolet and visible (UV‐Vis) spectrophotometer (DR 5000, Hach Company, Loveland, CO, USA) [38]. The absorbance was compared with the stand‐ ard curve (Y = 0.7563x − 0.00831 R2 = 0.998) established at the maximum absorption wave‐ length of MO 462 nm [39], subsequently the concentration and removal rate of MO was calculated separately. The TOC values of collected samples were determined by the TOC analyzer (multi NC 3100, Analytik Jena AG Company, Jena, Germany) via direct non‐ purgeable organic carbon (NPOC) measurement method to confirm the reduction of or‐ ganic substances in solutions [24]. The experimental duration of each plasma process was 60 min. Samples were taken at 0 min, 10 min, 20 min, 30 min, and 60 min, the reactor effluent from MO wastewater treatment was analyzed after filtration of glass fiber membrane (0.45 µm). The samples were scanned using an ultraviolet and visible (UV-Vis) spectrophotometer (DR 5000, Hach Company, Loveland, CO, USA) [38]. The absorbance was compared with the standard curve (Y = 0.7563x <sup>−</sup> 0.00831 R<sup>2</sup> = 0.998) established at the maximum absorption wavelength of MO 462 nm [39], subsequently the concentration and removal rate of MO was calculated separately. The TOC values of collected samples were determined by the TOC analyzer (multi NC 3100, Analytik Jena AG Company, Jena, Germany) via direct nonpurgeable organic carbon (NPOC) measurement method to confirm the reduction of organic substances in solutions [24].

### *2.4. Data Analysis*

1. Removal rate of MO (η) %

$$\mathfrak{n} = \frac{\mathbf{C}\_0 \mathbf{-} \mathbf{C}\_t}{\mathbf{C}\_0}$$

where η was the removal rate of MO in the solution, C<sup>0</sup> and C<sup>t</sup> were the concentration of MO solution at times of 0 and t, respectively.

2. Energy consumption (W) kWh/g

$$\mathbf{w} = \frac{\mathbf{I}\mathbf{U}\mathbf{t}}{\Delta \mathbf{m}}$$

where W was the energy consumption of removing organic matter per unit weight, I was the output current, U was the output voltage, t was the discharge time, ∆m was the removal weight of organic matter.

3. Mineralization rate (XTOC) %

$$\chi\_{\text{TOC}} = \frac{\text{TOC}\_0 - \text{TOC}\_0}{\text{TOC}\_0}$$

where XTOC was the mineralization rate of MO, TOC<sup>0</sup> and TOC<sup>t</sup> were total organic carbon at 0 min and t min, respectively.

### *2.5. RSM Experimental Design*

Design-experimental V8.0.6.1 statistical software (Stat-Ease, Minneapolis, MN, USA) was applied to carry out the experimental design, analysis of variance (ANOVA), mathematical modeling, and 3D response surface [13]. Plackett-Burman, Central-Composite, and Box-Behnkhen Design (BDD) were widely used methods for RSM experiments, in which BDD followed the least number of quadratic model fitting experiments at three levels. As shown in Table 1, four significant operating parameters of a novel plasma reactor involving time (min), voltage (kV), discharge gap (mm), and discharge needle numbers were optimized using BBD in this study, and the removal rate of MO was the factor of response surface analysis [40]. In this study, four parameters presented three variation levels equally, which were codified as −1, 0, and 1 [13].

**Table 1.** Response surface factor level table in novel plasma reactor.


### **3. Results and Discussion**

### *3.1. The Influence of Voltage*

Discharge plasma originated by high voltage on the gas-liquid surface would generate various active species in the solution [35]. Thus, the discharge voltage was a significant factor on the removal of azo dye during the plasma treatment. To evaluate the effect of discharge voltage on MO removal, systematic experiments were implemented on plasma treatment with different applied voltages of 11 kV, 13 kV, and 15 kV. The other experiment parameters of the discharge needle numbers were set as 5, and the discharge gap was 20 mm. As shown from the contour plot of the response surface in Figure 2, the colors changed gradually from purple to red, which demonstrated that the removal rates of MO were enhanced over time. In addition, there was an increase at the same altitude, the removal rates of MO were improved with the voltage increasing in a certain applied voltage range. The interaction of voltage and treatment time showed a positive effect for the removal rate of MO. The removal rates of MO were 85.0%, 90.3%, and 95.1% at 60 min when each applied voltage were 11 kV, 13 kV, and 15 kV, respectively. This phenomenon was consistent with supplementary free radicals generated by high applied voltage [41]. Because additional energy electrons could be produced with the application of increasing voltages, which could intensify the collision chance between energetic electrons and air molecules effectively, thereafter producing extra excited oxygen atoms that could react with water molecules to generate extra HO• [35,42]. On the other hand, the plasma treatment process could generate ultraviolet radiation that improved the removal rate of MO, and the intensity of ultraviolet radiation could be enhanced at relatively high applied voltage [35].

