Efficient removal of methylene blue dye by a hybrid adsorption–photocatalysis process using reduced graphene oxidetitanate nanotube composites for water reuse

Efficient removal of methylene blue dye by a hybrid adsorption–photocatalysis process using reduced graphene oxide/titanate nanotube composites for water reuse Accepted Manuscript Title Efficient remo[.]

Accepted Manuscript

Title: Efficient removal of methylene blue dye by a hybrid

adsorption–photocatalysis process using reduced graphene

oxide/titanate nanotube composites for water reuse

Authors: Chi Hieu Nguyen, Ruey-Shin Juang

DOI: https://doi.org/10.1016/j.jiec.2019.03.054

To appear in:

Received date: 8 October 2018

Revised date: 24 December 2018

Accepted date: 28 March 2019

Please cite this article as: Nguyen CH, Juang R-Shin, Efficient removal of methyleneblue dye by a hybrid adsorption–photocatalysis process using reduced graphene

oxide/titanate nanotube composites for water reuse, Journal of Industrial and Engineering Chemistry (2019), https://doi.org/10.1016/j.jiec.2019.03.054

This is a PDF file of an unedited manuscript that has been accepted for publication

As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain

Revisions submitted to Journal of Industrial and Engineering Chemistry

(Research Article)

(JIEC_2018_392_R1)

Efficient removal of methylene blue dye by a hybrid adsorption–photocatalysis process using reduced graphene oxide/titanate nanotube composites for water reuse

Chi Hieu Nguyen 1,2 , Ruey-Shin Juang 1,3,4 *

1Department of Chemical and Materials Engineering, Chang Gung University, Guishan, Taoyuan 33302, Taiwan

2Institute of Environmental Science, Engineering and Management, Industrial University of Ho Chi Minh City,

Ho Chi Minh City, Vietnam

3Division of Nephrology, Department of Internal Medicine, Chang Gung Memorial Hospital, Linkou, Taiwan

4Department of Safety, Health and Environmental Engineering, Ming Chi University of Technology, Taishan, New Taipei City 24301, Taiwan

*Corresponding author: Professor Ruey-Shin Juang

E-mail: rsjuang@mail.cgu.edu.tw

Address: Department of Chemical and Materials Engineering, Chang Gung University, 259 Wenhua First Road,

Guishan, Taoyuan 33302, Taiwan

Graphical Abstract

This scheme proposes possible photocatalytic mechanism over rGO/TNT composites TiO2 nanotubes are dispersed well on the surface of rGO sheets Tubular structure of TiO2 improves adsorption ability for dye Under photon irradiation, electrons (e-) are excited from the valence band (VB) to conduction band (CB), leading to the formation of holes (h+) in the VB The holes with strong oxidation ability scavenge water molecules on the surface

of TNTs and generate highly reactive hydroxyl radical (·OH) Also, they attack and convert dye molecules to products The photoinduced electrons (e-) reduce the absorbed oxygen molecule to produce superoxide radicals

by-ACCEPTED MANUSCRIPT

(·O2), which may directly oxidize organic pollutant into small molecules; and, part of ·O2 can react with H and generate H2O2, which is further excited by electrons and changed into ·OH radicals These active radicals attack dye molecules and degrade them into the intermediates and final products such as CO2 and H2O Normally, the electron-hole pairs for TNTs are ready to be recombined, resulting in poor photoactivity Herein, the photoinduced electrons can be trapped by rGO, leading to electron-hole separation Therefore, the adsorption and photocatalytic ability of rGO/TNT composites is enhanced

Highlights

 Reduced graphene oxide/titanate nanotube (rGO/TNT) composites are hydrothermally prepared as novel photocatalysts

 Unlike methyl orange (MO), rGO/TNT composite achieves higher decolorization/mineralization of

 Unlike MO, removal of MB using rGO/TNT composites is due to the synergy of adsorption and photocatalysis

 GO content in rGO/TNT composite has a great effect on adsorption and photocatalytic activity for cationic dye

 Incorporating rGO in rGO/TNT composite will extend its photocatalytic ability to the visible light region

Abstract

ACCEPTED MANUSCRIPT

In this study, reduced graphene oxide/titanate nanotube (rGO/TNT) composites were fabricated by a one-step facile hydrothermal process The fine structures and physicochemical properties of the prepared materials were comprehensively determined The incorporation of a small amount of rGO into the composites led to the enhancement of the absorption intensity of visible light and the separation of photogenerated charged carriers The rGO/TNT composite with the optimal amount of rGO of 3% by weight showed the highest photocatalytic activity for both methylene blue (MB) and methyl orange Moreover, 3 wt.% rGO/TNT exhibited higher adsorption capability and photocatalytic activity for MB, a cationic dye, than TNTs and commercial TiO2 P25 The maximum amount of MB adsorbed on 3 wt.% rGO/TNT was 26.3 mg/g at 25°C, and the adsorption rapidly reached equilibrium after 40 min of contact time Approximately 100% decolorization and 77.4% mineralization over 3% rGO/TNT composite were achieved under 100 W UV irradiation for 1 h, whereas 95.0% decolorization and 78.8% mineralization were achieved under visible light irradiation for 3 h The degradation pathways of MB over P25, TNTs, and 3 wt.% rGO/TNT composite were finally proposed and compared

