**Photocatalysts for Organics Degradation**

Special Issue Editors

**Barbara Bonelli Maela Manzoli Francesca S. Freyria Serena Esposito**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Special Issue Editors* Barbara Bonelli PoliTO BiomED Interdepartmental Lab Italy

Serena Esposito University of Cassino and Southern Latium Italy

Maela Manzoli Universita degli Studi di Torino ` Italy

Francesca S. Freyria Massachusetts Institute of Technology USA

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Catalysts* (ISSN 2073-4344) (available at: https://www.mdpi.com/journal/catalysts/special issues/organics degradation).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Article Number*, Page Range.

**ISBN 978-3-03928-286-9 (Pbk) ISBN 978-3-03928-287-6 (PDF)**

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

## **Contents**


## **About the Special Issue Editors**

**Barbara Bonelli** (Professor of Chemistry Fundamentals for the Technologies). BB holds a PhD in Chemistry from the Universita degli Studi di Torino (Italy) and has been enrolled at Politecnico ` di Torino (Italy) since April 2001. Her main scientific interests are the physico-chemical aspects related to heterogeneous catalysis and gas adsorption, the characterization of materials by means of spectroscopic techniques and other surface techniques. She is the co-author of more than 150 papers in peer-reviewed international journals (h-index 34).

**Maela Manzoli** Professor of Industrial Chemistry at the Department of Drug Science and Technology of the University of Turin, Italy). Her studies focus on the surface properties of polycrystalline solids of catalytic interest (DRUV-Vis, FTIR spectroscopy, nitrogen physisorption), as well as their textural, morphological and structural characterization (SEM, HRTEM, XRD, XANES, EXAFS) under reaction conditions. Particular interest is dedicated to supported noble metal nanoparticles, applied to a variety of catalytic processes assisted by MW, US or mechanochemistry. She is the co-author of three book chapters and about 120 papers in peer-reviewed international Journals (h-index 38).

**Francesca S. Freyria**, after an M.Eng. in Environmental Engineering, received the European Ph.D. degree in Materials Science and Technology at Politecnico of Torino (Italy) under the supervision of Professor B. Bonelli. In 2014, she joined Professor Bawendi's group at Massachusetts Institute of Technology (Cambridge, USA) as postdoc researcher with an MIT Energy Initiative fellowship. In 2019 she won a Marie Skłodowska-Curie Individual Fellowship to develop new hybrid antenna nanomaterials for artificial photosynthesis. Her broader research interests include the study of new heterostructured nanomaterials and mesoporous materials, and how to endow them with new properties for environmental remediation and solar energy applications.

**Serena Esposito** is an Assistant Professor at the Department of Applied Science and Technology, Politecnico di Torino (Italy). Her research activities deal with the definition of synthesis strategies to prepare nanomaterials with tailored physico-chemical features. Porous, magnetic, ceramic or metal-ceramic nanomaterials are mostly obtained by the sol-gel technique, and they are used in catalysis, fuel cells, biological separations and water remediation.

## *Editorial* **Photocatalysts for Organics Degradation**

#### **Barbara Bonelli 1,\*, Maela Manzoli 2, Francesca S. Freyria 1,3 and Serena Esposito <sup>1</sup>**


Received: 11 October 2019; Accepted: 14 October 2019; Published: 21 October 2019

Organics degradation is one of the challenges of Advanced Oxidation Processes (AOPs), which are mainly employed for the removal of water and air pollutants. Compared to stand-alone processes, AOPs are more effective if combined with other removal means, especially to degrade recalcitrant (stable) pollutants in subsequent steps.

Integrated systems able to solve the aforementioned issues in a single step could be less expensive and more efficient, but their development requires advancements from the point of view of both materials and the process. In this Issue, a system consisting of integrated resin adsorption/Dielectric Barrier Discharge (DBD) plasma regeneration was proposed to treat some textile dyes, showing that the DBD plasma could maintain the efficient adsorption performance of the resin while degrading the adsorbed dye [1].

Some AOPs imply the presence of catalyst, especially in photocatalytic processes: the goal of photocatalysis is to find efficient and stable materials (under the reaction conditions), which are able both to stabilize excitons (i.e., the generated electron/hole pairs) and to exploit solar light. However, the last two goals remain very ambitious and require breakthrough advances from the point of materials science (synthesis methods) and physical chemistry. Moreover, a multi-technique approach could help in studying the surface and textural properties of the photocatalyst in order to be able to unravel the phenomena regulating excitons formation and stabilization.

Different solutions are reported in the literature, including the production of nanocomposites [2,3] and of mixed phases of TiO2 [4]. The former have to be properly designed, like in Z-Scheme g-C3N4/Fe-TiO2 [2] for the photodegradation of phenol, and in heterojunction nanostructured composites for photocatalytic fuel cells [3]: both systems were able to absorb in the Vis region. As a whole, the formation of heterojunctions in the nanocomposites simultaneously favors the photogenerated electron/hole separation and maintains the reduction and oxidation properties.

Occurrence of mixed phases is another means to promote and stabilize excitons, like in Degussa P25, where the mixed rutile/anatase phase is considered responsible for its good photocatalytic performance. Recently, mixed TiO2 phases containing brookite have been proved to display improved photocatalytic efficiency, like in N-doped anatase/brookite nanoparticles [4], obtained with high surface area by a low temperature sol-gel technique. Again, the development of new nanomaterials has been shown to have an impact on the progress of photocatalytic efficiency. Such advancements may be obtained by a plethora of synthesis methods, leading to different nanomaterials, like mixed Ni/Fe-based Metal Organic Frameworks (MOFs) [5] and Sr aluminates co-doped with Eu and Dy [6]. The former are porous networks, with high specific surface areas, where a thorough physico-chemical characterization by multiple techniques showed [5] that the occurrence of mixed-metal cluster Fe2NiO was able to enhance the photocatalytic performance further, via an electron transfer effect. The latter materials were instead prepared by different methods, namely with a hydrothermal reaction at low T and using a sol-gel method [6], pointing out the importance of developing new synthetic routes to obtain engineered (nano)materials for photocatalytic applications.

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Saturated Resin Ectopic Regeneration by Non-Thermal Dielectric Barrier Discharge Plasma**

#### **Chunjing Hao, Zehua Xiao, Di Xu, Chengbo Zhang, Jian Qiu and Kefu Liu \***

Department of Light Sources & Illuminating Engineering, Fudan University, Shanghai 200433, China; 17110720032@fudan.edu.cn (C.H.); zhxiao14@fudan.edu.cn (Z.X.); 15210720020@fudan.edu.cn (D.X.); 13110720012@fudan.edu.cn (C.Z.); jqiu@fudan.edu.cn (J.Q.)

**\*** Correspondence: kfliu@fudan.edu.cn; Tel.: +86-21-5566-5184

Received: 26 October 2017; Accepted: 22 November 2017; Published: 27 November 2017

**Abstract:** Textile dyes are some of the most refractory organic compounds in the environment due to their complex and various structure. An integrated resin adsorption/Dielectric Barrier Discharge (DBD) plasma regeneration was proposed to treat the indigo carmine solution. It is the first time to report ectopic regeneration of the saturated resins by non-thermal Dielectric Barrier Discharge. The adsorption/desorption efficiency, surface functional groups, structural properties, regeneration efficiency, and the intermediate products between gas and liquid phase before and after treatment were investigated. The results showed that DBD plasma could maintain the efficient adsorption performance of resins while degrading the indigo carmine adsorbed by resins. The degradation rate of indigo carmine reached 88% and the regeneration efficiency (RE) can be maintained above 85% after multi-successive regeneration cycles. The indigo carmine contaminants were decomposed by a variety of reactive radicals leading to fracture of exocyclic C=C bond, which could cause decoloration of dye solution. Based on above results, a possible degradation pathway for the indigo carmine by resin adsorption/DBD plasma treatment was proposed.

**Keywords:** indigo carmine; resin; Dielectric Barrier Discharge; adsorption; regeneration

#### **1. Introduction**

Industrial production processes, especially in printing and dyeing manufacturing, generate large quantities of refractory wastewater every year [1–3]. Organic chemical dyestuffs are applied as coloring material in textile industry, and are hard to degrade in normal ways, such as adsorption [4,5], biological [6], and chemical methods [7,8]. These methods have many disadvantages. On the one hand, in biological treatment, is difficult to cultivate suitable active species. On the other hand, chemical disposal often brings the problem of secondary pollution. In addition, the physical adsorption method is only a phase transfer of contaminants, and adsorbents are usually by chemical regeneration, resulting in secondary contamination of chemical reagents. Hence, systems of advanced oxidation processes (AOPs) with conventional approaches are integrated for the decomposition of organic dye contaminants, such as Fenton oxidation process (H2O2 + Fe2+) [9], ozone and UV (O3 + UV), photocatalytic oxidation (TiO2 + UV), and Non-thermal plasma (NTP) [10–12]. These methods are mainly based on the strong oxidative properties of hydroxyl radical and degradation of organic molecules. The history of Dielectric Barrier Discharge can be traced back to 1857 [13]. In 1987, Sidney first presented the technique of high voltage pulsed discharge to dispose of sewage [14]. After that, many research teams have studied the applications in various areas [15–17]. DBD plasma is widely used in the environmental protection because it produces a large number of high energy electrons, intense UV radiation, and a variety of chemical free radicals (e.g., hydroxyl radical, high energy oxygen atoms, etc.), which can rapidly react with most of the bio-refractory organic pollutants. Nevertheless, DBD plasma technology alone needs higher energy consumption, and wastewater quality factors, such as concentration, conductivity, and pH value, greatly affect the degradation effect. In particular, relatively low concentration and Chemical Oxygen Demand (COD) may greatly waste discharge energy, and more energy probably heats the wastewater solution. Hence, one of the most promising technologies, combining adsorption and DBD plasma to degrade pollutants, was introduced. Research on highly concentrated pollutants and regeneration for saturated adsorbent has been reported [18,19]. At present, the adsorbents applied in the wastewater treatment are Granular Activated Carbon (GAC), zeolite, activated alumina, etc. [20,21]. However, although Activated Carbon (AC) has been widely used in the industry, the adsorption performance of saturated AC greatly decreases after multiple regeneration. Moreover, the regeneration of AC is difficult, for example the use of heating regeneration method resulting in high carbon loss rate, or the use of pharmaceutical regeneration method resulting in high costs and secondary pollution. AC is also conductive, which is not conducive to DBD plasma discharge [22]. Based on some related literature [23–25], resin has strong adsorption properties, and can keep strong ability to absorb contaminants through repeated regeneration. The general regenerative method is eluted by mixed solution, which can lead to chemical secondary pollution [26]. Unlike the general methods, DBD discharge regeneration can achieve double effect, in which concentrated pollutants onto resins are decomposed and saturated resins are regenerated to restore the adsorption performance. At present, no literature has mentioned the study on regeneration of resin by plasma. In this article, we conducted an in-depth study that confirmed the combination of resin adsorption and DBD regeneration process can greatly improve the degradation efficiency of pollutants and reduce operating costs.

In this paper, a flat-plate reactor to investigate a facile wastewater treatment technique was designed. There are five aspects researched: (1) the adsorption behavior of resin about indigo carmine solution; (2) the regeneration efficiency for multiple cycles; (3) the functional groups and structure properties of resin before and after DBD plasma treatment; (4) the analysis of intermediate products in gas and liquid phase; and (5) a possible degradation pathway of indigo carmine contaminants by resin adsorption/DBD plasma discharge treatment system. The technique of integrated resin adsorption/DBD plasma regeneration method has very broad prospects in the field of environment protection.

#### **2. Results**

#### *Degradation Pathway Process*

The possible reaction pathway for the degradation of indigo carmine solution by absorption/DBD plasma regeneration system was proposed (Figure 1). The pathway included all of the detected intermediates and showed the active radicals as oxidant, especially the hydroxyl radical formed in the DBD discharge process. Other weak oxidants were also possible, such as H2O2 and HO2. According to the LC-MS analysis results, isatin 5-sulfonic acid (*m*/*z* 226) was the main aromatic product produced when a hydrogen radical attacked the C=C bond of indigo carmine. Isatin 5-sulfonic acid then lost SO4 <sup>2</sup><sup>−</sup> and converted to isatin. Further oxidation of the intermediate products led to a mixture of carboxylic acid and amine. Finally, those carboxylic acid and amine were degraded to inorganic molecule, including of carbon dioxide, ammonium, nitrate, etc.

**Figure 1.** Degradation pathway of indigo carmine in an integrated resin adsorption/DBD plasma.

#### **3. Discussion**

#### *3.1. Effect of Regeneration on Adsorption Capacity and Kinetics of Resin*

By comparing the adsorption isotherms of virgin resin with a series of DBD regenerated resins, the effect of DBD plasma on the adsorption capacity was analyzed. Figure 2 depicts the adsorption isotherms of indigo carmine on virgin and series of adsorption/DBD regenerated resins. It was

observed that the adsorption capacity of regenerated resin is reduced, and, as the regeneration cycle progresses, the *qe* value of the resin samples decreased.

**Figure 2.** Adsorption isotherms of indigo carmine on virgin and different saturated/DBD regeneration resins.

On the other hand, the adsorption type of indigo carmine onto resin samples after DBD plasma treatment was also studied. Generally, the Freundlich model was a kind of adsorption isotherm model, which was generally expressed by Freundlich equation (see, e.g., [27]):

$$q\_{\mathfrak{e}} = K\_F \mathbb{C}\_{\mathfrak{e}'}^{1/n} \, \tag{1}$$

where *qe* is the amount of adsorption equilibrium state, mg/g; *Ce* is the concentration of equilibrium solution, mg/L; *KF* (L/g) is the Freundlich parameter interaction with adsorption and adsorption capacity; and the exponential term of 1/*n* (dimensionless) is related to the adsorption force. ln *qe* and ln *Ce* plotted in a straight line from the slope and intercept of the straight line were the values of 1/*n* and ln *KF*, respectively. The fitting curve of the linear correlation coefficient was R2. The above three constants are listed in Table 1. The results showed that all isotherms fitted well to the Freundlich equation, which indicated that regeneration process did not seem to alter adsorption processes. All 1/*n* values were less than 1, which indicated further adsorption of indigo carmine onto resins. The adsorption isotherms of indigo carmine onto resins confirmed this phenomenon (Figure 2).

**Table 1.** Freundlich constants for adsorption of indigo carmine onto resin.


#### *3.2. Effect of Regeneration on the Regeneration Efficiency*

The residual concentration changes of indigo carmine were analyzed on virgin and DBD regenerated resins (Figure 3). Five DBD treated cycle experiments were conducted. The first to fifth DBD plasma regeneration experiments were abbreviated as DBD1, DBD2, DBD3, DBD4, and DBD5, respectively. There was only a little change in adsorption rate for DBD regenerated resins, demonstrating that adsorption rate almost kept the same level after five cycles of regeneration. Hence, the DBD regeneration efficiency could directly reveal the impact of DBD discharge process, which was calculated using the following Equation (2):

$$RE = \frac{q\_r}{q\_{\upsilon}} \times 100\% \tag{2}$$

where *qv* and *qr* are the amounts of adsorption equilibrium state of indigo carmine on virgin and regenerated resins, respectively (mg/g).