*Water* **2022**, *14*, 3152 6 of 16

However, high applied voltage would cost additional energy consumption and reduce energy utilization [43]. Therefore, the applied voltage was selected as 15 kV. age [35]. However, high applied voltage would cost additional energy consumption and reduce energy utilization [43]. Therefore, the applied voltage was selected as 15 kV.

range. The interaction of voltage and treatment time showed a positive effect for the re‐ moval rate of MO. The removal rates of MO were 85.0%, 90.3%, and 95.1% at 60 min when each applied voltage were 11 kV, 13 kV, and 15 kV, respectively. This phenomenon was consistent with supplementary free radicals generated by high applied voltage [41]. Be‐ cause additional energy electrons could be produced with the application of increasing voltages, which could intensify the collision chance between energetic electrons and air molecules effectively, thereafter producing extra excited oxygen atoms that could react with water molecules to generate extra HO• [35,42]. On the other hand, the plasma treat‐ ment process could generate ultraviolet radiation that improved the removal rate of MO, and the intensity of ultraviolet radiation could be enhanced at relatively high applied volt‐

**Figure 2.** The influence of voltage on the removal rate of MO. **Figure 2.** The influence of voltage on the removal rate of MO.

### *3.2. The Influence of Discharge Gap 3.2. The Influence of Discharge Gap*

Suitable discharge gap is important for MO removal in plasma, which influences the electric filed in the gap [36]. In this study, systematic experiments were conducted in plasma treatment with various discharge gaps of 19 mm, 20 mm, and 21 mm. The other parameters of the experiment including the applied voltage of 15 kV, and the number of discharge needles of 5. As shown from the contour plot of the response surface in Figure 3, the optimum removal rates of MO could achieve 89.1%, 96.7%, and 79.2% when the discharge gap was 19 mm, 20 mm, and 21 mm, respectively. The MO removal rates im‐ proved firstly when the discharge gap between 19 mm and 20 mm, but decreased subse‐ quently when the discharge gap increased to 21 mm. It was because that corona discharge would change into unstable and noisy spark discharge when the discharge gap was low excessively, the generated spark discharge was not conductive to the active species due to the gap breakdown [44], whereas the removal rate of MO decreased to 79.2% when the discharge gap amplified to 21 mm. The discharge gap was related to electric filed intensity which could influence the movement speed of gas molecules in the discharge area [45], thereafter, the increased discharge gap would weaken the electric field intensity which declined the movement speed of gas molecules in the air under the same input energy and then generate scarcer active species [44,45]. Therefore, 20 mm was chosen as the opti‐ mal discharge gap for the plasma treatment. Suitable discharge gap is important for MO removal in plasma, which influences the electric filed in the gap [36]. In this study, systematic experiments were conducted in plasma treatment with various discharge gaps of 19 mm, 20 mm, and 21 mm. The other parameters of the experiment including the applied voltage of 15 kV, and the number of discharge needles of 5. As shown from the contour plot of the response surface in Figure 3, the optimum removal rates of MO could achieve 89.1%, 96.7%, and 79.2% when the discharge gap was 19 mm, 20 mm, and 21 mm, respectively. The MO removal rates improved firstly when the discharge gap between 19 mm and 20 mm, but decreased subsequently when the discharge gap increased to 21 mm. It was because that corona discharge would change into unstable and noisy spark discharge when the discharge gap was low excessively, the generated spark discharge was not conductive to the active species due to the gap breakdown [44], whereas the removal rate of MO decreased to 79.2% when the discharge gap amplified to 21 mm. The discharge gap was related to electric filed intensity which could influence the movement speed of gas molecules in the discharge area [45], thereafter, the increased discharge gap would weaken the electric field intensity which declined the movement speed of gas molecules in the air under the same input energy and then generate scarcer active species [44,45]. Therefore, 20 mm was chosen as the optimal discharge gap for the plasma treatment. *Water* **2022**, *14*, 3152 7 of 16

**Figure 3.** The influence of discharge gap on removal rate of MO. **Figure 3.** The influence of discharge gap on removal rate of MO.

*3.3. The Influence of Discharge Needle Numbers*

discharge needles numbers in this study.