Keywords: Adsorption; Photocatalysis; Reduced graphene oxide; Titanate nanotubes; Cationic dye; Degradation

pathways

ACCEPTED MANUSCRIPT

1 Introduction

Numerous environmental issues caused by water source pollution have recently attracted considerable interest from scientists and engineers Textile industries are well-known manufacturers that use and discharge large amounts of water resources The total annual dye production exceeds 700,000 t worldwide, and approximately 20% of this production is directly released through effluents from manufacturing processes [1,2] One of the major problems of textile effluents is the abundant class of colored organic compounds in water, which can impede light transmission, consequently damaging the photosynthetic process of aquatic flora The majority of synthetic dyes are toxic, no biodegradable, or likely decomposed to form harmful by-products under anoxic conditions From the viewpoint of environmental sustainability and water reuse, the efficient removal of dyes from effluents is highly desired prior to discharging into receiving waters [3,4]

Traditional physicochemical methods, such as chemical precipitation, electrocoagulation, membrane filtration, and adsorption, are currently applied to remove the color of textile effluents [5–11] However, effluents are not mineralized but merely transferred from one kind of pollution to another Thus, further treatments are required to remediate these concentrated pollutants In recent years, heterogeneous photocatalysis has been recognized as an efficient process for removing organic pollutants thoroughly [12–14] Most importantly, the application of this process to textile effluent treatment might result in the decolorization and complete mineralization of dyes [15,16]

Of many heterogeneous photocatalysts used, titanium dioxide (TiO2) has been proven to be promising because of its high chemical reactivity and stability, nontoxicity, and cost-effectiveness [14,17,18] However, some drawbacks of TiO2 photocatalysis, including the high bandgap energy (3.2 eV in the anatase phase) and easy recombination of photoinduced electron–hole pairs, always limit its feasibility for practical application [14,19] Moreover, individual catalyst is appropriate only for degrading dyes with low concentrations but fails to treat high-strength solutions [20] Thus, modifications of the existing materials have been explored to improve the photocatalytic capability and extend the application range An efficient method is to dope transitional metals (e.g.,

Cu2+ and Fe+3), noble metals (e.g., Pt, Pd, and Ag), or nonmetallic elements (e.g., S and N) on photocatalysts [21–25] The main objectives of doping are to reduce the bandgap of TiO2 by transposing the absorption band from the

UV light region to the visible light region and to improve the separation of electron–hole pairs In addition to doping of TiO2, adjusting its morphology, combining it with other metal oxides (e.g., ZnO and Al2O3) [26–28], and fixing it on mesoporous materials with high surface area, such as zeolites and activated carbon, are some

ACCEPTED MANUSCRIPT

alternatives to increase TiO2 photoactivity [29] Each type of modification possesses relevant mechanisms to improve the optoelectronic properties of TiO2

Recent studies have indicated that titanate nanotubes (TNTs) exhibit good photocatalytic activity because of its higher surface area than commercial TiO2 P25 [30–33] However, TNTs possess similar drawbacks to TiO2 By contrast, graphene contains six C atoms, is created by sp2 hybridization, and has a 2-D planar honeycomb structure with a thickness of a single atomic layer [34–36] Fabricating graphene to form inorganic composites can thus produce attractive materials because of its unique electronic properties [4,36–38] Currently, the Hummers’ method is most commonly used to fabricate graphene sheets, and the final product is a single graphene oxide (GO) that contains many hydrophilic groups Exfoliated GO exhibits poor electronic conductivity; therefore, the reduction of GO to form reduced graphene oxide (rGO) is required to store and shuttle electrons [39–41] Previous studies have shown that GO/TiO2 composites exhibit exothermic mixing [42,43] Moreover, TiO2 can retard the mass loss of rGO at high temperatures because of the chemical reactions between rGO and TiO2 through Ti–O–C bonds The beneficial effect of combining rGO with TiO2 for the photodegradation of pollutants, particularly azo dyes, has been proven [42,44] The heterojunction between rGO and TNTs is expected to enhance photocatalysis through the synergy of both materials Such an improved photocatalytic activity is caused by the combination of absorption and separation of photoinduced charged carriers because rGO acts as electron reservoir preventing recombination; meanwhile, holes that remain in TNTs can initiate the oxidation [4,45] Moreover, the formation of Ti–O–C bonds and its combination with carbon-based materials extend visible light harvesting [8,14,29]

Although a few studies reported the enhanced photocatalytic activity of rGO/TNT composites for some dyes [46–48], to the best of our knowledge, in-depth insight into the synergistic effects of adsorption and photocatalysis,

as well as the roles of rGO/TNT composites, on the improvement of degradation in the visible light region is lacking For example, anionic dyes were rarely investigated, although rGO/TNT composites were observed to have the potential to treat dye-bearing effluents However, this finding is insufficiently convincing due to the key roles of the chemical structures of pollutants on adsorption capability and photocatalytic activity This work aimed

to synthesize some rGO/TNT composites through one-step hydrothermal method Methylene blue (MB, a cationic dye) was selected as the model dye to assess the hybridity of adsorption capability and photocatalytic activity of composites Methyl orange (MO, an anionic dye) was chosen as the control dye because it has comparable molar mass to MB but different chemical characteristics In particular, the currently prepared composites exhibited different adsorption abilities for MO and MB The transformation products formed during MB photocatalysis over P25, TNTs, and rGO/TNT composites were finally analyzed and the pathways were proposed and compared

ACCEPTED MANUSCRIPT

2 Materials and methods

2.1 Materials

TiO2 P25, whose surface area and average particle size were 50 m2/g and 21 nm, respectively, was provided

by Degussa Co (Germany) Graphite powder (99%), NaOH (99.8%), H2SO4 (98%), H3PO4 (85%), and ethanol (99%) were obtained from Sigma Aldrich Co (St Louis, MI, USA) Potassium permanganate (98%) and H2O2

