rGO Functionalized ZnO–TiO 2 Core-Shell Flower-Like Architectures for Visible Light Photocatalysis

: Core-shell heterostructures with a complex, ﬂower-like morphology, comprising a ZnO core and a TiO 2 shell decorated with reduced graphene oxide (rGO) sheets by hydrothermal wrapping, are reported to extend the absorption properties of the semiconductors toward the visible light range. The ternary photocatalysts were characterized by X-ray diffraction, ﬁeld emission scanning electron microscopy, Raman spectroscopy, diffuse reﬂectance UV–Vis, and attenuated total reﬂectance-Fourier transform infrared spectroscopy. Its photocatalytic performance was evaluated under visible light irradiation using methylene blue dye as a model pollutant. The rGO-modiﬁed ZnO–TiO 2 photocatalyst exhibited superior photoactivity compared to that of the parent ZnO–TiO 2 core-shell structures, which was dependent on its graphene content. The enhanced photocatalytic response was attributed to the higher absorption in the visible light range, as well as the pronounced electron and hole separation in the ternary system.


Introduction
The intense industrialization of the last decades has resulted in the continuous introduction of industrial and agro-industrial wastewaters in aqueous matrices; heterogeneous photocatalysis has been proposed as an efficient alternative to conventional methods for their effective treatment. TiO 2 has been the most widely used photocatalyst, but the maximization of its photocatalytic response is hindered by its inadequate absorption in the visible light range, as well as the high degree of recombination of the photogenerated charge carrier pairs. The synthesis of core-shell heterostructures comprising TiO 2 and ZnO has been evaluated as an effective approach to diminish charge recombination due to the favorable energy differences between the two semiconductors [1][2][3][4][5]. The modification of TiO 2 or ZnO with reduced graphene oxide (rGO) sheets has been shown to present synergistic advantages such as a suppressed recombination of the photogenerated charge carriers and the extension of the photoresponse toward the visible light range [6][7][8][9][10][11].

ZnO Flower-Like Structures and ZnO-TiO2 Core-Shell Heterostructures
The SEM and TEM images of the ZnO and the ZnO-TiO2 core-shell architectures are shown in Figure 2. Well-defined flower-like ZnO structures with a smooth surface ( Figure  2a,c and Figure S1a) were observed. After coating with a TiO2 shell, non-agglomerated core-shell structures were obtained with a uniform porous titania shell of 59 ± 7 nm thickness (Figure 2b,d and Figure S1b). Based on the findings of our previous work, where the photoactivity of ZnO-TiO2 core-shell, flower-like architectures with variable titania shell The Raman spectrum of the GO (Figure 1b) exhibited peaks at~1355 and 1600 cm −1 , assigned to the typical D and G bands of graphitic materials, respectively [30,31]. Specifi-cally, the D band can be attributed to the modes of the sp 2 atoms in the rings, while the G band is due to the stretching vibration of the sp 2 -hybridized C-C bonds [32].
The ATR-FTIR spectrum of the GO presented in Figure 1c shows typical peaks associated with the oxygen functionalities introduced onto the rGO sheets upon the oxidation of graphite. In particular, the broad band at~3400 cm −1 can be assigned to O-H stretching vibration, while the peaks at~1720 and 1040 cm −1 are ascribed to C=O and C-O stretching vibrations, respectively. Moreover, the peak at 1615 cm −1 is derived from the vibrations of the remaining unoxidized graphitic domains [33,34].

ZnO Flower-Like Structures and ZnO-TiO 2 Core-Shell Heterostructures
The SEM and TEM images of the ZnO and the ZnO-TiO 2 core-shell architectures are shown in Figure 2. Well-defined flower-like ZnO structures with a smooth surface (Figure 2a,c and Figure S1a) were observed. After coating with a TiO 2 shell, nonagglomerated core-shell structures were obtained with a uniform porous titania shell of 59 ± 7 nm thickness (Figure 2b,d and Figure S1b). Based on the findings of our previous work, where the photoactivity of ZnO-TiO 2 core-shell, flower-like architectures with variable titania shell thickness were evaluated, a shell thickness of~59 nm was found to exhibit the highest photocatalytic performance [2].