**Figure 4.** The influence of discharge needles number on removal rate of MO.

The numbers of discharge needle electrodes are related to the plasma channels be‐ tween the discharge needle and liquid surface [37]; thus, it is important to get a suitable

of discharge needle numbers on MO removal, different discharge needle numbers of 3, 4, and 5 were conducted in this plasma treatment. The other parameters were related to the applied voltage of 15 kV, and the discharge gap of 20 mm. As shown in the contour plot of the response surface in Figure 4, the removal rates of MO were improved with the growth of the treatment period and discharge needles number, subsequently the removal rate of MO was up to 95.1% at 60 min when the discharge needles number was 5. The results verified that the increase in discharge needle numbers expanded the discharge re‐ gion to promote the chemical reactions between active species and MO [46]. However, the removal rates of MO were decreased to 82.5% and 82.8% individually at 60 min when the discharge needle numbers were 3 and 4, respectively. In general, needle‐like electrodes generated fully propagated plasma channels at the liquid surface, and all types of active species were produced near plasma channels, thereafter the removal of MO mainly oc‐ curred in this circular area [33,47]. With the decline of discharge needle numbers, the plasma channels decreased, so the removal rates of MO were inferior due to depressed reaction region of active species with MO [37]. Therefore, 5 were chosen as the optimal

#### *3.3. The Influence of Discharge Needle Numbers 3.3. The Influence of Discharge Needle Numbers*

**Figure 3.** The influence of discharge gap on removal rate of MO.

*Water* **2022**, *14*, 3152 7 of 16

The numbers of discharge needle electrodes are related to the plasma channels between the discharge needle and liquid surface [37]; thus, it is important to get a suitable discharge needle numbers to enhance the removal rate of MO. To estimate the influence of discharge needle numbers on MO removal, different discharge needle numbers of 3, 4, and 5 were conducted in this plasma treatment. The other parameters were related to the applied voltage of 15 kV, and the discharge gap of 20 mm. As shown in the contour plot of the response surface in Figure 4, the removal rates of MO were improved with the growth of the treatment period and discharge needles number, subsequently the removal rate of MO was up to 95.1% at 60 min when the discharge needles number was 5. The results verified that the increase in discharge needle numbers expanded the discharge region to promote the chemical reactions between active species and MO [46]. However, the removal rates of MO were decreased to 82.5% and 82.8% individually at 60 min when the discharge needle numbers were 3 and 4, respectively. In general, needle-like electrodes generated fully propagated plasma channels at the liquid surface, and all types of active species were produced near plasma channels, thereafter the removal of MO mainly occurred in this circular area [33,47]. With the decline of discharge needle numbers, the plasma channels decreased, so the removal rates of MO were inferior due to depressed reaction region of active species with MO [37]. Therefore, 5 were chosen as the optimal discharge needles numbers in this study. The numbers of discharge needle electrodes are related to the plasma channels be‐ tween the discharge needle and liquid surface [37]; thus, it is important to get a suitable discharge needle numbers to enhance the removal rate of MO. To estimate the influence of discharge needle numbers on MO removal, different discharge needle numbers of 3, 4, and 5 were conducted in this plasma treatment. The other parameters were related to the applied voltage of 15 kV, and the discharge gap of 20 mm. As shown in the contour plot of the response surface in Figure 4, the removal rates of MO were improved with the growth of the treatment period and discharge needles number, subsequently the removal rate of MO was up to 95.1% at 60 min when the discharge needles number was 5. The results verified that the increase in discharge needle numbers expanded the discharge re‐ gion to promote the chemical reactions between active species and MO [46]. However, the removal rates of MO were decreased to 82.5% and 82.8% individually at 60 min when the discharge needle numbers were 3 and 4, respectively. In general, needle‐like electrodes generated fully propagated plasma channels at the liquid surface, and all types of active species were produced near plasma channels, thereafter the removal of MO mainly oc‐ curred in this circular area [33,47]. With the decline of discharge needle numbers, the plasma channels decreased, so the removal rates of MO were inferior due to depressed reaction region of active species with MO [37]. Therefore, 5 were chosen as the optimal discharge needles numbers in this study.

**Figure 4.** The influence of discharge needles number on removal rate of MO. **Figure 4.** The influence of discharge needles number on removal rate of MO.

### *3.4. Analysis of RSM*

RSM refers to a mathematical and statistical approach used to evaluate various aspects, including designing experiments, developing models, examining many independent variables, and assessing the optimum conditions for responses [48]. In the present study, four significant influencing parameters were selected and optimized at three levels through the ANOVA and mathematical modeling, for which, the removal rates of MO in the plasma treatment were applied as the response value. The test results were listed in Table 2.