(30 vol%) were supplied by Showa Chemical Co (Okayama, Japan) The dyes MB and MO (purity > 98%) were obtained from Alfa Aesar Co (Heysham, England) All chemicals were used as received Deionized (DI) water produced by the Millipore Milli-Q system (Darmstadt, Germany) was employed throughout this study

2.2 Preparation of rGO/TNT composites

2.2.1 Preparation of GO sheets

GO was prepared by an improved Hummers’ method [49,50] Briefly, 400 mL of concentrated H2SO4 and

H3PO4 solutions was mixed in an ice bath under continuous agitation at less than 15°C, to which 3.0 g of graphite powder and 18.0 g of KMnO4 were added Afterward, the resulting solution was heated to and kept at 50°C and magnetically agitated for 12 h When the reaction was completed, the suspension was poured slowly into a flask containing 400 g of iced DI water Then, the solution was cooled to 25°C, to which 30 vol% H2O2 (6 mL) was added to remove excess KMnO4 The suspension turned brilliant yellow immediately The final mixture was centrifuged at 10,000 rpm and sequentially washed with 30 vol% HCl, DI water, and ethanol until the washing solution became neutral Finally, the recovered precipitate was dried in a vacuum oven at 60°C for 24 h

2.2.2 Preparation of rGO/TNT composites

The rGO/TNT composite was hydrothermally synthesized as stated earlier [48] but with some modifications Briefly, 3.0 g of P25 and a certain amount of GO (i.e., 1%, 2%, 3%, and 5% by weight) were added to 90 mL of NaOH (10 M) under vigorous agitation for 1.5 h and sonicated for 1 h to make it homogeneous The resulting solution was shifted into a stainless-steel autoclave (Teflon-lined), which was kept in an oven at 135°C for 24 h Afterward the autoclave was cooled to 25°C The obtained solid was washed with 0.1 M HNO3 until the pH of the washing solution became 1.5 Then, the mixture was agitated continuously for 24 h to ensure ion exchange The suspended solids were recovered by centrifugation at 10,000 rpm and rinsed with DI water several times until the washing solution became neutral The final products were dried in a vacuum oven at 80°C for 24 h and calcined at

ACCEPTED MANUSCRIPT

300°C for 1 h in a tube furnace under Ar atmosphere to refine the crystal form The solid catalysts were denoted as 1%, 2%, 3%, and 5% rGO/TNT Pure TNTs were prepared by the same procedures used previously but without the addition of GO

2.3 Characterization of the synthesized catalysts

The morphology of the prepared catalysts was observed under a field emission scanning electron microscope (FE-SEM; SU8220, Hitachi, Japan) and a transmission electron microscope (TEM; JEOL JEM1230, Tokyo, Japan) The crystalline structure was identified by an X-ray powder diffractometer (XRD; Bruker D2 PHASER, Germany) with CuKα radiation source Functional groups in the prepared samples were identified by a Fourier transform infrared spectroscope (FTIR; Bruker Tensor 27 IR, Germany) In addition, the surface composition of the catalysts was determined by an X-ray photoelectron spectroscope (XPS), and the XPS spectra were recorded

by a Fison VG ESCA210 spectrometer with MgKα radiation

Optical properties of as-prepared catalysts were analyzed according to the UV–visible diffuse reflectance spectra recorded by a Jasco V650 spectrophotometer (Tokyo, Japan) equipped with an ISV-722 integrating sphere The photoluminescence (PL) spectra were recorded by a F-4500 fluorescence spectrophotometer (Hitachi, Japan)

at 25°C Photoelectrochemical tests of the synthesized catalysts were conducted on a CHI 660D workstation with

a three-electrode mode, following the procedures described earlier [51] The BET surface area of the catalysts was determined from N2 adsorption–desorption isotherms at −196°C on an accelerated surface area and porosimetry analyzer (Micromeritics ASAP2020, USA) The sample was degassed at 300°C for 3 h before measurements

2.4 Adsorption of dyes on the synthesized catalysts

Batch adsorption of dyes was conducted in the dark to prevent photocatalysis and photolysis Briefly, 0.1 g of catalysts was mixed with 100 mL of dye solution (10–60 mg/L) under magnetic stirring at 25±1°C The pH was kept constant at 6.8±0.2 Samples (3 mL) were taken at preset time intervals and filtered through a 0.22 µm syringe filter (Chromophil® Xtra, Germany) to remove catalyst particles The concentrations of dyes in the sample were determined by a UV–visible spectrophotometer at the wavelength of 665 nm for MB and 464 nm for

MO The amount of dye adsorbed on the catalysts at equilibrium, qe (mg/g), is calculated as follows:

2.5 Photodegradation experiments

The photocatalytic activity of TNTs and rGO/TNT composites was assessed via the oxidation of dyes under

UV and visible light irradiation A mercury lamp (Sen Lights Co., Japan) without and with a UV cutoff filter, which had a light intensity of 6.5 and 0.009 mW/cm2, was used as the source of UV and visible light, respectively For comparison, P25 was adopted as a reference catalyst Photodegradation experiments were conducted in a batch reactor as described previously [54] Briefly, the lamp was placed in a quartz tube, which was immersed in the reactor The initial solution pH was adjusted using 0.1 M NaOH or HNO3 An aliquot of the catalysts was added to 20 mg/L of dye solution (1.0 L) Prior to photocatalysis, the suspension was agitated in the dark for 1 h to ensure adsorption equilibrium Samples (3 mL) were taken at preset time intervals, and the concentrations of dyes were measured spectrophotometrically to evaluate the decolorization efficiency and degradation kinetics In addition, the total organic carbon (TOC) was measured by a TOC Torch analyzer (Teledyne Tekmar, Mason, OH, USA) to determine the mineralization efficiency