**Figure 3.** Residual concentration changes of indigo carmine solutions by virgin and different saturated/DBD regeneration resins.

All regeneration efficiencies of this process by series of regeneration cycles are presented in Figure 4. The residual concentration of the indigo carmine solution adsorbed by the virgin and regenerated resin was basically achieved, which was less than 20%. At the same time, it was also observed that, as the number of regeneration cycles increased, the degradation efficiency remained almost unchanged, indicating that the structural properties of resins remained stable and DBD plasma did not cause serious damage to the active sites on the surface of resins (discussed below). As can be seen in Figure 4, the regeneration efficiency of the resin was maintained at 80% or more even after five regeneration cycles. The experimental setup was high voltage value of 18 kV, current of 4.32 A, frequency of 1 kHz, and the degradation rate of 86%. The energy efficiency of resin adsorption/DBD plasma treatment was 139.5 g/kWh, whereas the DBD plasma treating the same concentration of indigo carmine was 56.5 g/kWh, based on the previous work [2]. The energy efficiency of adsorption/DBD regeneration was greater than 2.5 times the DBD plasma system.

The UV-Vis spectra of the resin samples at each treatment cycle are shown in Figure 5. The peaks were caused by the residual indigo carmine and intermediates onto resins after DBD regeneration. The wavelengths of 610, 450, 280, and 250 nm onto regenerated resins were observed before and after treatment of the UV-Vis spectra. The wavelength of 610 nm was characteristic absorption peak of indigo carmine. Moreover, the chromophoric group and unsaturated bond of indigo carmine correspond to the wavelengths of 610 and 250 nm, respectively. The formula of the indigo carmine is shown in the inset of Figure 5, and the bond in the bracket is the chromophoric group. The absorption intensity of all of the peaks decreased through every regeneration cycles. These results showed that chromophoric and unsaturated bonds of indigo carmine were almost broken up, which illuminated that the saturated resin was regenerated sufficiently, maintaining great degradation efficiency after multiple successive discharge cycles, which is a promising technique.

**Figure 4.** The regeneration efficiencies of resins after DBD plasma multiple cycles.

**Figure 5.** The UV-Vis spectra of virgin and DBD regeneration treatment resins.

#### *3.3. Changes in the Structural Properties of Resins*

The chemical bonds of the resin were characterized with FT-IR (fourier transform infrared spectroscopy) spectrometer. The FT-IR spectra of the three kinds of resins sample, containing virgin, saturated adsorbed resin, and adsorbed/DBD plasma regenerated resin, are depicted in Figure 6. The peaks at the wavelength of ~3420, ~2940, ~1650, ~1450, ~1100, and ~680 cm−<sup>1</sup> for all of the resin samples indicated that the resin surface functional groups were not destroyed. The broadening bond around the main peak, ~3420 cm−1, could be mainly caused by O-H stretching vibration peak in water [25]. The peak at ~3420 cm−<sup>1</sup> was a multi-absorption peak, which was widened by overlapping with nitrogen hydrogen bond (N-H) and O-H stretching vibration peaks. The absorption peak at ~2940 cm−<sup>1</sup> was mainly caused by porogen (polyethylene glycol), and residual organic liquid paraffin on the resin surface. The N-H bending vibration absorption peak corresponded to the position at ~1650 cm−1. The band of 1450 cm−<sup>1</sup> was primarily linked to the aromatic ring of C=C functional groups. At the peak of 1100 cm<sup>−</sup>1, it was generally matched with C-O stretching in the lactate and ether

groups [28]. The adsorption intensity of the saturated adsorbed resins had enhanced compared with the virgin resin, which demonstrated clearly that adsorbed contaminants onto resins could increase the intense of hydrogen bond, double bond of carbon, and carbon oxygen bond. After the multiple regenerative cycles plasma treatment, the intensity of all the absorption peaks were decelerated compared with saturated resin, which was possible on account of the adsorbed indigo carmine onto resin achieved a certain degree of degradation. Note that the bond around 680 cm−<sup>1</sup> attributed C-N bonds of indigo carmine were cleaved partially and indigo carmine was decomposed to some decolorized intermediates [29].

**Figure 6.** FT-IR spectra of the four kinds of virgin, plasma, saturated and DBD5 resins.

Apart from the analysis of functional groups, the structural characteristics of virgin, saturated, and DBD5 resins are listed in Table 2. The analysis showed that virgin and DBD5 resins exhibited similar specific surface area, total pore volume, pore size, and adsorption capacity. The analysis showed that DBD plasma regeneration process did not destroy the structure of resin. Therefore, the reason of the reduced performance of the saturated resin was that the adsorbed organic molecules occupied the adsorption site.

**Table 2.** Structural characteristics of virgin, saturated, and regenerated samples.


#### *3.4. Identification of Intermediates by GC-MS and LC-MS*

As shown in Figure 7, the GC-MS (gas chromatography-mass spectrometer) analysis exhibited six peaks related to formic acid (*m*/*z* = 29) at tr = 1.6 min, acetic acid (*m*/*z* = 43) at tr = 1.75 min, benzaldehyde (*m*/*z* = 105) at tr = 3.17 min, octamethyl-cyclotetrasiloxane (*m*/*z* = 281) at tr = 4.5 min, 4-ethyl-benzaldehyde (*m*/*z* = 134) at tr = 5.63 min, and phthalic anhydride (*m*/*z* = 104) at tr = 6.45 min. The peak position was almost identical to a previous study [30]. Note that carboxylic acids came from the heterocyclic ring opening of isatin-5-sulfonic acid sodium salt dihydrate, which was been confirmed during DBD plasma degradation of indigo carmine. Aldehyde and acid anhydride could be

formed from the oxidation of their CO-NH-CO groups. Octamethyl was a kind of siloxane copolymer, which was probably formed when silica wool was heated. The cooperation of the plasma with the resin would still produce some intermediate products. The results are listed in Table 3.

**Figure 7.** Total ion chromatogram of decomposed compositions by GC–MS analysis.


**Table 3.** Analysis of degradation products by GC-MS.

Figure 8 displays the LC-MS (liquid chromatography-mass spectrometer) analysis of the indigo carmine solution and the molecular formula of indigo carmine. Note that main charged anion of *m*/*z* 423 and its isotopic variants including *m*/*z* 423.9 (m+1) and 425 (m+2) well fitted those calculated for C16H8N2O8S2. The anion of *m*/*z* 423 was detected as the primary species in dying solution.

**Figure 8.** LC-MS analysis of initial indigo carmine solution.

The LC-MS analysis of the indigo carmine aqueous solution adsorbed by virgin resin and the molecular formula of the predominant component is shown in Figure 9. Whereas the anion of *m*/*z* 423 was not detected, ions of *m*/*z* 228.2, 229.3, 250.3, and 338.5 were clearly observed. Obviously, the components of *m*/*z* 228.2 (m+1) and *m*/*z* 229.3 (m+2) were isotopologs of isatin 5-sulfonic acid with molecular formula of C8H5NO5S [31]. The cation of *m*/*z* 250.3 was an isotopolog of 5-Isatinsulfonic acid sodium salt, which proved the fracture of C=C bond. Based on these results, the continuous formation of intermediates adsorbed onto virgin resin had a much smaller π-electron conjugated system than the initial molecule, which could result in the indigo carmine solution decoloration, as experimentally observed. To analyze the residual pollutants on the surface of the plasma regenerated resin, LC-MS of indigo carmine solution adsorbed onto resin regenerated by DBD plasma and the molecular formula of the main byproducts are shown in Figure 10. The anion of *m*/*z* 226.1 is doubly charged, as evidenced by the presence of the (M+1) isotopologs of *m*/*z* 226. The *m*/*z* for the doubly charged anions *m*/*z* 226.1 and 243.9 was 18 units, which indicated the latter molecule could be formed from the former via the incorporation of two hydroxyl groups. The anion of *m*/*z* 113.1 could probably be fitted with cyclohexylmethanamine with molecular formula of C7H15N. The peaks at other locations might be caused by residual surfactant on the surface of resin. Therefore, the formation of these intermediate products in aqueous solution was owing to fracture of the chromophoric C=C group and incorporation of oxygen atoms, hydroxyl groups, etc. Hence, the analysis of indigo carmine degraded by DBD plasma treatment by LC-MS allowed us to detect unknown byproducts and analyze the degradation pathway in the reaction. Soem of the intermediates in aqueous solution by resin adsorption/DBD plasma regeneration are listed in Table 4.

**Figure 9.** LC-MS analysis of indigo carmine solution adsorbed by virgin resin sample.

**Figure 10.** LC-MS analysis of indigo carmine solution adsorbed by plasma regenerated sample.

**Table 4.** Analysis of degradation products by LC-MS.



**Table 4.** *Cont.*

#### **4. Materials and Analysis Methods**

#### *4.1. Materials*

The resins used in this experiment were manufactured by Shaanxi LanShen Special Resin Factory, China. The type of the resin was LS-109D. The resins were pretreated based on the following steps: Firstly, the resin was soaked in the anhydrous ethanol for 24 h and washed with ethanol mixed with water in a volume ratio of 1:5 until the effluent was clear with the absence of ethanol. Secondly, the above resin was soaked in 4% hydrochloric acid for 2 h and washed to neutral with deionized water, and then in 4% sodium hydrate soaked for 2 h and washed to neutral with deionized water. Finally, the resin was dried at 60 ◦C to constant weight and placed in the dryer for reserve. The initial resin is abbreviated as Virgin. Indigo carmine was purchased from the Sinojpharm Chemical Reagent Co., Ltd. (Shanghai China). The analytical grade of all other reagents was used in the experiment (Aladdin Reagent Co., Ltd. Shanghai China). The concentration of 1000 mg/L stock solutions was made by dissolving indigo carmine powder into deionized water. The adopted concentrations in the adsorption experiment were acquired by diluting the stock solution with deionized water.

#### *4.2. The DBD Regeneration Reaction System*

The schematic diagram of DBD regeneration system is shown in Figure 11. It primarily included pulsed power supply, oxygen cylinder, and the regeneration reactor. The schematic diagram of the adsorbent-packed DBD reactor is shown in Figure 12. It was a flat type of DBD reactor. The ground electrode and high voltage electrode of DBD reactor was copper sheet. The discharge electrode was placed onto quartz barrier (80 mm × 30 mm × 2 mm). The discharge gap space between the ground and high voltage electrode was kept at 3 mm. In addition, the flat reactor port filled with quartz wool was to prevent the resin from blowing out by the carrier gas during the discharge process. As the carrier gas of oxygen, the flow rate was 3 L/min. The discharge voltage and current waveforms were recorded with the oscilloscope (Tektronix TDS 2014, Johnston, OH, USA), with a voltage probe (Tektronix P6021, Johnston, OH, USA) and a current probe (Tektronix P6021, Johnston, OH, USA), which are shown in Figure 13. The discharge parameters in the regeneration process were pulse

frequency of 1 kHz, pulse voltage of 16 kV, current of ~4.3 A, storage capacitance Cp of 3.8 pF, and reaction time of 5 min, with a high rise time of about 50 ns.

**Figure 11.** Bench-scale apparatus of DBD plasma for Indigo Carmine decomposition.

**Figure 12.** Schematic diagram of the DBD regeneration reactor: 1, high voltage electrode; 2, quartz glass; 3, plasma area; 4, resin; and 5, ground electrode.

**Figure 13.** Voltage and current waveforms delivered to the DBD regeneration reactor.

#### *4.3. Analytical Device*

The pH of solutions was measured with a FE20 meter (Mettler Toledo, Greifensee, Switzerland). The concentration of indigo carmine was detected by UV-Vis spectrophotometer (Specord® 200 Plus, Analytikjena, Jena, Germany) using the supernatant from the filtered solution and detection at the maximum wavelength of 610 nm. The analytical samples separated from treated solution were filtered with Whatman 0.45 μm PTFE membrane filter before analysis. The intermediates of samples were analyzed by Liquid Chromatograph Mass Spectrometer (Agilent 6400, Agilent Technologies Inc., Santa Clara, CA, USA) analysis with a C18 column and ultraviolet detection at 610 nm. The mobile phase was the volume ratio of 7:3 (*v*/*v*) between acetonitrile and deionized water (with 0.01% formic acid) with a flow rate of 1 mL/min. Furthermore, the filtrate of samples was extracted thrice with dichloromethane and evaporated in a vacuum evaporator (BUCHI R-300, Buchi, Flawil, Switzerland) with 40 ◦C water bath, after which a gas chromatography (Agilent 6890N) coupled with a mass selective (Agilent 5975) apparatus and a capillary column (30 mm × 0.25 mm × 0.25 mm) was utilized for identification of byproducts in gas phase during the regeneration process. The functional groups of virgin, saturated and DBD regenerated resin were a by Fourier transform-infrared (FT-IR) spectroscopy. The analytic samples were prepared by mixing 1 mg of the samples with 500 mg of KBr in an agate mortar and scanned in a range from 4000 to 400 cm<sup>−</sup>1. The structural properties of resin were obtained from the physical adsorption of N2 at 350 K determined by a Tristar II 3020 equipment. The special surface area was calculated using the BET equation [32]. To evaluate the adsorption capacity, the adsorption equilibrium isotherms of indigo carmine onto resins were measured based on the method provided by Mangun [33].

#### *4.4. The Regeneration of Indigo Carmine Saturated Resin*

The regeneration reaction of resins was carried out in the DBD reactor. Before the regeneration process, 0.25 g of saturated resins were put into the reactor. The regeneration reaction started when the power supply was open, which would last for 10 min. The resins were regenerated for five cycles in total. The first to fifth DBD plasma regeneration experiments were abbreviated as DBD1, DBD2, DBD3, DBD4, and DBD5, respectively. During DBD plasma regeneration process, the reaction temperature was not more than 35 ◦C. All experiments were carried out at atmospheric pressure.

#### *4.5. Kinetics Adsorption*

The kinetics adsorption reaction of indigo carmine onto virgin and regenerated resins were operated in oscillatory reactor. The concentration of solution after adsorption was monitored by the mentioned method.

#### *4.6. Adsorption Equilibrium Isotherms*

The adsorption isotherms of indigo carmine onto virgin and regenerated resins were operated in oscillatory reactor. Exactly 0.25 g of resins were added into a series of conical bottles containing 100 mL of indigo carmine solution of different concentration. The concentration was 5, 10, 15, 20, 30, 40, 50, 60, 80, and 100 mg/L, respectively. The conical flasks with cover were shaken with a constant speed of 120 rpm at 40 ◦C for 12 h. Then, the suspension was filtered for further analysis. Based on the standard curves of indigo carmine samples, the concentration was analyzed with UV–Vis spectrophotometer, and the amount of indigo carmine adsorbed onto resins was inferred from Equation (3):

$$q\_{\mathcal{E}} = \frac{(\mathbb{C}\_0 - \mathbb{C}\_{\mathcal{E}})V}{m},\tag{3}$$

where *qe* is the amount of indigo carmine adsorbed per gram of resin, mg/g; *V* is the volume of the liquid phase, L; *C*<sup>0</sup> is the concentration of the initial solution before it contacts with resin, mg/L; *Ce* is the concentration of the solution at equilibrium condition, mg/L; and *m* is the amount of the resin, g.