The significance of experimental factors was judged by F-value and *p*-value in ANOVA analysis, F-value was the accuracy of generated quadratic polynomial equation computed statistically, and the experimental factor was significant when the *p*-value < 0.05. The ANOVA results for response surface of the quadratic model were listed in Table 3. MO independent variable of time (A and A<sup>2</sup> ), voltage (B and B<sup>2</sup> ), discharge gap (C<sup>2</sup> ), discharge needle numbers (D), and the interaction of AB were significant for the removal efficiency. However, the interaction of AC, AD, BC, BD, and CD were insignificant, and the independent variable of C and D<sup>2</sup> were also insignificant. Therefore, the influence parameters for the MO by the plasma technique followed the order of time (A) ≈ voltage (B) > discharge needle numbers (D) > discharge gap (C). The Predicted R<sup>2</sup> = 0.7119 was in reasonable agreement with the Adjusted R<sup>2</sup> = 0.9000 because the difference < 0.2. The signal of noise

ratio about Adep Precision = 16.954 > 4 demonstrating that the signal was acceptable due to among the desirable region. According to the parameters of *p*-value < 0.0001, R<sup>2</sup> = 0.9500, which demonstrated the prediction model was appropriate for describing the plasma experimental data. As shown in Figure 5a, all residuals spread along a straight line that demonstrated a normal probability distribution plot of MO obtained removal rates which were employed to validate the normality of studentized residuals. Furthermore, Figure 5b correspondingly exhibited the predict and actual values of removal rates as another confirmation.


**Table 2.** Results of the removal experiment of MO.

**Table 3.** MO ANOVA results for response surface of quadratic model.


*Water* **2022**, *14*, 3152 9 of 16

**Table 3.** MO ANOVA results for response surface of quadratic model.

Model 13,001.19 14 928.66 18.99 <0.0001 significant A‐Time 7866.37 1 7866.37 160.89 <0.0001 B‐Voltage 1694.80 1 1694.80 34.66 <0.0001 C‐discharge gap 5.49 1 5.49 0.11 0.7424 D‐discharge needle numbers 1019.92 1 1019.92 20.86 0.0004

> AB 271.92 1 271.92 5.56 0.0334 AC 154.75 1 154.75 3.17 0.0969 AD 85.56 1 85.56 1.75 0.2071 BC 76.30 1 76.30 1.56 0.2321 BD 136.89 1 136.89 2.80 0.1165 CD 43.10 1 43.10 0.88 0.3637 A2 443.32 1 443.32 8.86 0.0100

signal of noise ratio about Adep Precision = 16.954 > 4 demonstrating that the signal was acceptable due to among the desirable region. According to the parameters of *p*‐value < 0.0001, R2 = 0.9500, which demonstrated the prediction model was appropriate for describ‐ ing the plasma experimental data. As shown in Figure. 5a, all residuals spread along a straight line that demonstrated a normal probability distribution plot of MO obtained re‐ moval rates which were employed to validate the normality of studentized residuals. Fur‐ thermore, Figure 5b correspondingly exhibited the predict and actual values of removal

**Squares df Mean Square <sup>F</sup>‐Value** *<sup>p</sup>***‐Value** 

**Table 3.** *Cont.* B2 432.13 1 432.13 8.84 0.0101

**Source Sum of**

rates as another confirmation.

**Figure 5.** The normal probability plot of studentized residuals of MO of the model for the removal rates (**a**) and the experimental response values versus the predicted response value of MO removal rates (**b**). **Figure 5.** The normal probability plot of studentized residuals of MO of the model for the removal rates (**a**) and the experimental response values versus the predicted response value of MO removal rates (**b**).

The RSM for the removal rates of MO according to the four important parameters achieved from the plasma experiment were presented in Figure 6a–f. Four parameters were demonstrated in 3D surface plots. The interactions among the voltage, time, discharge gap, as well as discharge needle numbers were correlated to the ANOVA results for response surface of quadratic model. In addition, when the removal rate of MO was predicted as 99.3% by massive optimization values in software, correspondingly, the optimum conditions including the voltage of 14.8 kV, the plasma treatment time of 60 min, the discharge gap of 20 mm, discharge needles of 5 were confirmed. This conclusion demonstrated that the results of RSM experiment conditions were consistent with the results of previous single-factor experiment conditions. Moreover, the regression model could be conformed as a second-order response surface by the fitting analysis of multi-linear regression, as shown in the following Equation (10).

$$\begin{aligned} \text{Y} = 59.24 + 25.60 \text{A} + 11.88 \text{B} + 0.68 \text{C} + 9.22 \text{D} + 8.24 \text{AB} + 6.22 \text{AC} + 4.63 \text{A} - \text{D} - 4.57 \text{BC} - 5.85 \text{BD} - \times 100 \text{V} \\ \text{3.28} \text{CD} - 8.17 \text{A}^2 - 8.16 \text{B}^2 - 12.12 \text{C}^2 + 2.54 \text{D}^2 \end{aligned} \tag{10}$$