Reaction intermediates formed during MB photocatalysis over P25, TNTs, and rGO-TNT composite were identified by a high-resolution UPLC®-MS/MS system (Waters Xevo TQ-XS, Milford, MA, USA) equipped with

an ACQUITY UPLC® BEH C18 column (2.1×50 mm, particle size of 1.7 µm) Mass spectrometry was conducted with electrospray ionization in positive ion mode (ESI+) The assigned parameters, as well as the operating procedures and conditions, were the same as those described previously [51]

The reusability of 3% rGO/TNT composite was typically studied for five repeated cycles The catalysts used

in each cycle were washed several times with DI water Then, 30 mL of 30 vol% H2O2 was added to the catalyst solution (3 g/L), which was irradiated with UV for 2 h and rinsed several times with DI water The catalysts were regenerated by drying and calcining in an oven at 300°C for 1 h before the beginning of the next cycle

3 Results and discussion

3.1 Characterization of as-prepared catalysts

3.1.1 Structural and textural characterization

Fig 1 depicts the XRD patterns of as-prepared catalysts For GO, the diffraction peaks that appeared at 2θ

values of 11.3° and 42.5° (inset) were attributed to the (002) and (001) planes of GO, respectively, which indicate the existence of O-containing groups, such as hydroxyl, carbonyl, and epoxy Pristine rGO was similarly fabricated by hydrothermal treatment of GO in the absence of P25 In the rGO spectrum, the main diffraction peak

ACCEPTED MANUSCRIPT

at 2θ of 11.3° disappeared and the broad peak of rGO was observed at 25.1° [39], indicating that many containing groups were removed This finding shows that GO was successfully reduced to rGO by the hydrothermal method The XRD patterns of rGO/TNT composites present diffraction peaks at 2θ = 25.4°, 37.6°, 48.1°, 53.5°, 55.1°, 62.6°, 68.7°, 70.2°, and 75° assigned to the (101), (004), (200), (105), (211), (204), (116), (220), and (215) crystal planes of the anatase phase of TiO2, respectively (JCPDS Nos 21-1272 and 21-1276) The original rutile peaks in P25 disappeared completely for hydrothermally prepared catalysts Typical diffraction

O-peaks of GO (Fig 1, inset) or carbon species cannot be detected, even for 5% GO Moreover, rGO with

diffraction peak at 25.1°, which was formed after the hydrothermal process, could be hidden by a strong anatase peak at 25.4° [29,46,55] This phenomenon is caused by the relatively small amount and weak intensity of rGO Hence, the hydrothermal process successfully converted TiO2 nanoparticles into TNTs and simultaneously reduced GO to rGO The heights of the peaks of rGO/TNT composites were lower than those of TNTs, indicating that the crystalline structure of TNTs was less developed than that of rGO/TNT composites The average crystal

size of rGO/TNT composites and TNTs was estimated by the Debye–Scherrer formula [51] As presented in Table

1, the crystal size of rGO/TNT composites decreased because of the addition of rGO compared with that of TNTs Fig 2a shows the N2 adsorption–desorption isotherms of the prepared catalysts, which were classified as type

IV on the basis of the IUPAC classification Such an isotherm is commonly observed for mesoporous materials, and its characteristic feature is the hysteresis loop, which is related to the occurrence of pore condensation [56] The limiting adsorption regime occurs over a range of high P/P0 values, but the hysteresis loop in the isotherm of as-prepared catalysts is detected at P/P0 = 1, indicating the presence of macropores The shape of the adsorption branches is rather similar to type II, which is typical for nonporous or macroporous adsorbents, where monolayer–multilayer adsorption occurs This reveals that mesopores and macropores were created due to the agglomeration

of TNTs and rGO sheets, whereas the smaller pores (<10 nm) were assigned to the internal pores of TNTs

The textural properties of as-synthesized catalysts and P25 are summarized in Table 1 Notably, TNTs and

rGO/TNT composites had a larger surface area than P25 The surface area of TNTs (168.7 m2/g) was smaller than that of rGO/TNT composites The surface area of rGO/TNT composites reached a maximum at 3 wt.% rGO because rGO increases the surface area by dispersing TNTs on rGO sheets and preventing the coagulation of TNTs However, excess GO leads to agglomeration, which reduces the surface area of rGO/TNT composites The

pore size distributions derived from the adsorption branches of as-prepared materials are also shown in Fig 2b

For all samples, the shapes of the pore size distributions exhibited no significant differences The pore size of all samples concentrated mainly at diameters of 2–20 nm These components correspond to the internal pore volume

ACCEPTED MANUSCRIPT

distribution of the nanotubes and the large pores formed between nanotubes All rGO/TNT composites appeared

more macroporous than TNTs (Fig 2b), which was probably caused by the presence of rGO sheets