#### **5. Conclusions**

An integrated system of resin adsorption/DBD plasma regeneration method was applied for the degradation of indigo carmine solution. According to the GC-MS and LC-MS analytical results, above 85% of indigo carmine adsorbed on resin was decomposed into sulfonic acid and dehydroxylation byproducts by DBD plasma. Simultaneously, saturated resin was regenerated, and the adsorption capacity of adsorption/DBD plasma regenerated resin could be maintained at a relatively high level after multiple cycles. The functional groups, specific surface area, total pore volume, pore size, and adsorption capacity of regenerated resin did not suffer a large degree of damage. The multiple cycles of regenerative reaction indicated that resin maintained a stable and effective performance for indigo carmine adsorption. Finally, the possible degradation pathway of indigo carmine was proposed in the resin adsorption/DBD plasma regeneration process. This integrated method has a good prospect in the treatment of refractory organic wastewater.

**Acknowledgments:** The work was supported by Chinese National Nature Science Foundation under Grant 11075041.

**Author Contributions:** Chunjing Hao conceived, designed and performed the experiments; Zehua Xiao processed the reactor; Di Xu, Chengbo Zhang, Jian Qiu contributed reagents/materials/analysis tools, and Kefu Liu mainly summarized and refined the analysis; and Chunjing Hao wrote the paper.

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

#### **References**


© 2017 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 (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Fabrication of a Z-Scheme g-C3N4/Fe-TiO2 Photocatalytic Composite with Enhanced Photocatalytic Activity under Visible Light Irradiation**

### **Zedong Zhu 1, Muthu Murugananthan 2, Jie Gu <sup>1</sup> and Yanrong Zhang 1,\***


Received: 30 January 2018; Accepted: 8 March 2018; Published: 13 March 2018

**Abstract:** In the present study, a nanocomposite material g-C3N4/Fe-TiO2 has been prepared successfully by a simple one-step hydrothermal process and its structural properties were thoroughly studied by various characterization techniques, such as X-ray diffraction (XRD), Fourier Transform Infrared (FTIR) spectroscopy, electron paramagnetic resonance (EPR) spectrum, X-ray photoelectron spectroscopy (XPS), and UV-vis diffuse reflectance spectrometry (UV-vis DRS). The performance of the fabricated composite material towards the removal of phenol from aqueous phase was systematically evaluated by a photocatalytic approach and found to be highly dependent on the content of Fe3+. The optimum concentration of Fe3+ doping that showed a dramatic enhancement in the photocatalytic activity of the composite under visible light irradiation was observed to be 0.05% by weight. The separation mechanism of photogenerated electrons and holes of the g-C3N4/Fe-TiO2 photocatalysts was established by a photoluminescence technique in which the reactive species generated during the photocatalytic treatment process was quantified. The enhanced photocatalytic performance observed for g-C3N4-Fe/TiO2 was ascribed to a cumulative impact of both g-C3N4 and Fe that extended its spectrum-absorptive nature into the visible region. The heterojunction formation in the fabricated photocatalysts not only facilitated the separation of the photogenerated charge carriers but also retained its strong oxidation and reduction ability.

**Keywords:** titanium dioxide; graphitic carbon nitride; Fe doping; Z-scheme

#### **1. Introduction**

For the past several decades, many semiconducting materials have been employed as photocatalysts and their photocatalytic performance was proved to be appropriate for organic pollutant degradation, hydrogen production from water splitting, and the reduction of CO2 into fuels [1–3]. Among the studied materials, titanium dioxide (TiO2) has been widely investigated owing to its excellent photocatalytic performance, viability, nontoxic nature, and good chemical stability. However, TiO2 badly suffers from its wide band gap (3.0–3.2 eV) and low quantum efficiency, which limits its efficiency in practical applications. The conventional drawback of TiO2 as a photocatalyst is that it can be activated only in the ultraviolet light region. Hence, work on extending its absorptive behavior into the visible range and reducing the photoexcited electron–hole pair recombination rate has been carried out by several strategies, such as the doping of metal (Fe, Cu, V) [4–6] and nonmetal (N, S) [7,8] elements into the lattice of TiO2, the deposition of a noble metal (Pt, Au) [9,10] on its surface as a cocatalyst, and coupling it with an another semiconductor to form a heterojunction structure that narrows the band gap of TiO2.

The doping of transition metal into a semiconductor (TiO2) is one of the effective approaches to extend its absorptive behavior into the visible range besides improving the quantum efficiency. In particular, Fe has been considered to be a suitable candidate as the radius of both Fe3+ and Ti4+ (Fe3+: 0.69 Å; Ti4+: 0.745 Å) is almost the same, so that the incorporation of Fe into the crystal lattice of TiO2 becomes easier [11]. In addition, as the energy level of Fe2+/Fe3+ is much closer to that of Ti3+/Ti4+, Fe3+ could provide a shallow trap for photo-generated charge carriers that favors charge separation, which in turn improves the quantum yield efficiency [11].

In an another approach to enhance the efficiency of charge separation in TiO2, a heterojunction structure consisting of two different semiconductors has been demonstrated [12]. Once the heterojunction has formed between TiO2 and the coupled semiconductor material of a suitable band gap, the photoexcited electron of the lower conduction band (CB) potential of TiO2 will be promoted to the CB potential of the coupled semiconductor material, and similarly, the photoexcited hole of the higher valence band (VB) potential of TiO2 will be transferred to the VB of the coupled semiconductor material [13]. The oxidation and reduction abilities of the composite comes from those of the transferred respective photoexcited carriers, which are weaker than those of the original counterparts. As a result, though the charge separation efficiency of the composite is improved by the heterojunction, the oxidation and reduction abilities of the composite are decreased considerably [14–16]. Nevertheless, a coupling of two different semiconductors could lead to a formation of a typical Z-scheme system, in which the photoexcited electrons from the semiconductor with a less negative CB will transfer to the VB of the coupled semiconductor and combine with the photoexcited holes over there [17,18]. A composite following the Z-scheme system exhibits a higher redox capability than either of the components alone, thereby enhancing the charge separation efficiency and increasing the lifetime of charge carries as well. Owing to the above-mentioned advantages, the work on developing TiO2-based Z-scheme photocatalysts has emerged as an important research area in the recent past.

Ever since the debut work carried out on graphite carbon nitride (g-C3N4) in 2009 [19], the metal-free semiconductor has attracted the attention of scientists working in the photocatalytic domain due to its narrow band gap (2.7 eV), extreme negative CB position (−1.12 eV versus Normal Hydrogen Electrode (NHE)), structural flexibility, and good chemical stability. Although the activation of pristine g-C3N4 can be achieved in the visible light region up to 460 nm, its photocatalytic efficiency is limited due to the high recombination probability of photoexcited electron–hole pairs [20]. It is expected that coupling g-C3N4 with TiO2 would form a Z-scheme photocatalytic system and solve the problems normally encountered when using each of the semiconducting materials individually. In order to further improve the photocatalytic performance of the g-C3N4/TiO2 composite under visible-light irradiation, attempts on developing composites, such as g-C3N4-Ti3+/TiO2 and S-, N-, or Fe3+-doped TiO2/g-C3N4, have been made [21–24]. However, very few works have been done on the fabrication of g-C3N4 and Fe-doped TiO2 nanocomposite structures. Phenol is one of the most common organic water pollutants, because it is toxic even at low concentrations, and also its presence in natural waters can lead further to the formation of substituted compounds during disinfection and oxidation processes. Additionally, phenol is a model non-dye pollutant and a typical refractory aromatic compound considered to be a good probe molecule in testing photocatalytic activity for environmental purposes. The photocatalytic abatement of phenol vapors on anatase TiO2 and g-C3N4-Ti3+/TiO2 nanotubes has been the object of a study [25,26], and the mineralization process is complete in about 3–4 h and 7–8 h, respectively.

The present work focused on preparing a photocatalytic nanocomposite g-C3N4/Fe-TiO2 by a simple one-step hydrothermal process followed by a complete characterization using instrumental techniques such as X-ray Diffraction (XRD), X-ray Photoelectron Spectroscopy (XPS), Scanning Electron Microscopy (SEM), Transmission Electron Microscopy (TEM), Electron Paramagnetic Resonance (ESR), and UV-vis Diffused Reflectance spectrophotometry (UV-vis DRS). The photocatalytic activity of the as-prepared g-C3N4/Fe-TiO2 composites was investigated under the visible light region by preparative degradation experiments using phenol as model pollutant. To further confirm the enhanced activity of the as-prepared g-C3N4/Fe-TiO2 composites, a comparison experiment was carried out with an Fe-TiO2 particle photocatalyst.

#### **2. Results and Discussion**

#### *2.1. Phase Structures and Morphology*

Figure 1 shows the XRD patterns of the TiO2, Fe-TiO2, g-C3N4/TiO2 (CT), 0.05Fe-CT, and g-C3N4 samples. It can be seen that all of the TiO2-based samples exhibit identical diffraction patterns. The 2θ peaks observed at 25.3◦, 37.8◦, 48.0◦, 54.0◦, and 62.4◦ were well-matched with the standard data and correspond to the (101), (004), (200), (204), and (211) crystal planes of anatase TiO2, respectively [27]. The two prominent diffraction peaks observed at 13.6◦ and 27.7◦ for pure g-C3N4 could be attributed to the diffraction patterns of the (100) and (002) crystal planes, respectively [28]. No peak corresponding to the characteristics of g-C3N4 was observed in either the CT or 0.05Fe-CT samples, and this might be due to the relatively poor crystallization and less content of g-C3N4 within the composites [29]. In addition, there is no obvious change in the peaks of anatase TiO2 in both the composites, which indicates that neither the coupling of g-C3N4 nor the Fe-doping affects the phase structure of TiO2.

**Figure 1.** XRD patterns of TiO2, Fe-TiO2, CT, 0.05Fe-CT, and g-C3N4.

The images of SEM and TEM taken for the synthesized samples are displayed in Figures 2 and 3, respectively. The pure g-C3N4 sample exhibits a layered structure with a smooth surface that can be clearly seen in Figure 2a. This layered structure is expected to provide more sites for the growth of Fe-TiO2 nanoparticles. Figure 2b reveals that Fe-TiO2 materials consist of the aggregation of small nanocrystals. From the result shown in Figure 2c, it is found that the layered structure remains intact upon the incorporation of Fe-TiO2 nanoparticles; moreover, the surface of g-C3N4 becomes slightly rough due to the formation of Fe-TiO2 nanoparticles, suggesting that at least two semiconductors are in absolute physical contact with each other, which is the premise for the probable formation of either heterojunction or Z-scheme composites. Additionally, Energy Dispersive X-ray Spectroscopy (EDX) mapping of the composite shown in Figure 2d–h confirms the presence of Ti, O, C, N, and Fe elements in the 0.05Fe-CT sample. These results along with those of XRD and SEM suggest that Fe-TiO2 nanoparticles are successfully loaded on the surface of g-C3N4.

**Figure 2.** SEM images (**a**) g-C3N4; (**b**) Fe-TiO2; (**c**) 0.05Fe-CT; (**d**–**h**) EDX mapping of 0.05Fe-CT.

**Figure 3.** TEM (**a**,**b**) and EDX; (**c**) images of 0.05Fe-CT.

From the high resolution transmission electron microscopy (HR-TEM) image of the 0.05Fe-CT sample as shown in Figure 3a,b, a distribution of TiO2 nanoparticles with the size of ~5 nm on the surface of g-C3N4 was confirmed. The lattice spacing of TiO2 nanoparticles was found to be 0.351 nm, which matches with the (101) plane. The corresponding EDX pattern shows the existence of C, N, Ti, O, and Fe elements, which was in accordance with the results of the EDX mapping discussed earlier.

As seen in Figure 4a, the Fourier Transform Infrared (FTIR) spectrum for the g-C3N4, TiO2, CT, and 0.05Fe-CT samples shows strong bands in the region of 450–4000 cm−1. For g-C3N4, the bands observed around 1100–1650 cm−<sup>1</sup> could be assigned to C-N and C=N stretching vibrations; the band at 810 cm−<sup>1</sup> corresponds to s-triazinering vibrations; and the band around 3000–3300 cm−<sup>1</sup> is correlated to

N-H stretching vibration modes [30–33]. For the TiO2 sample, the band observed around 500–700 cm−<sup>1</sup> could be accounted for with Ti-O stretching and Ti-O-Ti bridging stretching modes; and the bands at 1630 and 3400 cm−<sup>1</sup> are correlated to the H-O-H bending stretch of surface-adsorbed water and its hydroxyl groups, respectively [34]. The characteristic bands observed for g-C3N4 and TiO2 appeared in both the CT and 0.05Fe-CT composite samples too.

**Figure 4.** FTIR spectra (**a**) and UV-vis DRS; (**b**) of g-C3N4, TiO2, CT, and 0.05Fe-CT.

As seen in Figure 4b, the optical property of the 0.05Fe-CT, CT, pure TiO2, and g-C3N4 samples was measured by UV-vis diffuse reflectance spectroscopy. The UV-vis DRS spectra of 0.05Fe-CT and CT are quite similar to that of pure TiO2, except for a slight movement of their main absorption edges toward the visible light region. In addition, upon the incorporation of g-C3N4, i.e., for the CT sample, a red shift moving up to 443 nm was observed, which indicates a reduced bandgap absorption edge of 2.80 eV estimated from 1240/λ (λ describes wavelength) [35]. Further, upon the doping of Fe to the CT sample, the light absorption of 0.05Fe-CT was extended to a still longer wavelength region and the appearance of the highest red shift, to a maximum of 461 nm (2.69 eV), was observed. These observations clearly indicate that the incorporation of composite material and the doping of Fe could graft a photocatalyst with the ability to utilize visible light effectively.

EPR as a highly sensitive spectroscopic technique for examining paramagnetic species can give valuable information about the lattice site wherein a paramagnetic doping ion is located. This technique can detect Fe ions to an extent of even less than 0.01 wt % in metal-oxide matrices [36]. The EPR spectra of TiO2, CT, 0.01Fe-CT, 0.03Fe-CT, 0.05Fe-CT, and 0.06Fe-CT are depicted in Figure 5. At high magnetic field, a symmetrical EPR signal is observed at g = 2.004 for both TiO2 and CT as well, which is an identification of the trapping of electrons on oxygen vacancies [37]. In addition, the EPR signal of CT is in accordance with that of TiO2, which strongly indicates that the presence of CNs has no influence on the phase structure of the TiO2. For the *x*Fe-CT samples, unsymmetrical signals are observed at g = 1.99, which can be assigned to the fact that the Fe3+ is substituted for Ti4+ in the octahedral surroundings/atmosphere [37,38], otherwise it could simply be an overlapping of the two kinds of EPR signals. Further, as no signal at other g values was observed, the existence of Fe ions as Fe2O3-type clusters (g = 2.16) could not be possible [39,40]. It is worth noting that the intensity of signals at g = 1.99 of *x*Fe-CT samples as shown in Figure 5a increases with increasing Fe3+ content, which indicates that the substitution of Fe3+ for Ti4+ in the TiO2 lattice was effectively accomplished by a hydrothermal approach of simply increasing the iron content in the solution mixture. As seen in Figure 5b, the EPR spectra at low magnetic field exhibited very weak signals of g value at 4.29, which suggests that Fe3+ was located in a strongly distorted rhombic environment [40]. It is clear that the specific signals of EPR spectra at both high and low magnetic field confirmed the successful incorporation of Fe3+ into the crystal lattice of TiO2 by a one-step hydrothermal method.