### *3.5. Advantages of Removal Effect of MO in Novel Plasma Reactor*

In order to verify the advantages of novel designed plasma reactor, the comparison of MO removal efficiencies applied with the same voltage of 15 kV was implemented between the novel and conventional plasma reactors (Figure S1). As shown in the Figure 7, the optimum removal rate of the novel plasma reactor could reach 95.1% at 60 min, while that of the conventional plasma reactor was only 40.5%. In addition, the energy consumption of the novel reactor for MO removal was only 0.26 kWh/g, which was much lower than that of the conventional plasma reactor with 1.1 kWh/g significantly. This might be related to the increased plasma area between the diameters of the grounded plate and the plasma discharge area, contributing to fully propagated plasma that occupied a roughly circular region on the gas-liquid surface with uniform leader distribution [33], furthermore,

increasing the mass transfer efficiency and interaction between the active species and MO contaminant. Therefore, the novel designed plasma reactor could show both advantages of higher removal efficiency and lower energy consumption during the process of MO removal treatment. Y = 59.24 + 25.60A + 11.88B + 0.68C + 9.22D + 8.24AB + 6.22AC + 4.63A − D−4.37BC − 5.85BD − 3.28CD − 8.17A2 − 8.16B2 − 12.12C2 <sup>+</sup> 2.54D2 (10)

The RSM for the removal rates of MO according to the four important parameters achieved from the plasma experiment were presented in Figure 6(a) – (f). Four parameters were demonstrated in 3D surface plots. The interactions among the voltage, time, dis‐ charge gap, as well as discharge needle numbers were correlated to the ANOVA results for response surface of quadratic model. In addition, when the removal rate of MO was predicted as 99.3% by massive optimization values in software, correspondingly, the op‐ timum conditions including the voltage of 14.8 kV, the plasma treatment time of 60 min, the discharge gap of 20 mm, discharge needles of 5 were confirmed. This conclusion demonstrated that the results of RSM experiment conditions were consistent with the re‐ sults of previous single‐factor experiment conditions. Moreover, the regression model could be conformed as a second‐order response surface by the fitting analysis of multi‐

linear regression, as shown in the following Equation (10).

*Water* **2022**, *14*, 3152 10 of 16

**Figure 6.** RSM for degradation efficiency of MO as a function of voltage and time (**a**), discharge gap and time (**b**), discharge needle numbers and time (**c**), discharge needle numbers and discharge gap (**d**), discharge gap and voltage (**e**), discharge needle numbers and voltage (**f**). **Figure 6.** RSM for degradation efficiency of MO as a function of voltage and time (**a**), discharge gap and time (**b**), discharge needle numbers and time (**c**), discharge needle numbers and discharge gap (**d**), discharge gap and voltage (**e**), discharge needle numbers and voltage (**f**). *Water* **2022**, *14*, 3152 11 of 16

of MO removal treatment. **Figure 7.** The comparison of novel and conventional reactor for the removal rates of MO (the novel plasma reactor (V = 50 mL) and the conventional plasma reactor (V = 150 mL) under the same con‐ ditions of initial concentration = 100 mg/L, voltage = 15 kV, discharge gap = 20 mm, and discharge needle numbers = 5). **Figure 7.** The comparison of novel and conventional reactor for the removal rates of MO (the novel plasma reactor (V = 50 mL) and the conventional plasma reactor (V = 150 mL) under the same conditions of initial concentration = 100 mg/L, voltage = 15 kV, discharge gap = 20 mm, and discharge needle numbers = 5).

#### *3.6. The Influence of Catalysts on MO 3.6. The Influence of Catalysts on MO*

ergistic reaction of HO• and SO4•<sup>−</sup>.

The MO removal efficiency could be further improved by adding catalysts due to the rapid interaction between active species and organic contaminants. For instance, Fe2+ could react with H2O2 in a solution, similarly to the Fenton reaction, to produce supple‐ mentary HO• [49], which was beneficial to improve the removal of organic matter greatly. PS could be easily activated by UV and heated by plasma, resulting in oxidizing to SO4•<sup>−</sup> strongly [50]. SO4•<sup>−</sup> showed a commendable oxidization effect on the improvement of deg‐ radation efficiency due to the advantages of strong stability [51]. Therefore, it is necessary to explore the influence of catalysts of PS and Fe2+ in the same experimental conditions The MO removal efficiency could be further improved by adding catalysts due to the rapid interaction between active species and organic contaminants. For instance, Fe2+ could react with H2O<sup>2</sup> in a solution, similarly to the Fenton reaction, to produce supplementary HO• [49], which was beneficial to improve the removal of organic matter greatly. PS could be easily activated by UV and heated by plasma, resulting in oxidizing to SO<sup>4</sup> •− strongly [50]. SO<sup>4</sup> •− showed a commendable oxidization effect on the improvement of degradation efficiency due to the advantages of strong stability [51]. Therefore, it is necessary to explore the influence of catalysts of PS and Fe2+ in the same experimental