Fig 3 illustrates the FTIR spectra of GO, TNTs, and rGO/TNT composites Both rGO/TNT and TNTs

exhibited similar spectra with three different peaks at 3,460, 1,630, and 500.1 cm−1 The peak at 500.1 cm−1 was caused by the crystal lattice vibration of TiO6 octahedra, which was strongly affected by the incorporation of ions (via ion exchange) into TNTs or by the alteration of the tubular structure, including the diameter of nanotubes The peak at 1,630 cm−1 resulted from the hydroxyl group bending mode of water molecules adsorbed on the surface of TNTs [46] The intensity of the peak at 1,630 cm−1 for rGO/TNT composites was stronger than that for TNTs, which can be attributed to the skeletal vibration of graphene sheets The appearance of a broad band at 3,460 cm−1corresponded to O–H stretching of the surface hydroxyl groups on TiO2 The hydroxyl groups on the catalyst surface that were created through ion exchange can improve the adsorption capability of the catalyst for organic species Notably, all peaks of GO disappeared from the FTIR spectra of rGO/TNT composites probably because the amount of GO in the rGO/TNT composite is lower than the limit detected by the instrument or hidden by the peaks of TNTs

The Raman spectra of GO, rGO, and rGO/TNT composites are presented in Fig 4 For all rGO-TNT

composites, the Raman peaks at 150, 198, 396, 507, and 632 cm−1 attributed to the Eg(1), Eg(2), B1g, A1g + B1g, and

Eg(3) modes, respectively, were assigned to the anatase phase of TiO2 [57,58] No Raman peaks due to the rutile or brookite phase were detected, which is consistent with the XRD results New peaks of the D and G bands located

at 1,346 and 1,588 cm−1, respectively, were detected for the rGO/TNT composites, denoting that the composite contained TiO2 and rGO The intensity ratio (ID/IG) of rGO alone was similar to that of rGO in rGO/TNT composites Remarkably, the ID/IG value of GO and 5% rGO/TNT was 0.96 and 1.08, respectively The ID/IG ratio

of rGO increased slightly, indicating that the number of O-containing groups and the average size of sp2 domains

on rGO decreased [29,59] The IG/ITNT ratio of an integrated area of tangential G mode relative to the most intense

Eg anatase mode IG/ITNT of rGO/TNT composites was used to quantify the variation of rGO loading on TNTs [57] Evidently, the increase in rGO content in the composite corresponded to the increase in IG/ITNT ratio In detail, the

IG/ITNT ratio increased by 4.2-fold, whereas the rGO loading increased from 1% to 5%

The XPS results showed the surface composition of rGO/TNT composites (Fig 5), which contained three main elements, i.e., Ti, O, and C with photoelectron peaks of Ti2p, O1s, and C1s, respectively Fig 5b shows the

spectra of Ti2p Two peaks at 458.3 and 464.0 eV corresponded to Ti2p3/2 and Ti2p1/2 of Ti(IV), respectively, which confirmed the formation of TiO2 The XPS spectra of C1s for the composite (Fig 5d) depicts one strong

ACCEPTED MANUSCRIPT

symmetric peak at 284.48 eV, which can be ascribed to C–C bonds Compared with the XPS spectra of C1s in GO, the C–O and C=O peaks at high binding energies of GO disappeared, indicating a considerable reduction in the

number of C–O species in rGO (Fig S1 in the Supplementary material) [46,60,61] This finding again proved

the successful transformation of GO to rGO by the hydrothermal process By contrast, the peaks of Ti2p, O1s, and C1s for 3% rGO/TNT shifted compared with TNTs, indicating the addition of GO

3.1.2 Morphology

Fig 6 depicts the FE-SEM and TEM images of as-prepared catalysts The FE-SEM image of GO before the

attachment of TNTs (Fig 5a) shows that its structure is similar to a thin sheet Hence, depositing TiO2 on GO along its wrinkles and edges was easier [59,62] TNTs with small tubular structure nearly covered the rGO sheets,

as shown in the TEM image of 3% rGO/TNT (Fig 6d) The length of the tubular structure of TNTs varied from

several dozens to hundreds of nanometers, and its internal and external diameters varied from 5.4 nm to 6.5 nm and from 6.9 nm to 9.0 nm, respectively Such a crystallite size is consistent with that observed from the previously presented XRD analysis

Previous studies reported that the formation of a 1-D structure involved the dissolution of the 3-D structure of TiO2 by breaking Ti–O–Ti bonds and rearranging the TiO6 octahedra into 2-D nanosheets [55,63,64] These sheets could scroll or fold into a nanotubular morphology Herein, rGO served as the matrix for the deposition of TNTs Thus, TNTs cover the majority of the surface area of rGO due to the high TNT loading on rGO/TNT composite

3.1.3 Optical properties

The UV–visible absorption spectra of P25, TNTs, and rGO/TNT composites are depicted in Fig 7a The light

absorbance of P25 and TNTs did not show a significant difference However, the absorption intensity of rGO/TNT composites was higher than that of TNTs and P25 over the entire visible light region The incorporation of rGO into TNTs had a considerable effect on the optical properties of TNTs The absorption of visible light by rGO/TNT composites increased with the increase in the amount of rGO The improved activation of rGO/TNT composites

by visible light was responsible for the interaction between rGO and TiO2

In contrast to that of TNTs and P25, the absorption edge of rGO/TNT composite showed a slight redshift, denoting a decrease in bandgap energy The bandgap energies of all catalysts were determined by the transformed Kubelka–Munk function as stated previously [51] The bandgaps of P25 and TNTs were equivalent, but those of

ACCEPTED MANUSCRIPT

rGO/TNT composites slightly narrowed because of the addition of rGO (Table 1) Under visible light irradiation,

the low bandgap energy creates more electron–hole pairs Consequently, the photocatalytic capability is enhanced