**Figure 5.** EPR spectra of different test field (**a**,**b**) of TiO2, 0.01Fe-CT, 0.03Fe-CT, 0.05Fe-CT, and 0.06Fe-CT.

In order to examine the chemical states of elements involved in the as-prepared samples, XPS measurements were performed. The comparison of the Ti 2p spectra for samples TiO2, Fe-TiO2, and 0.05Fe-CT is shown in Figure 6a. The Ti 2p3/2 and Ti 2p1/2 peak positions of the TiO2 sample were 458.55 eV and 464.25 eV, whereas they shifted to a higher binding energy of 458.65 eV and 467.35 eV for the Fe-TiO2 and 0.05Fe-CT samples, respectively. The small shifts of binding energy might be due to the effect of the Fe3+ in the interstitial and/or substitutional site in the TiO2 crystal lattice and formed the Ti-O-Fe bonds in the crystal lattice [41,42]. Due to the low doping level, the signals of Fe were too weak to be observed (not shown). As for the O 1s spectra presented in Figure 6b, two peaks of the binding energy at 529.95 and 532 eV for the 0.05Fe-CT sample were associated with the O2 − in TiO2 and the -OH terminal on the surface [42]. For the 0.05Fe-CT sample, the formation of the new Ti-O-Fe bonds in the crystal lattice might change the electron densities of Ti4+ cations and O2 − anions, which caused a slightly higher shift of O 1s peaks compared to those for TiO2 at 529.75 and 531.65 eV, respectively, and which might be a cause for the enhanced photocatalytic activity [42,43].

**Figure 6.** *Cont*.

**Figure 6.** High resolution XPS spectra of (**a**) Ti 2p, (**b**) O 1s, (**c**) C 1s, and (**d**) N 1s.

From the XPS spectra of C 1s in Figure 6c, three peaks centered at 285, 286.2, and 288.9 eV can be observed in all three samples. The main C 1s corresponds to adventitious carbon species presenting a band located at 284.4 eV [44]. The small shoulder at 286.4 and 288.9 eV could be accounted for with the C-N-C and C-(N)3 groups of g-C3N4, respectively [26,45,46]. In addition, the regional spectrum of N 1s for 0.05Fe-CT as seen in Figure 6d could be fitted into two peaks at 399.2 and 400.4 eV, with the former ascribed to C-N=C [34] and the latter to the N-(C)3 of the g-C3N4 [26]. No peak concerning chemical interaction between Ti and C (Ti-C) or N (Ti-N) is seen for the 0.05Fe-CT sample in the XPS (Figure 6c,d) spectra. Taking into account of the results of FTIR, SEM, and DRS studies together, the deposition of g-C3N4 could only be on the surface of the Fe-TiO2, and there was a chemical reaction between them as no apparent characteristic Ti-C(N) coordination peaks were seen.

#### *2.2. Photocatalytic Activity Test*

The photocatalytic activity of the prepared samples was evaluated in terms of photodegradation of phenol with a concentration of 10 mg dm−<sup>3</sup> under visible light irradiation. Phenol was chosen as a model pollutant because it is the basic molecule of phenolic compounds, which are known to be highly toxic, persistent, and biorecalcitrant, widely used in preservative, herbicide, and pesticide products, and being considered to be a grave threat to the health of humankind [47]. The study was carried out under similar experimental conditions using the respective photocatalytic materials.

The change of concentration of phenol during the photodegradation process under visible light irradiation is shown in Figure 7a. The decomposition of phenol was achieved to an extent of 38.6% and 72% at 80 min for pure g-C3N4 and TiO2, respectively. The effective photodegradation of phenol observed (79.2%) in the case of g-C3N4 (CT) could account for the impact of the hybrid structure on enhancing the photocatalytic activity of the material. Upon the incorporation of Fe3+ into CT, i.e., *x*Fe-CT, the observed photocatalytic activity was higher compared to that of TiO2 and CT, separately. Furthermore, the photocatalytic activity of the *x*Fe-CT increased as the content of Fe3+ increased

initially from 0.01% to 0.05%, and thereupon there was a decline trend up to 0.06%. Additionally, the 0.05Fe-CT sample showed a highest photocatalytic activity of complete degradation of phenol at 80 min under visible light irradiation. These findings confirmed that the enhanced photocatalytic activity of the composite materials cause a synergetic effect of both g-C3N4 and Fe3+.

**Figure 7.** (**a**) Change of concentration of phenol during the photodegradation process; (**b**) the kinetic constants for all samples; (**c**) the cyclic test of 0.05Fe-CT for its stability.

The degradation of phenol follows a pseudo-first-order reaction [28,48] of *ln*( *ct c*0 ) = −*kt*, where *C*<sup>0</sup> is the initial concentration of phenol, *Ct* is the concentration of phenol at time *t*, and *k* is kinetic constant. The kinetic constants (*k*) of all samples were calculated and are given in Figure 7b. It is 0.019 min−<sup>1</sup> for CT composites higher than that of either pure g-C3N4 (the *k* value was 0.0044 min<sup>−</sup>1) or TiO2 (the *k* value was 0.016 min−1). When Fe3+ ions were introduced, an enhanced photocatalytic activity was observed from all of the g-C3N4/Fe-TiO2 composites and the rate constants were 0.016 min<sup>−</sup>1, 0.034 min−1, 0.035 min−1, 0.04 min−1, and 0.029 min−<sup>1</sup> for the samples of 0.01Fe-CT–0.06Fe-CT, respectively. It was found that the incorporation of Fe3+ enhanced the photocatalytic activity of the obtained heterojunctions, and the highest performance was observed in 0.05Fe-CT.

The stability of 0.05Fe-CT was examined by catalyst recycling experiments under similar operating conditions for the photodegradation of phenol. After each cycle, the catalyst was separated by centrifugation, washed with ethanol and millipore deionized water, then dried and reused for a fresh run of photodegradation of phenol with a concentration of 10 mg dm−3. As seen in Figure 7c, no obvious decline in the degradation efficiency was observed after five cycles, suggesting that the combination of g-C3N4 and Fe-TiO2 has high-level photocatalytic stability for phenolic compound degradation.

#### *2.3. Photocatalytic Mechanism*

In order to elucidate the photocatalytic mechanism of Fe-CT composites, the main active species generated over g-C3N4, CT, 0.05Fe-CT, and TiO2 were quantified by adding a suitable scavenger during the photocatalytic degradation of phenol. The reactive species corresponding to both g-C3N4 and Fe-TiO2 were nullified as references, so that the reactive species of 0.05Fe-CT alone could be figured out. The scavengers used in this study were sodium oxalate (OA, 0.5 mmol dm−3), p-benzoquinone (BQ, 0.5 mmol dm−3), and isopropanol (IPA, 1 mmol dm−3) against photogenerated holes (h+) [49], superoxide anion radicals (•O2 <sup>−</sup>) [18], and hydroxyl radicals (•OH) [49], respectively. It needs to be mentioned that the applied concentration of each scavenger did not cause any removal of phenol in the respective control experiment [23].

As shown in Figure 8, for pure g-C3N4, the addition of OA caused a slight decrease of the photocatalytic efficiency from 38.6% to 37.1% at 80 min, which indicates that h+ was not the major reactive specie; the introduction of IPA caused a decrease to 32.5%, which indicates that •OH made a considerable contribution towards the photocatalytic degradation. When BQ was added into the reaction solution, the degradation efficiencies of phenol showed a significant fall to 10.1%. From these results, it is very clear that •O2 − is the major reactive specie in the photocatalytic reaction of pure g-C3N4. The order of influence was •O2 <sup>−</sup> <sup>&</sup>gt; •OH > h+.

**Figure 8.** Trapping of reactive specie experiments for g-C3N4, CT, 0.05Fe-CT, and TiO2.

For the TiO2, the efficiency of phenol degradation was shown to be 72.1% when no scavenger was added. With the addition of OA and IPA into the reaction solution of a separate run, the photocatalytic efficiency of phenol degradation decreased to 23.8% and 52.9%, respectively. In the presence of BQ, the degradation rate of phenol was slightly decreased to 71%. Obviously, the major reactive species for pure TiO2 are h+ and •OH.

As seen in Figure 8, for the g-C3N4/TiO2 photocatalytic system, the degradation efficiency of phenol was inhibited in the order BQ > OA > IPA when these three scavengers were added in the separate run, which indicates that •O2 <sup>−</sup>, •OH, and h+ were all of the active species generated in the g-C3N4/TiO2 photocatalytic system.

It is clear in Figure 8 that the photocatalytic efficiency of phenol for the 0.05Fe-CT photocatalyst was 100% at 80 min without any scavengers. With the addition of scavenger IPA, BQ, and OA in the separate run, the photocatalytic efficiencies of phenol decreased to 90.1%, 58.5%, and 17.6%, respectively. The inference is that both g-C3N4/TiO2 and Fe-CT showed an identical trend in the presence of scavengers, the major reactive species were •O2 <sup>−</sup> as well as h<sup>+</sup> in the photocatalytic reaction of g-C3N4/Fe-TiO2, and the order of influence was •O2 <sup>−</sup> > h+ <sup>&</sup>gt;•OH.

To further determine the photocatalytic mechanism of Fe-CT composites, a quantitative estimation of •OH was carried out by the photoluminescence (PL) method using terephthalic acid (TA) as a probe molecule during the photocatalysis process. The PL signals of g-C3N4, TiO2, Fe-TiO2, CT, and 0.05Fe-CT samples recorded at 80 min of the photocatalysis process are shown in Figure 9a. It could be easily understood that no •OH was generated during the photocatalysis process using g-C3N4 as there was no corresponding PL signal. The absence of •OH radicals in the photocatalysis process of g-C3N4 could be well-explained by taking into account the position of the VB edges of g-C3N4 and the actual potential of the OH−/•OH couple (+1.83 V/+2.7 V) (versus NHE) formation. Thus, the photogenerated holes on the surface of g-C3N4 were not strong enough to oxidize the OH<sup>−</sup> or H2O into •OH [26,50]. However, the formation of •OH was observed in the TiO2, Fe-TiO2, CT, and 0.05Fe-CT samples, among which 0.05Fe-CT showed the greatest quantity of •OH generation, which confirms the Z-scheme of transferring photoexcited charge carriers between g-C3N4 and Fe-TiO2. Otherwise, if 0.05Fe-CT worked only under the general heterojunction system, the oxidation ability of the composite would have been the same as that of g-C3N4, wherein the production of •OH is not possible.

As seen in Figure 9b, the linear potential part of the Mott–Schottky plot based on impedance measurements was used to determine the flat-band positions of the samples [51]. The positive slope of the straight lines indicates that both TiO2 and Fe-TiO2 are n-type semiconductors, i.e., the flat-band potential [52] of the samples approximately equates to the lowest potential of the CB. Thus, the CB level of TiO2 and Fe-TiO2 are measured to be ca +0.05 V (versus NHE) and −0.01 V, respectively. The negative shift of flat-band potentials (*E*fb) after Fe doping suggests a similar shift of the Fermi level, which facilitates the charge separation at the semiconductor/electrolyte interface [53].

**Figure 9.** (**a**) Photoluminescence (PL) spectra of g-C3N4, CT, Fe-TiO2, and 0.05Fe-CT in a <sup>1</sup> <sup>×</sup> <sup>10</sup>−<sup>3</sup> mol dm−<sup>3</sup> basic solution of terephthalic acid under visible light irradiation after 80 min; (**b**) The Mott–Schottky plots of TiO2 and Fe-TiO2 for determining the flat-band potentials of samples.

These accumulated electrons in the CB of TiO2 (Figure 10a) could not effectively reduce the O2 to yield •O2 <sup>−</sup> due to its CB being less negative than that of •O2 <sup>−</sup>/O2 potential (−0.28 V versus NHE) [54]. The VB of TiO2 is more positive than that of H2O/•OH potential (+2.7 V versus NHE) and capable enough to oxidize H2O to form •OH [34]. Our experiment mentioned above has also confirmed that

h+ but not the •O2 − generated was the major reactive specie in the photocatalytic degradation of the phenol molecule for pure TiO2; on the other hand, the VB levels (ca. +1.58 V) of g-C3N4 are not positive enough to drive the oxidation of H2O to form •OH, but its CB level (ca. −1.12 V) is negative enough to reduce O2 to produce •O2 <sup>−</sup> [18,49]. Also, it has been observed that the •O2 − was a major reactive specie in the photocatalytic reaction for pure g-C3N4. For the composite of CT, the formation of •OH and the significant contribution of both h<sup>+</sup> and •O2 − in the photocatalytic reaction showed that the composite followed the Z-scheme. If the charge carriers of the CT were transferred as per the so-called usual model, the electrons in the CB of g-C3N4 would have migrated to the CB of TiO2 and accumulated over there, which could possibly not reduce the O2 to yield •O2 −; holes in the VB of TiO2 would migrate to the VB of g-C3N4, which could not oxidize <sup>−</sup>OH/H2O to give •OH.

For the Fe-doped TiO2 (Fe-TiO2), a prominent decrease in the band gap and a red shift of the threshold absorption were observed in UV-vis DRS analysis. In addition, the extent of doping of Fe3+, which actually existed in the form of O••<sup>v</sup> in the band gap of TiO2, could enhance the photocatalytic activity of the material in the visible region [55,56]. As a result, the Fe-TiO2 showed higher photocatalytic activity and a greater quantity of •OH generation than those of TiO2.

Based on the above results, the Z-scheme mechanism of the g-C3N4/Fe-TiO2 composites is illustrated in Figure 10b in detail. Due to their narrow band gaps, both g-C3N4 and Fe-TiO2 can be easily excited to yield photogenerated electron–hole pairs under visible-light irradiation. Since both the CB and VB positions of Fe-TiO2 are lower than those of g-C3N4, the photogenerated electrons (e−) in the CB of Fe-TiO2 tend to transfer and recombine with the photogenerated holes (h+) in the VB of g-C3N4. The photogenerated holes left behind in the VB of Fe-TiO2 can directly oxidize phenol into harmless metabolite products. Simultaneously, the remaining photogenerated electrons in the CB of g-C3N4 can reduce the adsorbed O2 to yield •O2 −, which is again a powerful oxidative species for phenol degradation. The g-C3N4/Fe-TiO2 composites following a Z-scheme mechanism enable a fast separation and transfer of the photogenerated electron−hole pairs and in turn show strong oxidation and reduction abilities for the efficient photocatalytic degradation of organic pollutants.

**Figure 10.** (**a**) Electronic band structure of the respective catalysts; (**b**) Z-scheme photocatalytic mechanism for g-C3N4/Fe-TiO2 composites.

#### **3. Materials and Methods**

#### *3.1. Chemical and Material*

Analytical grade (AR) chemicals viz. Ferric nitrate (Fe(NO3)3·9H2O), tetra-butyl titanate (TBOT), absolute ethyl alcohol (C2H6O), nitric acid (HNO3), melamine, isopropanol, 5% Nafion, phenol sodium oxalate (OA), p-benzoquinone (BQ), and isopropanol were purchased from Sinopharm Chemical Reagent CO. Ltd., Shanghai, China and used as received. Millipore deionized water was used for preparing the stock solutions and the entire experimental part.