optimized by RSM. As shown in Figure 8, four plasma system involving plasma,

at 60 min. The removal rates of MO could reach 97.0% and 97.3% under the plasma/Fe2+ system and plasma/PS system, respectively, when the concentration of Fe2+ and PS was both 0.02 mmol/L. Compared with the single plasma system, their removal rates of MO were enhanced slightly. This phenomenon might be related to the advanced oxidation of catalysts addition. For one thing, Fe2+ could efficiently catalyze H2O2 into advanced oxi‐ dation radical of HO• by the Fenton reaction, whereafter this active HO• could strengthen the removal efficiency of MO, as shown in Equations (11) and (12) [52]. For the otherthing, as shown in Equation (13), PS could be activated by plasma to produce SO4•<sup>−</sup> radical [16,50], thus the removal rate of MO was enhanced by the SO4•<sup>−</sup> radical attack on the MO molecules efficiently. However, compared with the plasma/PS system, the plasma/Fe2+ system presented higher removal velocity. It was because that Fe2+ could catalyze H2O2 into a large number of active HO• radicals in a short time [29]. In addition, HO• was a nonselective radical with the oxidation potential of E0 = 1.8–2.7 V, which inclined to break‐ down –N=N– bond of MO with low bond energy into a single structure resulting in de‐ coloring initially. Whereas, SO4•<sup>−</sup> radical was a strong one‐electron oxidant that would degrade aromatics selectively due to its higher oxidation potential of E0 = 2.5–3.1 V [24,53,54]. Moreover, XTOC of plasma/Fe2+ and plasma/PS system at 60 min were 16.1% and 20.7%, respectively. Higher XTOC presented by plasma/PS carried out further efforts to mineralize MO resulting in lower removal velocity. This also could explain the optimum removal efficiency of plasma/PS/Fe2+ system was attributable to the simultaneously syn‐

conditions optimized by RSM. As shown in Figure 8, four plasma system involving plasma, plasma/Fe2+, plasma/PS, and plasma/Fe2+/PS altogether performed remarkable treatment efficiency, and plasma/PS/Fe2+ exhibited the maximum removal rate for MO up to 99.2% at 60 min. The removal rates of MO could reach 97.0% and 97.3% under the plasma/Fe2+ system and plasma/PS system, respectively, when the concentration of Fe2+ and PS was both 0.02 mmol/L. Compared with the single plasma system, their removal rates of MO were enhanced slightly. This phenomenon might be related to the advanced oxidation of catalysts addition. For one thing, Fe2+ could efficiently catalyze H2O<sup>2</sup> into advanced oxidation radical of HO• by the Fenton reaction, whereafter this active HO• could strengthen the removal efficiency of MO, as shown in Equations (11) and (12) [52]. For the other thing, as shown in Equation (13), PS could be activated by plasma to produce SO<sup>4</sup> •− radical [16,50], thus the removal rate of MO was enhanced by the SO<sup>4</sup> •− radical attack on the MO molecules efficiently. However, compared with the plasma/PS system, the plasma/Fe2+ system presented higher removal velocity. It was because that Fe2+ could catalyze H2O<sup>2</sup> into a large number of active HO• radicals in a short time [29]. In addition, HO• was a nonselective radical with the oxidation potential of E<sup>0</sup> = 1.8–2.7 V, which inclined to breakdown –N=N– bond of MO with low bond energy into a single structure resulting in decoloring initially. Whereas, SO<sup>4</sup> •− radical was a strong one-electron oxidant that would degrade aromatics selectively due to its higher oxidation potential of E<sup>0</sup> = 2.5–3.1 V [24,53,54]. Moreover, XTOC of plasma/Fe2+ and plasma/PS system at 60 min were 16.1% and 20.7%, respectively. Higher XTOC presented by plasma/PS carried out further efforts to mineralize MO resulting in lower removal velocity. This also could explain the optimum removal efficiency of plasma/PS/Fe2+ system was attributable to the simultaneously synergistic reaction of HO• and SO<sup>4</sup> •−. *Water* **2022**, *14*, 3152 12 of 16 Fe2++H2O2→Fe3++HO• +OHି (11) Fe3++H2O2→Fe2++HO2 • (12)

$$\text{Fe}^{2+} + \text{H}\_2\text{O}\_2 \rightarrow \text{Fe}^{3+} + \text{HO}^\bullet + \text{OH}^- \tag{11}$$

$$\text{Fe}^{3+} + \text{H}\_2\text{O}\_2 \rightarrow \text{Fe}^{2+} + \text{HO}\_2^{\bullet} \tag{12}$$