Fig 7 shows the transient photocurrents under UV irradiation All prepared catalysts exhibited rapid and

steady photocurrent responses The photocurrents of all rGO/TNT composites were enhanced compared to those

of TNTs and P25 In particular, the photocurrent of 3% rGO/TNT was the largest and increased by 4.1-fold in contrast to that of TNTs The photocurrent is generated by photoinduced electrons in the conduction band of the catalysts, confirming that the incorporation of rGO enhanced the mobility and separation of the electron–hole pairs [64–66]

3.2 Adsorption capability and photocatalytic activity

3.2.1 Adsorption capability of as-prepared catalysts

Preliminary tests showed that adsorption occurred rapidly within 10 min and equilibrium was reached within

40 min The adsorption of MO on rGO/TNT composites was negligibly small Under the conditions studied, only

1% to 4% of the initial MO concentration (10–60 mg/L) was adsorbed This is not the case for MB Fig 8a shows

the adsorption isotherms of MB on various catalysts at 25°C The adsorption capability of rGO/TNT composites for MB increased with the increase in the amount of GO up to 3% The enhanced adsorption was attributed to (1) the π–π conjugation between the aromatic regions of 2-D GO and MB and (2) the change in the morphology of TNTs resulting from the increase in the surface area after rGO hybridization [45] Based on the fitted parameters

(Table 2) of the Langmuir and Freundlich equations (Section S1 in the Supplementary material), the isotherms were better depicted by the Langmuir model (Fig 8b), which indicated monolayer coverage of MB on the surface

of rGO/TNT

To further analyze the adsorption behavior, the zeta potentials of the prepared catalysts were measured The

points of zero charge of P25, TNTs, and 3% rGO/TNT were at pH 6.9, 5.7, and 4.8, respectively (Fig S2) In

aqueous solution, TNTs tend to develop a negative charge because of the dissociation of titanic acid GO was

negatively charged in the pH range of 3.0–11.5 (Fig S2) Hence, the rGO/TNT composite was more negatively

charged than TNTs because of GO, considerably affecting the adsorption of charged molecules on the surface of TNTs Therefore, the affinity of cationic species (e.g., MB) was higher than that of anionic species (e.g., MO)

The adsorption capability of 3% rGO/TNT for MB increased with the increase in pH (Fig 9) because MB is a cationic dye and the point of zero charge of 3% rGO/TNT is pH 4.8 (Fig S2) Therefore, MB was readily

adsorbed on the catalysts with more negatively charged surface

ACCEPTED MANUSCRIPT

3.2.2 Photocatalytic activity of as-prepared catalysts

In the first series of tests, MB and MO were adsorbed on all samples for 60 min in the dark Only 8% of the

initial MB concentration was adsorbed on P25, which is less than those on TNTs and rGO/TNT composites (Fig

10) Moreover, 3% rGO/TNT exhibited the highest adsorption capability with 65% decolorization; for example,

decolorization was only 40% using 5% rGO/TNT However, the adsorption capability of all catalysts for MO was low (only 1% to 4% of the initial MO concentration) Blank tests (under UV irradiation without catalysts) showed that photolysis of MB and MO was negligible Under UV irradiation for 30 min, 100% of MB was degraded over 3% rGO/TNT Complete decolorization was achieved by all prepared catalysts, except for P25, after 60 min in the dark and 60 min UV irradiation After adsorption of 20 mg/L of MB on the catalysts, photocatalysis followed the

first-order kinetics (Section S2) with a kapp value of 0.048 min−1 (for 5% rGO/TNT), 0.054 min−1 (for P25), 0.067 min−1 (for TNTs), 0.10 min−1 (for 1% rGO/TNT), 0.11 min−1 (for 2% rGO/TNT), and 0.14 min−1 (for 3%

rGO/TNT) (Fig S3)

Meanwhile, the degradation of MO decreased sharply from 99.4% to 58.6% after 180 min UV irradiation when P25 was converted into TNTs, which can be partly attributed to the surface charge of TNTs Although TNTs

had a larger surface area than P25, the surface of TNTs was more negatively charged than that of P25 (Fig S2)

Notably, MO is anionic and the repulsive force between dye molecules and catalysts increases at pH 6.8 [67–69] Hence, the active species on the catalyst surface cannot easily attack dyes; as a result, the photocatalytic activity of TNTs was reduced After combining rGO with TNT, the photocatalytic activity of rGO/TNT composites was enhanced compared with that of TNTs because of the improved separation of charged carriers The highest photodegradation of MO was observed over 3% rGO/TNT with 99.2% decolorization and 64.5% mineralization

after 180 min UV irradiation (Fig 10b) However, all catalysts presented lower photocatalytic activity than P25

As shown in Fig S3, the degradation of MO also followed the first-order kinetics in the order of 0.025 min−1 (for P25) > 0.020 min−1 (for 3% rGO/TNT) > 0.013 min−1 (for 2% rGO/TNT) > 0.010 min−1 (for 1% rGO/TNT) > 0.008 min−1 (for 5% rGO/TNT) > 0.005 min−1 (for TNTs) This finding shows that the formation of TNTs with tubular morphology was inefficient for MO degradation compared with P25 Thus, we can conclude that the adsorption of pollutants on photocatalysts plays a key role in its catalytic performance

For comparison, the composite of GO and TNTs (GO-TNTs) was also adopted for the photodegradation of

MB and MO here (Fig S4) In contrast to TNTs, no apparent differences between photocatalysis of MB and MO

over TNTs were observed Remarkably, MB removal over 5% TNTs was higher than that over 3%