#### *3.2. Catalyst Preparation*

The g-C3N4 was synthesized from melamine by a direct heating step. Five grams (5 g) of melamine powder, taken in an alumina crucible, was placed in a muffle furnace and heated at 500 ◦C for 2 h. After cooling down to room temperature, the yellowish product was ground into powder form and again heated in a muffle furnace at 500 ◦C for another 2 h.

The composite particles were synthesized through a one-step hydrothermal process. A 20 mL volume of TBOT was gradually dropped into a mixture containing 167.5 mL of C2H6O, 5.0 mL specific concentration of Fe(NO3)3, 1.25 mL of HNO3, and 0.047 g of g-C3N4 under vigorous stirring, and the stirring process continued for another 1 h. Then, the mixture was transferred into a 500 mL teflon-lined stainless steel autoclave vessel and it was kept at 200 ◦C for 6 h. After the hydrothermal process, the precipitate was centrifuged, washed several times with ethanol and water, dried at 80 ◦C overnight, and ground well. The as-prepared samples were denoted as *x*Fe-CT, where *x* stands for the weight percentage of Fe (*x* = 0.01, 0.03, 0.04, 0.05, 0.06) with respect to TiO2 content, and CT denotes the g-C3N4/TiO2.

#### *3.3. Characterization*

The phase purity and crystal structure of the as-obtained samples were examined by the XRD technique using Rigaku Ultima IV X-ray diffraction (Rigaku Corporation, Tokyo, Japan) equipped with Cu *K*α radiation (40 kV, λ = 1.5406 Å). The 2θ scanning angle range was 20–80◦ with a step of 0.05 s<sup>−</sup>1. The morphology was examined using a field emission scanning electron microscope (FE-SEM, NANOSEM 450, FEI Corporation, Eindhoven, the Netherlands) operating at an accelerating voltage of 30 kV. TEM characterizations were done using an H-7000FA microscope (Hitachi, Tokyo, Japan) operating at the accelerating voltage of 75 kV. The UV-visible spectrum was obtained on a UV-2550 UV-visible spectrophotometer (Shimadzu Corporation, Kyoto, Japan) at room temperature and the spectrum range analyzed was 200–800 nm. The infrared absorption spectra were measured in a frequency range from 500 cm−<sup>1</sup> to 4000 cm−<sup>1</sup> on a Bruker V-70 FTIR spectrophotometer (Bruker, Karlsruhe, Germany). The X-band electron paramagnetic resonance (EPR) spectra were recorded at room temperature using a Bruker A300-10/12 EPR spectrophotometer (Bruker Corporation, Karlsruhe, Germany). The microwave frequency was fixed at 100 KHz, the power was 10 mW, and the field modulation ranged between 1.3–1.9 G and 3.2–3.7 G. The X-ray photoelectron spectroscopy (XPS) data were collected by an Axis Ultra instrument (Kratos Analytical, Manchester, UK) under an ultra-high vacuum (<10−<sup>8</sup> Torr) using a monochromatic Al Ka X-ray source (*hν* = 1486.6 eV) operating at 150 W.

Mott–Schottky plots have been investigated on an electrochemical workstation (CS310, CorrTest, Wuhan, China) under a three-electrode configuration by employing TiO2 or Fe-TiO2, Ag/AgCl, and Pt mesh as the working, reference, and counter electrode, respectively. Herein, the working electrodes had undergone a two-step treatment. Initially, 5 mg of synthesized sample was mixed with 800 μL of isopropanol, followed by 200 μL of millipore deionized water, and finally 20 μL of 5% Nafion, and then the mixture underwent an ultrasonication treatment; 6 μL of the mixture was dropped onto glassy carbon electrode and dried in an air environment. The supporting electrolyte used was Na2SO4 with a concentration of 0.5 mol dm−3. The potential scanning measurements for the electrode were performed from −0.2 V to 0.8 V in dark conditions, and the impedance-potential characteristics of the electrode were recorded at a frequency of 10 Hz.

#### *3.4. Photocatalytic Activity Measurement*

The photocatalytic experiment on phenol degradation under visible light was carried out in a glass container having a volume capacity of 200 mL to evaluate the activity of the g-C3N4/Fe-TiO2 composites. The light source was a 300 W PLS-SXE 300 xenon lamp (Perfect light, Wuhan, China) with a 400 nm cut filter to remove the UV irradiation that was suspended over a height of 10 cm

from the reaction solution surface. Typically, 5 mg of as-prepared photocatalyst was added into 50 mL of phenol-contaminated (10 mg dm−3) working solution. The glass container was placed in an ice-water bath, and the entire setup was placed on a magnetic stirrer operated at a constant stirring rate of 380 rpm. Prior to light irradiation, the suspension was stirred for 1 h to establish an adsorption/desorption equilibrium between phenol and photocatalyst under dark conditions. After visible light irradiation for a defined period of time (every 10 min), the reaction solution (1 mL) for analysis was siphoned out, and then the suspensions were removed by centrifugation and the clear supernatant solution was used for analysis. The concentration of phenol was measured by high performance liquid chromatography (HPLC) (Shimadzu, Kyoto, Japan) equipped with a UV detector and a C18 reverse-phase column (4.6 mm i.d. ×150 mm, Agilent, CA, USA). The mobile phase used in the HPLC was water and methanol (volumetric ratio of 50:50), and the injection volume of the sample was 20 μL with a flow rate of 0.5 mL min<sup>−</sup>1. The wavelength of the UV absorbance detector was fixed at 270 nm.

The quantity of •OH in the photocatalytic process was determined by the photoluminescence (PL) technique using terephthalic acid as a probe molecule. Terephthalic acid reacts with •OH to produce a highly fluorescent product 2-hydroxy terephthalic rapidly and specifically, which reflects as the PL signal at 425 nm excited by 315 nm of light. Detailed experimental information is given in our previous work [57].

#### **4. Conclusions**

The photocatalytic g-C3N4/Fe-TiO2 composite was successfully synthesized by a one-step hydrothermal process and found to exhibit excellent photocatalytic activity and stability for the photocatalytic degradation of organic pollutants. The composite with an optimum content of Fe3+ of 0.05 wt % exhibits the highest photocatalytic activity towards phenol degradation owing to a stronger spectral absorption of visible light wavelengths, an enhancement in the carrier density, and a decrease in the charge transfer resistance between the interface of solid and electrolyte. The formation of Z-scheme g-C3N4/Fe-TiO2 heterojunctions possesses a higher efficiency of charge separation and transfer as well as stronger oxidation and reduction abilities. This work may give new insight into the development of Z-scheme composite photocatalysts, which is of a great interest to the scientific community for photocatalysis.

**Acknowledgments:** This work was supported by the International Science & Technology Cooperation Program of China (Nos. 2013DFG50150 and 2016YFE0126300) and the Innovative and Interdisciplinary Team at HUST (2015ZDTD027). The authors thank the Analytical and Testing Center of HUST for the use of SEM, XRD, TEM, FTIR, and DRS equipment.

**Author Contributions:** Yanrong Zhang and Muthu Murugananthan conceived and designed the experiments; Jie Gu performed the experiments; and Zedong Zhu contributed analysis tools and wrote the paper.

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

#### **References**


© 2018 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 (http://creativecommons.org/licenses/by/4.0/).

## **Highly Efficient and Visible Light Responsive Heterojunction Composites as Dual Photoelectrodes for Photocatalytic Fuel Cell**

#### **Honghui Pan 1, Wenjuan Liao 1, Na Sun 1, Muthu Murugananthan <sup>2</sup> and Yanrong Zhang 1,\***


Received: 11 November 2017; Accepted: 8 January 2018; Published: 18 January 2018

**Abstract:** In the present work, a novel photocatalytic fuel cell (PFC) system involving a dual heterojunction photoelectrodes, viz. polyaniline/TiO2 nanotubes (PANI/TiO2 NTs) photoanode and CuO/Co3O4 nanorods (CuO/Co3O4 NRs) photocathode, has been designed. Compared to TiO2 NTs electrode of PFC, the present heterojunction design not only enhances the visible light absorption but also offers the higher efficiency in degrading Rhodamine B–a model organic pollutant. The study includes an evaluation of the dual performance of the photoelectrodes as well. Under visible-light irradiation of 3 mW cm−2, the cell composed of the photoanode PANI/TiO2 NTs and CuO/Co3O4 NRs photocathode forms an interior bias of +0.24 V within the PFC system. This interior bias facilitated the transfer of electrons from the photoanode to photocathode across the external circuit and combined with the holes generated therein along with a simultaneous power production. In this manner, the separation of electron/hole pair was achieved in the photoelectrodes by releasing the holes and electrons of PANI/TiO2 NTs photoanode and CuO/Co3O4 NRs photocathode, respectively. Using this PFC system, the degradation of Rhodamine B in aqueous media was achieved to an extent of 68.5% within a reaction duration of a four-hour period besides a simultaneous power generation of 85 μA cm<sup>−</sup>2.

**Keywords:** polyaniline; titanium dioxide; copper(II) oxide; cobalt oxide(II,III); photocatalytic fuel cell

#### **1. Introduction**

Water pollution, a serious issue of global concern, is no doubt a grave threat to human health and societal progress. The contamination of natural water systems is mainly due to lack of effective and viable techniques and excessive discharge of wastewater containing toxic organic contaminants. Developing an effective purification technique to maintain a green ecological environment and simultaneously recover the chemical energy stored abundantly in toxic organics that usually let out as wastewater has become an urgent need of the hour [1–3]. A novel device, the so-called photocatalytic fuel cell (PFC), constituted with a photoanode and a photocathode, for wastewater treatment along with simultaneous electricity generation under solar irradiation, is an emerging and attractive technique in the energy and environmental domain [4–6]. In this system, the electron/hole pairs can be generated from the photoelectrodes under light irradiation in a defined wavelength region. The electrons produced from the photoanode leave the holes and transfer through the external circuit to the photocathode, and the holes at the photoanode are released for degradation of organic compounds [7]. In addition, the PFC system, in which an n-type semiconductor generally used as

photoanode with a Fermi level higher than that of the cathode, could develop interior bias which facilitates the transfer of electrons from photoanode to photocathode thereby producing a concurrent generation of electricity [8]. The existing PFC systems constituting TiO2 and Pt as photoelectrodes [3,8] have severe limitations because TiO2 responds most to ultraviolet (UV) region light and suffers from the high probability of electron/hole pair recombination [9]. Using Pt as photocathode is obviously not a viable approach, which would eventually restrict its application to a large-scale level [10].

To overcome the above mentioned drawbacks, the studies on either developing visible light responsive photoanodes [11–14] or replacing Pt by p-type semiconductor as the photocathode [15,16] become equally important, leading to the development of dual photoelectrodes for PFC system. However, so far, these PFCs have been identified with its shortcomings that the photoactivity and photostability of the electrodes are poor, which limits their application. For instance, the visible-light driven PFC system using CdS/TiO2 or WO3/W as photoanode and Cu2O/Cu as photocathode suffers from poor stability of the CdS, WO3 and Cu2O due to their photocorrosion nature in aqueous media [17]. The limited usage of these electrodes could be attributed largely to the inherent drawbacks of the material that result in poor response in visible-light region, weak stability and undesirable photoactivity, which eventually limit the performance of the PFC system.

Forming a heterojunction by two different semconductors is an effective strategy to facilitate the hole/electron seperation and enhance the photocatalytic activity [18,19]. For instance, polyaniline (PANI), a conducting polymer, might be a good choice for TiO2 sensitization [20] due to high absorption coefficients in the visible-light region, high mobility of charge carriers and good environmental stability. PANI/TiO2 nanocomposite could be obtained by mixing commercial TiO2 nanopowder with PANI by a chemical oxidative polymerization step [20,21]. On the other hand, oxides of copper and cobalt, which are well known for their p-type semiconducting behavior, could be used as photocathode [22–24] replacing the noble metals and also as photocatalysts for degradation of pollutants [25]. These materials can withstand in the multiple processing steps and have a compatibility nature with other material systems. All these notable characteristics behavior make them attractive and interesting base materials for heterostructures. Chopra et al. [25,26] recently established a fact that CuO nanowire−Co3O4 nanoparticle heterostructure has shown a unique photoactivity under visible-light irradiation. The p–p junctions formed by the combination of CuO and Co3O4 could efficiently reduce the probability of recombination of photogenerated electron/hole pairs, which in turn enhances the photocatalytic activity.

In this work, a pair of materials, viz. PANI/TiO2 nanotubes (NTs) as photoanode and CuO/Co3O4 nanoparticles (NRs) as photocathode fabricated based on Ti substrate, was proposed as a novel PFC system, which exhibits an effective degradation behavior toward Rhodamine B and shows an efficient generation of electricity.

#### **2. Results and Discussion**

#### *2.1. Characterization of PANI/TiO2 NTs Photoanode*

The microstructure and elements distribution of TiO2 and PANI/TiO2 NTs were studied by using Scanning Electron Microscope (SEM) and Energy Dispersive Spectrometer (EDX) (Figure 1), respectively. The TiO2 nanotube arrays were covered by a layer of discretely adhered PANI. As seen in the inset (Figure 1a), the cross-sectional view of TiO2 nanotube arrays substrate clearly displays the vertically oriented nanotubes with a length of about 900 nm and a wall thickness of 10 nm. Additionally, from the EDX analysis, the existence of elements, viz. C, N, Ti and O, was confirmed and especially the minimum quantity of PANI with respect to the content of TiO2 substrate material was confirmed by the relatively low intensities observed against the elements C and N. Figure 2 shows the X-ray diffraction patterns (XRD) recorded for both TiO2 and PANI/TiO2 NTs materials. The peaks presented for PANI/TiO2 NTs reflect characteristics of anatase TiO2 and the predominant peak of 2θ at 25.2◦ indicates a fine preferential growth of the Titania nanotube (TNTs) in 101 orientation

(JCPDS no. 21-1272). The fact that no diffraction peak was observed for PANI might be due to its amorphous phase in the composite. The position and shape of the peaks observed in XRD patterns for PANI/TiO2 NTs were almost identical with that of TiO2, indicate that the incorporation of PANI has no influence in the lattice structure of TiO2, which would be an added advantage for the hybrid photocatalytic material.

**Figure 1.** (**a**) SEM image of the TiO2 (the inset shows the cross-sectional image) and (**b**) polyaniline (PANI)/TiO2 nanotubes (NTs), (**c**) EDX analysis of (**b**).