$$\mathrm{^{1}S\_{2}O\_{8}^{2-}} \stackrel{\text{plasma}}{\rightarrow} 2\mathrm{SO}\_{4}^{\bullet-} \tag{13}$$

$$\rm{SO}\_4^{\bullet-} + \rm{OH}^- \rightarrow \rm{HO}^\bullet + \rm{SO}\_4^{2-} \tag{14}$$

$$\rm{HO}\_4^{\bullet-} + \rm{H}\_2\rm{O} \rightarrow \rm{HSO}\_4^- + \rm{HO}^\bullet \tag{15}$$

$$\rm{SO}\_4^{\bullet-} + \rm{SO}\_4^{\bullet-} \rightarrow \rm{S}\_2\rm{O}\_8^{2-} \tag{16}$$

$$\stackrel{\circ}{\mathrm{Fe}^{2+}} + \stackrel{\circ}{\mathrm{S}\_2\mathrm{O}\_8^{2-}} \rightarrow \mathrm{Fe}^{3+} + \mathrm{SO}\_4^{\bullet-} + \mathrm{SO}\_4^{2-} \tag{17}$$

tems.

**Figure 8.** The removal rates of MO under plasma, plasma/PS, plasma/Fe2+ and plasma/PS/Fe2+ sys‐ **Figure 8.** The removal rates of MO under plasma, plasma/PS, plasma/Fe2+ and plasma/PS/Fe2+ systems.

treatment process. As shown in Figure 9, the initial MO solution presented two absorption peaks of 268 nm and 463 nm. The maximum absorption at 463 nm corresponded to the π‐ π conjugated chain of the –N=N– structure, and the other one at 268 nm corresponded to

With the reaction proceeding, the wavelength of MO at 463 nm was diminished grad‐ ually, which demonstrated the decrease of MO concentration. However, the wavelength of the benzene ring monomer at 268 nm was amplified progressively, possibly because the strong π‐π conjugated system between –N=N–might be destroyed by HO• and O• in the plasma treatment process, thereafter MO was degraded into SO3 and phenylsulfinate ions [24]. Moreover, the peak of MO at 268 nm occurred blue shift slightly, demonstrating the obvious characteristic absorption peak of benzene ring. This phenomenon might be related to the occurrence of new productions involving monomer compounds of benzene‐ sulfonic acid and phenolsulfonic acid [55], due to the MO oxidized by large amounts of

the characteristic absorption peak of the benzene ring [13].

free radicals generated during the plasma treatment.

*3.7. Removal Mechanism of Plasma for MO Removal*

### *3.7. Removal Mechanism of Plasma for MO Removal*

The MO removal mechanism was proposed by the UV-Vis spectrophotometry analysis between 200 and 900 nm in the optimal experiment conditions during the plasma treatment process. As shown in Figure 9, the initial MO solution presented two absorption peaks of 268 nm and 463 nm. The maximum absorption at 463 nm corresponded to the π-π conjugated chain of the –N=N– structure, and the other one at 268 nm corresponded to the characteristic absorption peak of the benzene ring [13]. *Water* **2022**, *14*, 3152 13 of 16

**Figure 9.** Absorbance spectra of MO solution over the reaction time. **Figure 9.** Absorbance spectra of MO solution over the reaction time.

**4. Conclusions** In summary, the removal of MO in the simulated wastewater by novel plasma tech‐ nique could be feasible. Intensive studies on simulated aqueous solutions containing MO were implemented on the conditions of removal treatment involving the voltage, the numbers of discharge needle, and the distances of discharge gap. The optimal removal rates for MO could achieve 95.1% with novel plasma reactor, when the treatment duration was 60 min, the voltage was set at 15 kV, discharge gap as 20 mm, and discharge needle numbers were 5, respectively. An accurate predicted method of RSM was established to optimize the plasma treatment process for the removal of MO from simulated wastewater with the four important influencing parameters above involved. The influence factors for With the reaction proceeding, the wavelength of MO at 463 nm was diminished gradually, which demonstrated the decrease of MO concentration. However, the wavelength of the benzene ring monomer at 268 nm was amplified progressively, possibly because the strong π-π conjugated system between –N=N–might be destroyed by HO• and O• in the plasma treatment process, thereafter MO was degraded into SO<sup>3</sup> and phenylsulfinate ions [24]. Moreover, the peak of MO at 268 nm occurred blue shift slightly, demonstrating the obvious characteristic absorption peak of benzene ring. This phenomenon might be related to the occurrence of new productions involving monomer compounds of benzenesulfonic acid and phenolsulfonic acid [55], due to the MO oxidized by large amounts of free radicals generated during the plasma treatment.