GO-ACCEPTED MANUSCRIPT

TNTs and TNTs, which is opposite to the case of rGO/TNT composites This finding can be possibly attributed to the increase in the amount of MB adsorbed on the surface of GO Meanwhile, the amount of MO adsorbed on the surface of GO was smaller [4,45] Hence, the change in the degradation efficiency was negligible Moreover, 3% rGO/TNT exhibited higher photocatalytic activity for both MB and MO than GO-TNTs These results confirmed that rGO/TNT composites were fabricated successfully and rGO played an important role in improving the separation of photogenerated electron–hole pairs

Comparing the photoactivity of the catalysts on the basis of the MB degradation data shown in Fig 10 is

difficult because MB concentrations left in the solution are different after adsorption Other tests were conducted

with UV/visible light irradiation at the beginning of the test As shown in Fig 11a, decolorization decreased over

rGO/TNT composites in the order of 5% < 1% < 2% < 3% GO, although all catalysts achieved more than 95% decolorization after 60 min UV irradiation At an initial MB concentration of 20 mg/L, the hybrid adsorption–photocatalysis process exhibited kapp values of 0.044 min−1 (for 5% rGO/TNT), 0.061 min−1 (for TNTs), 0.069 min−1 (for 1% rGO/TNT), 0.097 min−1 (for 2% rGO/TNT), and 0.11 min−1 (for 3% rGO/TNT) In contrast to that under UV irradiation, MB removal under visible light irradiation was inefficient The highest decolorization of 95% was obtained for 3% rGO/TNT after 180 min irradiation, followed by 2% rGO/TNT > 5% rGO/TNT > 1%

rGO/TNT > TNTs > P25 (Fig 11b) Notably, 5% rGO/TNT had a higher photoactivity under visible light

irradiation than TNTs and 1% rGO/TNT, although it presented a lower photoactivity under UV irradiation This finding can be attributed to the decrease in bandgap energy and increase in absorption intensity of visible light with the increase in the amount of rGO The kapp value obtained using 3% rGO/TNT was 2.5 and 8.5 times larger than that using P25 under UV and visible light irradiation, respectively

In fact, the decolorization of dyes was faster than its mineralization, as verified by the TOC data (Fig 12),

because the color of a dye is caused by chemical bonds and the associated chromophores and auxochromes Dye molecules come in contact with hydroxyl radicals and oxidize incompletely, leading to the production of many

intermediates In this case, the removal of these products took a long time As depicted in Fig 12, mineralization

was also in the order of 5% < 1% < 2% < 3% of rGO loading under UV irradiation and 1% < 5% < 2% < 3% under visible light irradiation P25 exhibited the lowest mineralization, i.e., only 49.7% after 60 min UV irradiation and 12.3% after180 min visible light irradiation

Similar to decolorization, the highest mineralization was 77.4% (60-min UV irradiation) and 78.8% (180-min visible light irradiation) over 3% rGO/TNT This finding indicates some intermediates to be present in the solution during MB degradation, resulting in incomplete mineralization of MB, although complete decolorization of MB

ACCEPTED MANUSCRIPT

was achieved (Fig 11) In contrast to that under UV irradiation, decolorization was lower but mineralization was

higher under visible light irradiation The long irradiation time ensures a complete reaction The variations

between mineralization of MB and irradiation time are shown in Fig 12b Mineralization occurred rapidly within

the first 60 min and then slowed down for all catalysts, which can be ascribed to the pre-adsorption of MB on the surface of catalyst Thus, we can conclude that photocatalysis of MB (or cationic dyes) over rGO/TNT composites

was caused by the synergistic effect of adsorption and photocatalysis Table 3 compares the performance of

rGO/TNT composites with other adsorbents/photocatalysts reported previously The results indicate that the rGO/TNT composite is a promising candidate for the removal of dyes from effluents

3.2.3 PL properties

The efficient separation of photogenerated charged carriers is an important factor that affects photocatalytic

capability Hence, the PL spectra were recorded Fig 13 shows the results of TNTs and 3% rGO/TNT in the range

of 350–550 nm, which indicated the similarity among typical peaks at 381, 396, 438, 450, 467, 483, and 492 nm for both catalysts The strong peaks located at 396 and 467 nm were responsible for the emission and transition of bandgap energy, and their signals were reflected by charge transfer transition from Ti3+ cations to oxygen anions in the TiO6 8− complex [70,71] The combination of rGO and TNTs led to a sharp decrease in PL intensity A low PL intensity indicates a slow electron–hole recombination or a high separation efficiency That is, the electron–hole recombination with 3% rGO/TNT was slower than that with TNTs The photocatalytic activity was enhanced after rGO hybridization because of the improved separation of charged carriers and electron trap by rGO This finding explains the enhanced photocatalytic activity of the rGO/TNT composite

3.3 Mechanism of rGO/TNT photocatalysis

To explore the contribution of hydroxyl radicals (·OH), superoxide radicals (·O2), and holes (h+) during MB photocatalysis, 80 mM of isopropanol (IPA), 0.5 mM of p-benzoquinone (p-BQ), and 50 mM of iodide anions were added to MB solutions, respectively [29,65] IPA served as hydroxyl radical scavenger [50,72] Meanwhile, iodide anions in aqueous solutions served as hole scavengers [73] and p-BQ served as superoxide anion scavenger

[17,65] Fig 14 shows the results of MB photocatalysis over 3% rGO/TNT when IPA, p-BQ, and KI were added

In the case of UV irradiation, the addition of IPA significantly suppressed MB decolorization, which was only 78.4% after 60 min, indicating that the hydroxyl radical played a crucial role in MB degradation Moreover, the addition of p-BQ led to a decrease in MB removal from 98.2% to 88.7% Similarly, MB removal was only 83.9%