**Figure 2.** XRD patterns recorded for TiO2, and PANI/TiO2 NTs.

Figure 3 shows the Fourier Transform Infrared Spectroscopy (FTIR) spectra recorded for TiO2 and PANI/TiO2 NTs as well. The wide peak observed at 500–800 cm−<sup>1</sup> for TiO2 could be ascribed to the Ti–O bending mode of TiO2 sample. The strong characteristic absorption bands observed for PANI/TiO2 NTs, between 1200 and 1600 cm−1, were 1566, 1487, 1299, 1245 and 1127 cm−1. The bands at 1566 and 1487 cm−<sup>1</sup> could be correlated to C–C stretching mode of quinonoid and benzenoid units, repectively. The bands at 1299 and 1245 cm−<sup>1</sup> represented the C–N stretching mode of benzenoid unit while the band at 1127 cm−<sup>1</sup> reflects the plane bending vibration of C=N. The bands at 796 cm−<sup>1</sup> represented the C–H stretching mode of benzenoid rings [27]. Furthermore, as seen in Figure 4, the optical responses investigated by UV–vis Diffuse Reflectance Spectra (DRS) for TiO2 and PANI/TiO2 NT samples exhibit a notable absorption extension in the visible-light region at 420 nm upon the incorporation of PANI, which corresponds to a reduced bandgap absorption edge of 2.9 eV. It could be inferred from the red shift of the absorption wavelength that the PANI/TiO2 NTs would be an effective visible-light driven photocatalytic material.

**Figure 3.** FTIR spectra of the TiO2 and PANI/TiO2 NTs.

**Figure 4.** UV–vis diffuse reflection spectra of the TiO2 and PANI/TiO2 NTs.

In order to understand the separation and recombination of electron–hole pairs that take place in the photocatalytic materials, the photocurrent and the electrochemical impedance spectra (EIS) measurements were carried out under visible-light irradiation. The transient photocurrent responses of TiO2 and PANI/TiO2 NTs electrodes were recorded via several on–off cycles of irradiation, and the

representative traces observed are shown in Figure 5a. Obviously, the intensity of photocurrent response was found to be higher for PANI/TiO2 NTs (50 μA cm−2) than that for TiO2 NTs. The respective Nyquist plots of the TiO2 and PANI/TiO2 NTs photoelectrodes were shown in Figure 5b. The semicircle at high frequencies was characteristic of the charge transfer process and its diameter was equal to the charge transfer resistance. The PANI/TiO2 NTs sample showed a smaller semicircle than that of TiO2 sample in the Nyquist plots. This clearly confirms that the rate of electron transfer between the interface of PANI/TiO2 NTs and the electrolyte was improved as a result of the deposition of PANI which causes the enhanced photoelectrochemical activity of the former compared with that of the latter.

The effects analysis of radicals was carried out in the present study to establish the PEC degradation mechanism as it is proved to be an effective approach in predicting the photodegradation reaction pathways of organic molecules that take place on the surface of the photocatalyst. The nature of interaction between the chosen scavenger and the photocatalyst makes a prominent impact on the efficiency of organic pollutant degradation. The scavengers used in this study were sodium oxalate (Na2C2O4) of 0.5 mmol L−<sup>1</sup> [28,29], isopropanol of 1 mmol L−<sup>1</sup> [30], Cr(VI) of 0.05 mmol L−<sup>1</sup> [28], and p-benzoquinone of 0.5 mmol L−<sup>1</sup> [29,30] against h+, •OH, e<sup>−</sup> and O2 •−, respectively.

**Figure 5.** (**a**) Photocurrent responses of TiO2 and PANI/TiO2 NTs in 0.1 mol L−<sup>1</sup> Na2SO4 at a bias potential of +0.6 V (vs. saturated calomel electrode (SCE)); (**b**) Nyquist plots of TiO2 and PANI/TiO2 NTs measured at open circuit potential under irradiation.

As shown in Figure 6, in the absence of a scavenger, the PEC degradation of Rhodamine B on TiO2 sample at three hours was to an extent of 53%, and it was decreased to 29.4%, 35.8% and 32%, with the addition of scavengers Na2C2O4, isopropanol and p-benzoquinone, respectively, as a separate experiment. However, in the case of Cr(VI) addition, no prominent difference was observed in the efficiency of PEC degradation (52.6%), which could be attributed to the following facts. The addition of Cr(VI) accepts photoelectron and suppresses the reduction of oxygen that results in a decreased production of O2 •−, which in turn restrains the degradation of Rhodamine B. On the other hand, Cr(VI) inhibits the recombination of the photoinduced electron and the hole to a certain extent, which could reversely promote the efficiency of PEC degradation. Hence, the addition of Cr(VI) has no impact on the PEC degradation of Rhodamine B. It could be inferred that the major reactive species formed on pure TiO2 were h+, O2 •− and •OH.

For the PANI/TiO2 NTs, under similar experimental conditions, the PEC degradation of Rhodamine B decreased from 77%, an actual efficiency obtained without scavenger, to 27.7%, 45% and 28.4% with the addition of Na2C2O4, isopropanol and p-benzoquinone, respectively. Further, with the addition of Cr(VI) scavenger, the PEC degradation of Rhodamine B decreased to an extent of 60.4%. The results suggest that the major reactive species formed on PANI/TiO2 NTs photocatalyst were e−, <sup>h</sup>+, •OH, and O2 • with an order of influence as h<sup>+</sup> > O2 •− > •OH > e−.

**Figure 6.** Effects of different scavengers on the PEC degradation of Rhodamine B (0.05 mmol L−<sup>1</sup> Cr(VI): e<sup>−</sup> scavenger, 1 mmol L−<sup>1</sup> isopropanol: •OH scavenger, 0.5 mmol L−<sup>1</sup> p-benzoquinone: O2 •−, 0.5 mmol L−<sup>1</sup> sodium oxalate: h+ scavenger).

#### *2.2. Characterization of CuO/Co3O4 NRs Photocathode*

The SEM images recorded for as-prepared Co3O4 and CuO-coated Co3O4 (CuO/Co3O4) nanorods on the Ti substrate, are shown in Figure 7a,b respectively. As seen, the diameter of the former was observed to be about 150 nm. The CuO/Co3O4 nanorods were fabricated by conducting 30 cycles of pulsed electrodeposition in the aqueous media containing both CuSO4 and lactic acid, followed by an annealing step. As the deposition cycles increase, the CuO NPs started covering the surface of Co3O4 NRs (Figure 7c) gradually and upon 40 cycles, the entire surface was completely covered by CuO NPs which makes it weaken in the adsorption of incident light. The XRD pattern (Figure 8b) revealed the crystal structure and phase purity of both Co3O4 NRs and CuO/Co3O4 NR heterostructures. For Co3O4 NRs, all peaks in the pattern could be indexed using the Co3O4 anatase phase (JPCDS No: 42-1467), and the intense peak of 2θ at 19.0◦, 31.2◦ and 36.5◦ could be correlated to (111), (200) and (311) plane diffractions, respectively. With the loading of CuO NPs, an additional peak of 2θ at 35.5◦ was observed in the (111) orientation [31]. This indicates that the deposit made on Co3O4 NRs was only in the form of CuO and not as Cu or Cu2O. As seen in the UV–vis DRS recorded for CuO/Co3O4 NR sample (Figure 8b), a strong absorption was observed in the visible-light region with the band gap energy of 2.33 eV by a linear extrapolation in the absorption edge of the spectrum.

**Figure 7.** SEM images of the Co3O4 (**a**), CuO (30)/Co3O4 (**b**) and CuO (40)/Co3O4 (**c**).

**Figure 8.** (**a**) XRD of the Co3O4 and CuO (30)/Co3O4, (**b**) UV–vis DRS of CuO (30)/Co3O4.

The influence of content of CuO NRs on the PEC performance of CuO/Co3O4 was studied. The NRs were fabricated by pulsed electrodeposition of different cycles, viz. 10, 20, 30 and 40. Figure 9a shows the comparative transient photocurrent response observed in applying the alternative on–off visible-light illumination cycles at −0.25 V (vs. SCE). The CuO/Co3O4 NRs showed an instant photoresponse under irradiation, and the photocurrent densities started increasing initially as the coating cycle increases from 10 to 30, followed by a decrease with a further increase up to 40 cycles. The maximum photocurrent density of about 170 μA cm−<sup>2</sup> was observed for CuO/Co3O4 NRs at a coating cycle of 30. Figure 9b shows a linear sweep study for CuO/Co3O4 NRs processed in the potential range of −0.35 V to +0.01 V (vs. SCE) under chopped visible-light irradiation with a scan rate of 0.5 mV s<sup>−</sup>1. With a cathodic potential scanning, the photocurrent was observed to be increased gradually, which is in accordance with the property of a p-type semiconductor [31]. The CuO/Co3O4 NRs, prepared by 30 cycles of pulsed electrodeposition was chosen as the photocathode for the PFC system of present study as it exhibits the best photoactivity.

**Figure 9.** (**a**) PEC performance of the composite samples prepared at different pulse cycles at −0.25 V (vs. SCE) under visible-light irradiation in 0.1 mol L−<sup>1</sup> Na2SO4 aqueous solution and (**b**) Linear sweep voltammetry (LSV) curves of CuO(30)/Co3O4 in 0.1 mol L−<sup>1</sup> Na2SO4 solution in dark and under visible-light irradiation.

#### *2.3. Characterization of PFC System and Its Performances*

Figure 10 shows the Mott–Schottky (MS) plots depicted as 1/C<sup>2</sup> vs. potential at 100 Hz for the respective PANI/TiO2 NTs and CuO/Co3O4 NRs samples. The slopes of the linear part of the curves in the MS plot for the PANI/TiO2 NTs were positive, which is a characteristic behavior of typical n-type semiconductor. The linear parts of the curves were *x*-extrapolated to zero, to obtain the *V*fb value [32,33] of ca. −0.25 V vs. SCE for the PANI/TiO2 NTs (Figure 10a), which represents its conduction band edge (CB). Conversely, the p-type characteristic behavior of CuO/Co3O4 NRs was verified by a negative slope in the MS plot, as seen in Figure 10b. The valence band (VB) edge +0.58 V vs. SCE was approximately equal to the flatband position.

**Figure 10.** Mott–Schottky plots measured at a frequency of 100 Hz of (**a**) PANI/TiO2 NTs, (**b**) CuO/Co3O4 NRs in the dark.

The energy band positions of the photoanode and photocahotde are illustrated in Figure 11a. As the Fermi level of CuO/Co3O4 NRs is more positive than that of PANI/TiO2 NTs, an interior bias could be formed by connecting the two photoelectrodes directly, which would obviously drive the electrons generated from PANI/TiO2 NTs through the external circuit and combine with the holes generated in CuO/Co3O4 NRs. Meanwhile, the holes and the electrons remained in the respective photoelectrode can be very well utilized for degradation of organic pollutant. It is actually the key factor that makes the PEC technique successful by combining n-type photoanode and p-type photocathode.

The open circuit potential (*E*ocp) was established from the difference in the Fermi level of the two photoelectrodes [32,33]. To examine the photoelectric properties of the PFC, the photovoltage curves of the PFC system of PANI/TiO2-CuO/Co3O4 was measured in the dark and under irradiation. *E*ocp of the PANI/TiO2 NTs photoanode and the CuO/Co3O4 NRs photocathode were measured to be −0.13 V and 0.12 V, respectively, under visible-light irradiation (3 mW cm−2). It implies that the photovoltage between the photocathode and the photoanode would be +0.25 V which is consistent with the measured value (+0.24 V) of the PFC system, as shown in Figure 11b. As a result, the separation of the electron/hole pair in the photoelectrodes could be facilitated in parallel under visible-light irradiation.

**Figure 11.** (**a**) Energy level diagram of the PFC cell for organic compounds degradation and electricity generation, (**b**) The open-circuit voltage of PFC cell of PANI/TiO2-CuO/Co3O4 in dark and under visible-light irradiation.

#### *2.4. Degradation of Rhodamine B*

The performance of the PFC system was evaluated by a degradation study on Rhodamine B contaminated aqueous solution under visible-light irradiation. The degradation efficiency was monitored in terms of decolorization of Rhodamine B. The photocatalytic activity of various systems using different types of photocatalysts was compared under incandescent light irradiation as shown in Figure 12a. As seen, the photocatalytic activity of the system in which the photoelectrodes are not externally interconnected, was found to be inferior to the others and showing a decolorization of Rhodamine B of only 25.4%. For the PFC system of different photoelectrode couples TiO2-CuO/Co3O4, and PANI/TiO2-CuO/Co3O4 the decolorization was 51% and 68%, respectively at same reaction period. Figure 12b demonstrates that the short-circuit current density curve obtained for the present PFC system (PANI/TiO2-CuO/Co3O4) during the process of Rhodamine B decolorization, was relatively steady with a current density of 85 μA cm−<sup>2</sup> throughout the process. The consistent photocurrent density observed for the PFC confirmed its photostability and durability for long-time application.

**Figure 12.** (**a**) Comparison of the degradation rates of Rhodamine B in the photocatalytic decomposition processes using unconnected PANI/TiO2 and CuO/Co3O4 photoelectrodes, the PFC systems of TiO2CuO/Co3O4 and PANI/TiO2-CuO/Co3O4, (**b**) the generated electricity of the PFC.

#### **3. Materials and Methods**

#### *3.1. Chemical and Material*

Titanium foil with a thickness of 1 mm and a purity of 99.9% was purchased from Strem Chemicals (Newburyport, MA, USA). The chemicals such as ethylene glycol (EG), ammonia fluoride (NH4F), sodium sulfate (Na2SO4), phenylamine (C6H5NH2), cobalt nitrate hexahydrate (Co(NO3)2·6H2O), hexamethylenetetramine (C6H12N4, HMT), sodium persulfate (Na2S2O3) and HCl were purchased from Acros Organics (Pittsburgh, PA, USA) and used as received. The aqueous solution used was prepared by using a millipore deionized (DI) water with a resistivity of 18.2 MΩ cm.

#### *3.2. Preparation of PANI/TiO2-NTs*

The self-organized TiO2 nanotube arrays (TiO2-NTs) were fabricated on Ti foil substrate by anodization method using ethylene glycol (EG) as electrolyte media which contains 0.5 wt % NH4F and 10 vol % water. The fabrication process was described in detail in our previous studies [34]. The anodization of Ti foil was performed with a two-electrode electrochemical system employing Pt mesh as cathode at a constant operating potential of 20 V for a period of 2 h. The inter electrode gap was fixed as 3 cm for every electrolysis run. In the post treatment, the anodized sample was washed with millipore deionized water, dried at 70 ◦C and calcined at 450 ◦C for 2 h.

The PANI/TiO2 NTs composite was synthesized by a sequential chemical bath deposition (SCBD) method. Typically, the TiO2 NTs was successively immersed into four different glass beakers for about

30 min in each beaker. The first beaker contained aqueous solution of 0.27 mol L−<sup>1</sup> of phenylamine, and the third one contained an aqueous mixture of 0.23 mol L−<sup>1</sup> of sodium persulfate and 0.15 mol L−<sup>1</sup> of HCl, and the other two contained distilled water to rinse the samples to scavenge the excess of each precursor solution. Such an immersion treatment cycle was repeated thrice.

#### *3.3. Preparation of CuO/Co3O4*

The synthesis of Co3O4 electrode was accomplished by a simple hydrothermal process [35]. 7.2 g of Co(NO3)2·6H2O, 0.13 g of NH4F, and 0.3 g of HMT were dissolved in the order in a 50-mL acetone−deionized water (*v*/*v* = 50:50) mixture solution under continuous stirring using magnetic stirrer. Upon a formation of pink suspension, the stirring was continued for another 10 min. Then the suspension, together with a Ti film, was transformed to a teflon-lined stainless-steel autoclave vessel and kept for 24 h at 95 ◦C. The pink-depositions-covered-Ti film was obtained by these steps, and carefully rinsed with deionized water and dried at 70 ◦C, followed by a calcination process at 350 ◦C for 1 h in air environment. The transformation of pink depositions into black one upon calcinations confirmed the formation of Co3O4.