MO removal followed the order of time and voltage > discharge needle numbers > dis‐

#### charge gap. The removal rate of MO was predicted as 99.3% by optimization values in **4. Conclusions**

software, the optimum conditions were confirmed as the voltage of 14.8 kV, the plasma treatment time of 60 min, the discharge gap of 20 mm, and discharge needle numbers of 5. Compared with the conventional reactor, the novel plasma reactor showed the ad‐ vantages of high removal efficiency and low energy consumption for MO removal. The investigations of plasma associated with catalysts systems revealed the optimal removal rate for MO achieved 99.2% at 60 min by plasma/PS/Fe2+ system due to simultaneous syn‐ ergistic reaction of HO• and SO4•<sup>−</sup>. Furthermore, through analyzing the variation in the absorbances with UV‐Vis spectrophotometry during plasma process for the MO removal, it was found that –N=N– bond was initially attacked and broken by active species like HO•, and the oxidized intermediates of benzenesulfonic acid and phenolsulfonic acid might be generated. As a result, the present study demonstrated the excellent and efficient removal for MO through novel plasma treatment technique, and extensive investigations about novel plasma reactor design to improve mineralization rate should be further im‐ plemented in the future. **Supplementary Materials:** The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/w14193152/s1, Figure S1: The sketch map of conventional plasma reactor. **Author Contributions:** Conceptualization, Y.L. and J.B.; Data curation, Y.L., J.‐W.S. and C.‐L.L.; For‐ mal analysis, J.‐W.S., C.‐L.L. and L.‐X.S.; Funding acquisition, J.B.; Investigation, J.‐W.S., X.‐J.S. and In summary, the removal of MO in the simulated wastewater by novel plasma technique could be feasible. Intensive studies on simulated aqueous solutions containing MO were implemented on the conditions of removal treatment involving the voltage, the numbers of discharge needle, and the distances of discharge gap. The optimal removal rates for MO could achieve 95.1% with novel plasma reactor, when the treatment duration was 60 min, the voltage was set at 15 kV, discharge gap as 20 mm, and discharge needle numbers were 5, respectively. An accurate predicted method of RSM was established to optimize the plasma treatment process for the removal of MO from simulated wastewater with the four important influencing parameters above involved. The influence factors for MO removal followed the order of time and voltage > discharge needle numbers > discharge gap. The removal rate of MO was predicted as 99.3% by optimization values in software, the optimum conditions were confirmed as the voltage of 14.8 kV, the plasma treatment time of 60 min, the discharge gap of 20 mm, and discharge needle numbers of 5. Compared with the conventional reactor, the novel plasma reactor showed the advantages of high removal efficiency and low energy consumption for MO removal. The investigations of plasma associated with catalysts systems revealed the optimal removal rate for MO achieved 99.2% at 60 min by plasma/PS/Fe2+ system due to simultaneous synergistic reaction of HO• and SO<sup>4</sup> •−. Furthermore, through analyzing the variation in the absorbances with UV-Vis spectrophotometry during plasma process for the MO removal, it

X.W.; Methodology, Y.L. and J.B.; Project administration, Y.L. and J.B.; Resources, Y.L. and J.B.; Soft‐ ware, J.‐W.S. and L.‐X.S.; Supervision, Y.L., J.B., X.‐J.S. and X.W.; Validation, J.‐W.S. and X.‐J.S.; Vis‐ ualization, L.‐X.S.; Writing—original draft, Y.L. and J.‐W.S.; Writing—review & editing, Y.L. and was found that –N=N– bond was initially attacked and broken by active species like HO• , and the oxidized intermediates of benzenesulfonic acid and phenolsulfonic acid might be generated. As a result, the present study demonstrated the excellent and efficient removal for MO through novel plasma treatment technique, and extensive investigations about novel plasma reactor design to improve mineralization rate should be further implemented in the future.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/w14193152/s1, Figure S1: The sketch map of conventional plasma reactor.

**Author Contributions:** Conceptualization, Y.L. and J.B.; Data curation, Y.L., J.-W.S. and C.-L.L.; Formal analysis, J.-W.S., C.-L.L. and L.-X.S.; Funding acquisition, J.B.; Investigation, J.-W.S., X.-J.S. and X.W.; Methodology, Y.L. and J.B.; Project administration, Y.L. and J.B.; Resources, Y.L. and J.B.; Software, J.-W.S. and L.-X.S.; Supervision, Y.L., J.B., X.-J.S. and X.W.; Validation, J.-W.S. and X.-J.S.; Visualization, L.-X.S.; Writing—original draft, Y.L. and J.-W.S.; Writing—review & editing, Y.L. and J.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China (No. 21976124 and No. 21507092), the Natural Science Foundation of Liaoning Province of China (No. 2019-ZD-0217), and Liaoning Revitalization Talents Program (No. XLYC2007195).

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