ACCEPTED MANUSCRIPT

when KI was added The experimental results also indicated that both h and ·O2 played a significant role in MB photodegradation [74] Similarly, the presence of IPA and p-BQ significantly decreased MB decolorization,

whereas the addition of KI had only a slight effect on MB photodegradation (Fig 14b) This finding indicates that

the hydroxyl and superoxide radicals played a dominant role in MB degradation under visible light irradiation

As shown in Fig S5, the main peak that initially appeared at a retention time (RT) of 1.94 min represents the

parent MB molecule (m/z 284) [51,76] The weak peak at m/z 270 corresponds to self-degraded intermediate metabolite of MB After 30 min UV irradiation, the characteristic peaks appeared at m/z 270 (RT of 1.73 min), m/z 256 (RT of 1.24 min), and m/z 242 (RT of 1.10 min), which can be ascribed to the formation of Azure B, Azure A, and Azure C, respectively, because methyl groups can be substituted by hydrogen atoms The heights of these peaks increased within 30 min UV irradiation However, these peaks disappeared and new characteristic peaks appeared when UV irradiation continued The characteristic peaks of MB decreased with irradiation time and disappeared after 120 min using all catalysts tested No other intermediates were observed in the subsequent photocatalysis, indicating the formation of inorganic species Notably, 2-amino-4-(methyl amino)benzene sulfonic acid (m/z 202) and 4-phenolsulfonic acid (m/z 174) were detected as by-products during UV photocatalysis of

MB over 3% rGO/TNT (Fig S6) However, these two products were undetected when using the P25 and TNT catalysts On the basis of the identified products, the degradation pathways of MB were proposed (Fig 15) The

pathways of P25 and TNTs were similar; meanwhile, a different pathway was observed for 3% rGO/TNT This finding can be ascribed to considerable changes in the structure and morphology of as-prepared composites, which was caused by hydrothermal treatment and the presence of GO

3.5 Reusability tests of the catalysts

Fig 16 shows the degradation of MB over 3% rGO/TNT after five successive cycles of UV irradiation The

photocatalytic capability of the catalyst was maintained at more than 95%, which is consistent with previous findings on MB photocatalysis over mesoporous TiO2–curcumin particles [77] The slight decrease in

ACCEPTED MANUSCRIPT

decolorization was likely caused by the adsorption of by-products on the surface and pore channels of the composite These results indicate the high chemical and operating stability of the prepared rGO/TNT composites during repeated photocatalysis cycles

4 Conclusions

Hybrid nanostructures of rGO/TNT with different amounts of rGO were fabricated by a facile hydrothermal method using GO and P25 TiO2 as precursors By incorporating rGO, the hybrid structures exhibited a remarkably high application potential due to the synergetic benefit from the enhanced adsorption of cationic dye (MB) and the separation/transportation of photoinduced electron–hole pairs Significant improvement in the photodegradation

of MB by rGO/TNT composites was observed in contrast to other catalysts, such as TiO2, under UV and visible light The rGO/TNT composite with 3 wt.% rGO exhibited the most efficient decolorization and mineralization compared with other synthesized catalysts Approximately 100% decolorization and 77.4% mineralization were achieved after 1 h of UV irradiation, whereas 95% decolorization and 78.8% mineralization were achieved after 3

h of visible light irradiation For the removal of MO (anionic dye), P25 exhibited higher photocatalytic activity than rGO/TNT composites, although the 3% rGO/TNT composite showed the highest photocatalytic activity among as-prepared materials This finding can be attributed to the negligibly small adsorption of MO on 3% rGO/TNT composite Moreover, superior reusability of rGO/TNT composites was observed The present results indicated the excellent application potential of hybrid rGO/TNT composites as adsorbents and photocatalysts to efficiently and completely remove cationic dyes for water reuse

Acknowledgements

The authors thank the financial support of this work from the Ministry of Science and Technology, Taiwan (No 106-2221-E-182-052-MY3)

ACCEPTED MANUSCRIPT

References

[1] Q Wang, S Tian, P Ning, Ind Eng Chem Res 53 (2013) 643-649 https://doi.org/10.1021/ie403402q [2] M Shaban, M.R Abukhadra, S.S Ibrahim, M.G Shahien, Appl Water Sci 7 (2017) 4743-4756 https://doi.org/10.1007/s13201-017-0637-y

[3] A Asfaram, M Ghaedi, S Agarwal, I Tyagi, V.K Gupta, RSC Adv 5 (2015) 18438-18450 https://doi.org/10.1039/C4RA15637D

[4] D Robati, B Mirza, M Rajabi, O Moradi, I Tyagi, S Agarwal, V Gupta, Chem Eng J 284 (2016)

[13] S Rajendran, M.M Khan, F Gracia, J Qin, V.K Gupta, S Arumainathan, Sci Rep 6 (2016) 31641 https://doi.org/10.1038/srep31641

[14] Y Zhang, W Cui, W An, L Liu, Y Liang, Y Zhu, Appl Catal B 221 (2018) 36-46 https://doi.org/10.1016/j.apcatb.2017.08.076

[15] R Saravanan, E Thirumal, V.K Gupta, V Narayanan, A Stephen, J Mol Liq 177 (2013) 394-401 https://doi.org/10.1016/j.molliq.2012.10.018

[16] R Saravanan, M.M Khan, V.K Gupta, E Mosquera, F Gracia, V Narayanan, A Stephen, RSC Adv 5 (2015) 34645-34651 https://doi.org/10.1039/c5ra02557e

ACCEPTED MANUSCRIPT