CuO was prepared by a pulsed galvanostat method under high current conditions [36]. The electrodeposition was carried out in a conventional three-electrode electrochemical workstation (CS310, CorrTest, Wuhan, China) with a conditioned electrolyte solution of 0.4 mol L−<sup>1</sup> CuSO4 and 3 mol L−<sup>1</sup> lactic acid fixing the pH at 7 by NaOH and the temperature at 25 ◦C. The concentrated lactic acid acts as a complex agent for the stabilization of copper ions [37]. Upon subjecting to a negative current pulse for 0.5 s followed by a constant current density of 50 mA for 7 s, the surface of the Co3O4 was covered with Cu nanoparticles (NPs). The as-prepared electrode was carefully rinsed with millipore deionized water and dried at 70 ◦C, followed by a calcination process at 350 ◦C for 1 h. Then the samples were rinsed with ethanol, followed by a heat treatment at 450 ◦C for 1 h in air environment. In order to optimize the deposition of CuO NPs on the Co3O4, the samples were fabricated at different pulse cycles, viz. 10, 20, 30 and 40.

#### *3.4. Characterization*

The morphology and microstructure of the synthesized samples were characterized by field emission scanning electron microscopy (FE-SEM; NANOSEM 450, FEI, Eindhoven, The Netherlands). The phase and elemental composition of the samples were investigated using X-ray Diffraction Technique (XRD; PW3040/60 PANalytical, Almelo, The Netherlands) with Cu K α radiation (λ = 1.54056 Å). UV–visible spectrum scanning was carried out in the range of 200–800 nm using a UV-2550 model UV–visible spectrophotometer (Shimadzu Corporation, Kanagawa, Japan) at room temperature. The infrared absorption spectra were measured on a Bruker V-70 Fourier transform infrared (FTIR, Bruker, Karlsruhe, Germany) spectrophotometer in the frequency range of 400 to 4000 cm<sup>−</sup>1.

The photoresponsive test was carried out for the sample (either PANI/TiO2 NTs or CuO/Co3O4 NRs) used as working electrode in a three-electrode electrochemical work station (CS310, CorrTest, Wuhan, China), wherein saturated calomel electrode (SCE) and Pt foil was used as reference and auxiliary electrodes, respectively. The electrochemical impedance spectroscopic (EIS) studies were performed between 100 kHz and 0.01 Hz with a 5 mV rms sinusoidal modulation at the open circuit potential of the system under illumination. The linear sweep was evaluated under chopped light irradiation with a scan rate of 0.5 mV s−1. Mott–Schottky plots were measured at a frequency of 100 Hz. The electrochemical studies described above were carried out in a 0.1 mol L−<sup>1</sup> Na2SO4 aqueous solution at room temperature. The light source used was a 11 W incandescent lamp (PHILPS, Amsterdam, The Netherlands) that produced irradiation with an intensity of 3 mW cm−<sup>2</sup> to the test sample which was measured by a visible-light radiometer (FZ-A, Wuhan, China).

The photoelectrochemical characteristics of the PFC were examined by connecting PANI/TiO2 NTs electrode and CuO/Co3O4 NRs electrode directly. The short circuit current plot, and the open circuit potentials plot as well as the characteristic nature of photocurrent potentials were tested by digit precision multimeter (Tektronix DMM4050, Johnston, OH, USA) and the electrochemical workstation, respectively.

#### *3.5. Photoelectrocatalytic Degradation of Phenol under Visible-Light Irradiation*

Photoelectrocatalytic oxidation experiments were carried out in a glass container having volume capacity of 150 mL with a standard three-electrode configuration using synthesized PANI/TiO2-NTs as photoanode, a Pt foil and a SCE as counter and reference electrodes, respectively. The photoelectrochemical degradation experiments were performed with a working volume of 45 mL aqueous solution containing a model contaminant Rhodamine B (1 × <sup>10</sup>−<sup>5</sup> mol L−1) along with 0.1 mol L−<sup>1</sup> Na2SO4 as supporting electrolyte. The glass container was placed in a water bath wherein the temperature was constantly maintained as 298 K, and the entire set-up was placed on a magnetic stirrer operated at a constant stirring rate of 650 rpm during the process. Prior to the light irradiation, the experimental solution was stirred in the dark for ca. 30 min to establish the adsorption/desorption equilibrium between the organic contaminant and the surface of the PANI/TiO2-NTs under ambient air equilibrium. The degradation rate of Rhodamine B was followed by using a UV–vis spectrophotometer (UV2102 PCS, UNICO, Shanghai, China) in which the wavelength was fixed at 554 nm.

The PFC degradation of Rhodamine B (1 × <sup>10</sup>−<sup>5</sup> mol L−1) was performed by exposing the light on both the PANI/TiO2-NTs photoanode and CuO/Co3O4 photocathode with the illumination area of <sup>2</sup> × 2 cm2 under similar conditions to those followed in the photoelectrocatalytic experiment. The PFC current was measured by using a digit precision multimeter.

#### **4. Conclusions**

A highly efficient and visible-light responsive photocatalytic fuel cell (PFC) system involving a dual heterojuntion PANI/TiO2 photoanode and CuO/Co3O4 photocathode was constructed. The results obtained showed that a photocurrent of 50 μA cm−<sup>2</sup> was achieved using the PANI/TiO2 photoanode at a bias potential of +0.6 V (vs. SCE) in 0.1 mol L−<sup>1</sup> Na2SO4 electrolyte under visible-light irradiation of 3 mW cm−2, which was 150% higher than that of TiO2. Additionally, the optimized CuO/Co3O4 photocathode exhibited a photocurrent of 170 <sup>μ</sup>A cm−<sup>2</sup> at −0.25 V (vs. SCE). The PFC was constructed with the aim of providing an internal bias potential to the photoelectrocatalytic system and the performance and working mechanism of the same were systematically investigated. Under visible-light irradiation, the interior bias (+0.24 V) developed, drives the electrons of the PANI/TiO2-NT's photoanode across the external circuit to combine with the holes of the CuO/Co3O4 photocathode, which actually leads to electron/hole pair separation at respective photoelectrodes. The results obtained in the study suggest that the PFC system involving dual heterojuntion PANI/TiO2 photoanode and CuO/Co3O4 photocathode is very effective for wastewater treatment along with simultaneous electricity generation.

**Acknowledgments:** This work was supported by the International Science & Technology Cooperation Program of China (Nos. 2013DFG50150 and 2016YFE0126300) and the Innovative and Interdisciplinary Team at HUST (2015ZDTD027). The authors thank the Analytical and Testing Center of HUST for the use of SEM, XRD, TEM, FTIR and DRS equipments.

**Author Contributions:** Yanrong Zhang and Muthu Murugananthan conceived and designed the experiments; Wenjuan Liao and Na Sun performed the experiments; Honghui Pan contributed analysis tools; Wenjuan Liao wrote the paper.

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

#### **References**


© 2018 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 (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Low-Temperature Sol-Gel Synthesis of Nitrogen-Doped Anatase/Brookite Biphasic Nanoparticles with High Surface Area and Visible-Light Performance**

## **Liang Jiang, Yizhou Li, Haiyan Yang, Yepeng Yang, Jun Liu, Zhiying Yan, Xiang Long, Jiao He and Jiaqiang Wang \***

National Center for International Research on Photoelectric and Energy Materials (MOST), Yunnan Provincial Collaborative Innovation Center of Green Chemistry for Lignite Energy, Yunnan Province Engineering Research Center of Photocatalytic Treatment of Industrial Wastewater, The Universities' Center for Photocatalytic Treatment of Pollutants in Yunnan Province, School of Energy, School of Chemical Sciences & Technology, Yunnan University, Kunming 650091, China; liangjiang\_ynu@163.com (L.J.); zh111111ou@sina.com (Y.L.); ashoulu@sina.com (H.Y.); mondaysunday1234@163.com (Y.Y.); 18468068607@163.com (J.L.); zhyyan@ynu.edu.cn (Z.Y.); lxbl1991@163.com (X.L.); hejiao@ynu.edu.cn (J.H.)

**\*** Correspondence: jqwang@ynu.edu.cn; Tel.: +86-871-6503-1567

Received: 21 October 2017; Accepted: 30 November 2017; Published: 4 December 2017

**Abstract:** Nitrogen doping in combination with the brookite phase or a mixture of TiO2 polymorphs nanomaterials can enhance photocatalytic activity under visible light. Generally, nitrogendopedanatase/brookite mixed phases TiO2 nanoparticles obtained by hydrothermal or solvothermal method need to be at high temperature and with long time heating treatment. Furthermore, the surface areas of them are low (<125 m2/g). There is hardly a report on the simple and direct preparation of N-doped anatase/brookite mixed phase TiO2 nanostructures using sol-gel method at low heating temperature. In this paper, the nitrogen-doped anatase/brookite biphasic nanoparticles with large surface area (240 m2/g) were successfully prepared using sol-gel method at low temperature (165 ◦C), and with short heating time (4 h) under autogenous pressure. The obtained sample without subsequent annealing at elevated temperatures showed enhanced photocatalytic efficiency for the degradation of methyl orange (MO) with 4.2-, 9.6-, and 7.5-fold visible light activities compared to P25 and the amorphous samples heated in muffle furnace with air or in tube furnace with a flow of nitrogen at 165 ◦C, respectively. This result was attributed to the synergistic effects of nitrogen doping, mixed crystalline phases, and high surface area.

**Keywords:** anatase/brookite biphasic; nitrogen-doping; sol-gel method; visible light photocatalysis; degradation of dyes

#### **1. Introduction**

Heterogeneous photocatalytic processes involving TiO2 semiconductor particles have been shown to be a promising process for the treatment of dye effluents [1]. However, large band gap energy (3.2 eV) for anatase TiO2 limits its practical application for natural solar applications [2]. To develop more light-efficient catalysts, there is an urgent need to develop photocatalytic systems which are able to operate effectively under visible light irradiation. A number of systems have been reported to improve the visible-light activity of TiO2. Meanwhile, selecting the reasonable substrate and activity test are helpful to systematically and comprehensively assess the photocatalytic efficiency of the catalysts [3]. Nitrogen-doped (N-doped) TiO2 is one of the most typical examples of the visible-light photocatalysts, which is due to nitrogen doping can decrease the band gap energy and enhance the

photoactivity of TiO2 in the visible spectral range [4,5]. However, the low reactivity and quantum efficiency of N-doped TiO2 limit its practical application [6,7].

On the other hand, TiO2 exists in three main polymorphs, which are anatase, rutile, and brookite [8,9]. Phase mixing is well recognized as the most promising strategy for quantum efficiency improvement, which can be due to the enhanced charge carrier separation [6,10–14]. Particularly, it has been proven that the mixed phase of anatase/rutile TiO2 has synergistic effects and higher photocatalytic activity as compared to pure phase of either in anatase or rutile [15,16]. In contrast to anatase/rutile biphasic nanoparticles which have been intensively studied, the photocatalytic study of brookite and its phase mixing is quite limited, though it has been reported that anatase/brookite mixed-phase TiO2 has higher activity in visible light than P25 [8]. The reason may be mainly due to the difficulties in synthesis [17]. For example, anatase–brookite composite nanocrystals were synthesized by a sonochemical sol-gel method at very high heating temperature (500 ◦C) [12,18]. Highly crystalline phase-pure brookite and anatase/brookite mixed-phase TiO2 nanostructures were synthesized via a simple hydrothermal method with titanium sulphide as the precursors in sodium hydroxide solutions [19]. Interestingly, anatase-brookite heterojunction TiO2 photocatalysts were purposefully tailored by introducing different glycine concentrations through hydrothermal treatment at 200 ◦C for 20 h [20].

It is expected that a strategy coupling a binary structure with nitrogen doping could bring enhanced photocatalytic properties of TiO2. Recently, N-doped anatase/rutile TiO2 nanoparticles have been designed and synthesized [19,21]. Anatase/brookite mixed-phase nitrogen-doped TiO2 nanoparticles were also synthesized by a facile solvothermal route [22]. Interestingly, nitrogen plasma treatment was employed to prepare N-doped nanoporous TiO2 with large surface area and high-crystalline anatase/brookite phase [23].

Generally, a semiconductor catalyst with large specific surface area is beneficial for efficient photocatalysis, while in most synthetic processes, TiO2 with the brookite phase or a mixture of TiO2 polymorphs obtained hydrothermally at high temperature and with long time heat treatment have low surface area [14,19,23,24]. Hence, it is challenging to synthesize N-doped anatase/brookite TiO2 photocatalyst with large surface area and enhanced visible light activity at low temperature via simple and direct synthetic method.

Sol-gel is one of the most prominent methods used to prepare mixed phases of anatase/rutile TiO2 nanoparticles due to its simplicity and low equipment requirements. However, there are few reports on the simple and direct preparation of N-doped anatase/brookite mixed phase TiO2 nanostructures using sol-gel method at low heating temperature [15,16]. The goal of the present work is to synthesize anatase/brookite biphasic TiO2 nanoparticles by direct introduction of nitrogen in TiO2 lattice crystal during the sol-gel preparation at low temperature. In this work, the degradation of methyl orange (MO) in aqueous solution under visible light irradiations was selected to test the enhanced photocatalytic efficiency. It has been reported that amorphous TiO2 or a mixture composed of crystalline and amorphous TiO2 has high activity for the photocatalytic degradation of pollutants [25,26]. However, synthesized nitrogen-doped anatase/brookite biphasic nanoparticles of this work exhibited much higher photocatalytic efficiency than the prepared amorphous samples.

#### **2. Results and Discussion**

#### *2.1. Syntheses and Characterizations*

The synthesis process of this work was modified from a typical sol-gel method by using HNO3-catalyzed hydrolysis step of titanium tetraisopropoxide (TTIP) to reduce the hydrolysis rates [27]. Generally, heating is required to prepare crystalline TiO2. If low heating temperature in the range of 180–200 ◦C was employed in hydrothermal or solvothermal method, longer time (3–48 h) would be needed. Nevertheless, the obtained TiO2 samples still have low surface area (<125 m2/g) [18,21,23]. Moreover, a supercritical drying process was often used in the conventional sol-gel method [28]. By contrast, herein the aged gels were heated under nitrogen atmosphere with a much lower final autogenous pressure (about 350 psi), heating temperature (165 ◦C), and shorter heating time (4 h).

The crystal structures of sample NA-185 and NA-165 with anatase and brookite phases were identified by X-ray diffraction (XRD), as shown in Figure 1. The diffraction peaks of 2θ values at 25.3◦, 37.8◦, 48.1◦, 54.9◦, 62.8◦, 69.8◦, 75.4◦, 82.8◦ are assigned to the (101), (004), (200), (204), (220), (215), and (224) planes of anatase TiO2 (JCPDS 21-1272). Due to the overlapping of the planes of brookite (120), brookite (111), and anatase (101), the existence of the brookite phase was determined by the brookite (121) plane at 30.8◦ (JCPDS 29-1360). Both NA-145 and HA-165 are phase-pure anatase. From the XRD peak intensities [29], the brookite phase contents of sample NA-185 and NA-165 were calculated to be ~10% and ~6%, respectively. The crystal size was calculated by Scherrer equation (Table 1).

**Figure 1.** X-ray diffraction (XRD) patterns of as-prepared samples: (**a**) NA-185, (**b**) NA-165, (**c**) NA-145, (**d**) HA-165.

**Table 1.** The characteristics and the apparent first-order rate constant K (min<sup>−</sup>1) of samples.


**<sup>a</sup>** Determined by the Scherrer equation; **<sup>b</sup>** Calculated using the formula in reference [29].

The morphology and particle size of the samples revealed by Scanning electron microscopy (SEM) and Transmission electron microscopy (TEM) analysis (Figures S1 and S2). NA-185, NA-165, NA-145, and HA-165 all show aggregates consisting of small spheroidal nanoparticles with average size of approximately 6–8 nm, which was in agreement with the results calculated by Scherrer equation. To further confirm the existence of anatase and brookite phases, high-resolution TEM (HRTEM) analysis of NA-165 was performed. As shown in Figure 2a,b, the lattice fringes of 0.35 nm and 0.29 nm match the anatase (101) and brookite (121) plane, respectively. The results are in agreement with the XRD observations.

**Figure 2.** High-resolution TEM (HRTEM) images of NA-165. The lattice fringes of 0.35 nm and 0.29 nm match (**a**) the anatase (101) and (**b**) brookite (121) plane.

The X-ray photoelectron spectroscopy (XPS) measurements reveal the surface compositions and chemical states of the samples with the presence of N, O, Ti, and C. The N 1s peak of NA-165 at 400.1 eV can be attributed to the interstitial nitrogen in the form of Ti-O-N or Ti-N-O bonds (Figure 3a) [30]. Moreover, the Ti 2p2/3 and Ti 2p1/2 core levels were located at 458.4 and 464.2 eV (Figure 3b), which shift toward lower binding energies as compared to the reported pure TiO2 due to the nitrogen doping [31]. The nitrogen doping percentages of NA-185, NA-165, NA-145, and HA-165 were 0.66, 0.63, 0.73, and 0.52 at.%, respectively. Since the nitrogen content of NA-165 prepared under N2 using HNO3 as catalyst is higher than that of HA-165 prepared under N2 via the similar method under N2 but using HCl instead of HNO3 as catalyst, it implied that the nitrogen source in NA-165 may be from both N2 and HNO3 [22,32]. The O 1s XPS spectra of NA-165 shown in Figure 3c displays two peaks at 530.2 and 531.8 eV, which was attributed to the Ti–O bond and Ti-O-N or Ti-N-O, respectively [33]. The XPS results along with XRD patterns and HRTEM images reveal that nitrogen-doped anatase/brookite biphasic nanoparticles were successfully synthesized.

**Figure 3.** X-ray photoelectron spectroscopy (XPS) spectra of (**a**) N 1s, (**b**) Ti 2p, (**c**) O 1s region for NA-165.

The nitrogen adsorption-desorption isotherms shown in Figure S3 are all Type IV, implying that the samples may have mesoporous structures. The surface areas, average pore size, and pore volumes of the samples are summarized in Table 1 and Table S1. Obviously, the surface areas of the biphasic samples changed little with the increase of heating temperature, since the surface areas of NA-145, NA-165, and NA-185 were 249, 240, and 239 m2/g, respectively. HA-165 prepared using HCl instead of HNO3 as catalyst had lower surface area (216 m2/g). Compared with other methods, the employed heating temperature of this work was much lower, and the heat treatment time was shorter. Nevertheless, the surface area of NA-165 was also much higher than those of many other types of nitrogen-doped anatase/brookite biphasic TiO2 except for the one treated with nitrogen plasma [22].

The UV-vis diffuse reflectance spectra of NA-185, NA-165, NA-145, and HA-165 are shown in Figure 4 using P25 as a control group. The absorbance of the N-doped samples was stronger than that of P25 in the visible light region. The band gap energies of NA-185, NA-165, NA-145, HA-165, and P25 were 3.05, 3.03, 3.01, 3.09, and 3.12 eV, respectively, which were calculated from Equation (1):

$$\text{Eg} = 1240/\lambda,\tag{1}$$

where Eg and λ are the band gap energy (eV) and wavelength of adsorption edge (nm), respectively. The narrower band gap and stronger visible-light response of the samples can be ascribed to the effect of the nitrogen doping [3,31]. Among the two biphasic samples, the band gap energies were increased with increasing brookite content. The reason may be due to the band gap of brookite is larger than anatase [18].

**Figure 4.** The UV-vis diffuse reflectance spectra of NA-145, NA-165, NA-185, HA-165, and P25.

#### *2.2. Photocatalytic Activity*

The visible light photocatalytic activities of as-prepared samples were tested by photodegradation of MO (10 mg/L). For comparison, we have also studied the photocatalytic activities of P25 and the two samples prepared with the similar sol-gel method but heated at 165 ◦C in a muffle furnace with air (MF-165) or a resistance-heated tube furnace with a flow of nitrogen (TF-165). Figure 5a shows the removal rates of MO for NA-185, NA-165, NA-145, HA-165, P25, MF-165, and TF-165 are 95%, 92%, 83%, 53%, 49%, 27%, and 31%, respectively. The dark reaction adsorption rates of samples are all less than 8%, which implied that the removal of MO is mainly attributed to photocatalytic degradation rather than adsorption. Figure S4 shows the nitrogen-doped anatase/brookite biphasic samples of NA-185 (94%) and NA-165 (91%) with similar photocatalytic degradation rate of MO, which are higher than the other samples. The apparent first-order rate constant K (min−1) for NA-165 (0.021) is close to that of NA-185 (0.023), which is about 1.4, 3.5, 4.2, 9.6, and 7.5 times higher than those of NA-145, HA-165, P25, MF-165, and TF-165, respectively (Figure 5b and Table 1). Thus, 165 ◦C was chosen as a reasonable heating treatment temperature.

**Figure 5.** (**a**) Removal curves of methyl orange (MO). Error bars represent the standard deviation from three measurements; (**b**) Apparent first-order kinetics plot for the photocatalytic degradation of MO over different samples.

Table 2 summarizes the preparation methods, surface area, and visible-light photocatalytic activity of nitrogen-doped anatase/brookite biphasic TiO2 reported in recent years. The 4.2-fold visible light activity enhancement as compared to P25 suggests that NA-165 is a potential highly efficient photocatalyst. By contrast, if the aged gel was heated in amuffle furnace with air or obtained in a resistance-heated tube furnace with a flow of nitrogen at the same temperature (165 ◦C), respectively, only amorphous samples were obtained in spite of large surface area (Figure S5 and Table 1). Meanwhile, they were much less active compared to NA-165. This implies that the crystallinity may play a more important role. Moreover, the presence of brookite in the mixture can reduce the recombination of hole–electron pairs. The band gap was also widened with increasing brookite content [34]. This is why NA-165 (3.03 eV, 6% brookite content) and NA-185 (3.05 eV, 10% brookite content) exhibited similar MO photocatalytic degradation activity.

The photocatalytic stability of NA-165 was tested by cycling experiments. For each cycling run, NA-165 was separated by centrifugation, and dried at 90 ◦C. As shown in Figure 6, there was no significant decrease of photocatalytic degradation rate after three cycling runs. This result suggested that NA-165 was a stable photocatalyst for organic dye degradation under visible light.


**Table 2.** Comparison of nitrogen-doped anatase/brookite biphasic TiO2 prepared by various methods.

#### *2.3. Possible Reasons for the Enhancement of the Visible-Light Performance*

It is interesting to evoke some reasons why NA-165 has high visible-light performance, though mechanism of the enhancement is still far from understood. The first explanation is that the absorption edge of NA-165 shifts to the visible-light range, and then they possess narrower band gap, and have definite absorptions in the visible region due to the presence of nitrogen-doping, which has been confirmed by UV-vis diffuse reflectance spectra and XPS study. Secondly, the anatase/brookite biphasic nanoparticles are aggregated closely, as shown in Figures S1 and S2. The intimate contact can facilitate inter particle charge transfer from brookite to anatase and reduce the recombination of electron–hole pairs. Thirdly, the large surface area can provide more active sites and improve the diffusion and migration of MO in the process of photodegradation [36]. Furthermore, the competitive diffusion of the H2O and dye molecules, dye molecule structure, and photocatalytic degradation route are also the factors influencingthe photocatalytic process [37]. The photodegradation of MO under visible light was mainly driven by the active species O2 *•−*, h+, and *•*OH [38].

#### **3. Materials and Methods**

#### *3.1. Synthesis*

Titanium tetraisopropoxide (TTIP, ≥97%, Sigma-Aldrich, St. Louis, MO, USA) was of chemical grade. Acetone (≥99.5%, Tianjin Fengchuan Chemical Reagent Technologies Co., Ltd., Tianjin, China), HNO3 (65%, Xilong Scientific Co., Ltd., Shantou, China), and acetylacetone (≥98%, Tianjin Fengchuan Chemical Reagent Technologies Co., Ltd., Tianjin, China) were of analytical grade. All chemicals were used without further purification.

Nitrogen-doped anatase/brookite biphasic nanoparticles were prepared with a sol-gel process modified from a sol-gel combined solvothermal route [24]. Titanium tetraisopropoxide, acetone, HNO3 and acetylacetone with the volume ratio of 6.5:20:0.11:0.54 were mixed in a glass beaker. A mixture solution of deionized water and acetone (volume ratio of 1.2:7.5) was then added dropwise with vigorous stirring until reaching the gelling point. The gels were placed into a quartz-lined stainless-steel autoclave after being aged for 24 h at room temperature. Then, the temperature of the autoclave was increased to and held at 145, 165, or 185 ◦C for 4 h under nitrogen atmosphere after flushing the autoclave with nitrogen gas. The initial and final pressures were under atmospheric and autogenous pressure, respectively. After the heat treatment ended, the pressure was released quickly to remove the solvent vapour. The obtained material was cooled down to room temperature by nitrogen purging before being washed with deionized water and dried in vacuum at 90 ◦C for 4 h. According to the heat treatment temperature, the samples were denoted as NA-145, NA-165, and NA-185, respectively. If HNO3 was replaced by the same volume of HCl in the process of preparation, then the sample was denoted as HA-165.

#### *3.2. Photocatalytic Activity*

In each experiment, 50 mg photocatalysts and 50 mL of MO solution (10 mg/L) was placed in a glass vessel with a cooling water jacket and quartz cover. The suspensions were stirred in the dark for 30 min to reach the adsorption–desorption equilibrium. Then, the system was exposed for 120 min under visible light irradiation provided by a 500 W Xe lamp with a 420 nm cut off filter. At certain time intervals, 3 mL of the suspensions was collected and centrifuged (10,000 rpm, 20 min) to remove the photocatalysts. The separated solution was analysed and the maximum absorption was recorded at 464 nm by a spectrophotometer (Shimadzu UV-2600, Kyoto, Japan).

The removal rate of MO was calculated using Equation (2):

$$\text{iremoval rate} = \text{C} / \text{C}\_{0\prime} \tag{2}$$

where C and C0 are the initial and instantaneous absorbance of MO at 464 nm.

The photocatalytic degradation rates of MO and the first-order rate constant K (min−1) were calculated, respectively, using the Equations (3) and (4):

$$\text{photocatalitytic degradation rate} = \text{C} / \text{C} \text{e},\tag{3}$$

$$\ln\left(\text{Ce}/\text{C}\right) = \text{Kt},\tag{4}$$

where Ce and C are the adsorption–desorption equilibrium absorbance and instantaneous absorbance of MO at 464 nm, respectively. t is the irradiation time.

#### *3.3. Characterizations*

X-ray powder diffraction (XRD, Rigaku Co., Tokyo, Japan) analysis was conducted on a D/max-3B spectrometer with Cu Kα radiation at a range from 10◦ to 90◦ (2θ). Brunauer–Emmett–Teller (BET) surface area, pore volume, and pore size were measured by nitrogen adsorption/desorption using a Micromeritics Tristar II Surface Area and Porosity Analyzer (Micromeritics, Norcross, GA, USA). Transmission electron microscopy (TEM) was conducted on a Hitachi H-800 instrument (Japan Electron Optics Laboratory Co., Ltd., Tokyo, Japan). Scanning electron microscopy (SEM) images were taken on a FEIQuanta200FEG microscope (FEI, Hillsboro, OR, USA). X-ray photoelectron spectroscopy (XPS) was recorded using a Thermo Fisher Scientific K-Alpha<sup>+</sup> XPS system with Al Kα radiation and adventitious C1s peak (284.8 eV) calibration (Thermo Fisher Scientific Inc., Waltham, MA, USA). UV-Vis diffuse reflectance spectra were measured on a UV-2600 photometer (Shimadzu Corp., Kyoto, Japan).

#### **4. Conclusions**

The nitrogen-doped anatase/brookite biphasic nanoparticles with large surface area (240 m2/g) were successfully prepared during the sol-gel preparation at low temperature (165 ◦C, 4 h). The sample obtained without subsequent annealing at elevated temperature, which exhibited enhanced visible-light photocatalytic efficiency for the degradation of MO with 4.2-, 9.6-, and 7.5-fold visible light activities as compared to P25, MF-165, and TF-165, respectively. This was attributed to nitrogen doping, mixed crystalline phase, and high surface area. The recycling experiments suggested that NA-165 was a stable visible-light photocatalyst. The sample and low-temperature synthetic method developed in this work may provide a new pathway to prepare the stable photocatalyst for the degradation of organic dyes under visible light.

**Supplementary Materials:** The following are available online at www.mdpi.com/2073-4344/7/12/376/s1, Figure S1: Scanning electron microscopy (SEM) images of NA-185 (a), NA-165 (b), NA-145 (c) and HA-145 (d); Figure S2: Transmission electron microscopy (TEM) Transmission electron microscopy (TEM); Figure S3: Nitrogen adsorption–desorption isotherms for the prepared samples; Figure S4: Photocatalytic degradation of MO under visible light over different samples; Figure S5: XRD patterns of MF-165 and TF-165; Table S1: The Brunauer–Emmett–Teller (BET)analysis data of samples.

**Acknowledgments:** The work was supported by National Natural Science Foundation of China (Project 21403190, 21573193, 21367024, 21464016 and 21263027). The authors also thank Program for Innovation Team of Yunnan Province and Innovative Research Team (in Science and Technology) in the Universities of Yunnan Province, Key Laboratory of Advanced Materials for Wastewater Treatment of Kunming, the Key project from the Yunnan Educational Committee (Project ZD2012003), Yunnan Provincial Natural Science Foundation (Project 2015FB106) and Yunnan Applied Basic Research Projects (Project 2016FA002) for financial support.

**Author Contributions:** Jiaqiang Wang and Liang Jiang conceived and designed the experiments; Liang Jiang, Yizhou Li, and Haiyan Yang performed the experiments; Yepeng Yang, Xiang Long and Jun Liu analyzed the data; Liang Jiang, Zhiying Yan, and Jiao He wrote the paper; Jiaqiang Wang and Zhiying Yan modified the paper.

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

#### **References**


© 2017 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 (http://creativecommons.org/licenses/by/4.0/).