ZnO Flower-Like Structures and ZnO-TiO2 Core-Shell Heterostructures
The SEM and TEM images of the ZnO and the ZnO-TiO2 core-shell architectures are shown in Figure 2. Well-defined flower-like ZnO structures with a smooth surface (Figure 2a,c and Figure S1a) were observed. After coating with a TiO2 shell, non-agglomerated core-shell structures were obtained with a uniform porous titania shell of 59 ± 7 nm thickness (Figure 2b,d and Figure S1b). Based on the findings of our previous work, where the photoactivity of ZnO-TiO2 core-shell, flower-like architectures with variable titania shell thickness were evaluated, a shell thickness of ~59 nm was found to exhibit the highest photocatalytic performance [2].

rGO-Modified ZnO-TiO 2 Core-Shell Heterostructures
To investigate the effect of APTMS pre-treatment on the morphology of the ternary photocatalysts, the synthesis of rGO-modified ZnO-TiO 2 structures (denoted as ZT-rGO) was first carried out using ZnO-TiO 2 heterostructures with and without amine functionalization. As shown in Figure 3, when the surface of the ZnO-TiO 2 flower-like structures were not amine-functionalized before the introduction of the GO at a 2 wt % content, the rGO sheets did not wrap the inorganic material well (indicated by the white arrow in Figure 3a). On the contrary, the amine pre-functionalized flower-like structures were completely wrapped with the rGO sheets, which attached very well onto the inorganic surface ( Figure 3b).

rGO-Modified ZnO-TiO2 Core-Shell Heterostructures
To investigate the effect of APTMS pre-treatment on the morphology of the ternary photocatalysts, the synthesis of rGO-modified ZnO-TiO2 structures (denoted as ZT-rGO) was first carried out using ZnO-TiO2 heterostructures with and without amine functionalization. As shown in Figure 3, when the surface of the ZnO-TiO2 flower-like structures were not amine-functionalized before the introduction of the GO at a 2 wt % content, the rGO sheets did not wrap the inorganic material well (indicated by the white arrow in Figure 3a). On the contrary, the amine pre-functionalized flower-like structures were completely wrapped with the rGO sheets, which attached very well onto the inorganic surface ( Figure 3b). The observed increase in the adhesion of the rGO sheets onto the surface of the ZnO-TiO2 heterostructures can be assigned to the favorable electrostatic interactions between the amine-modified inorganic surface and the rGO sheets. The APTMS functionalization of the ZnO-TiO2 core-shell, flower-like heterostructures resulted in the formation of a positively charged inorganic surface onto which the negatively charged rGO sheets could readily attach [34][35][36][37][38]. The high contact degree between the rGO layers and the semiconductor resulted in an effective charge separation during the photocatalytic process, which is highly desirable [22][23][24][25]. Therefore, the APTMS functionalized ZnO-TiO2 core-shell, flower-like structures were employed in all the experiments discussed below.
The rGO sheets were attached onto the APTMS-modified ZnO-TiO2 core-shell, flower-like structures at different GO loadings. The obtained ternary catalyst structures are shown in Figure 4. The attachment of the rGO sheets on the flower-like structures was heavily affected by the rGO content of the composite. The rGO sheets completely wrapped the inorganic structures at 0.5 and 2 wt % graphitic loading (Figure 4a,b and Figure S1c); above this plateau value (Figure 4c), unattached rGO sheets (indicated by white arrows) were also observed.
The rGO-modified core-shell semiconductors were characterized by ATR-FTIR (Figure 4d). The spectra of the ZT-rGO structures exhibited new peaks at ∼1560 and 1230 cm −1 , which were attributed to the C=C skeletal vibration mode and C-O-C stretching vibration of rGO, respectively [39,40]. The peak at 1400 cm −1 can be attributed to the C-OH vibration mode of the graphitic material. As expected, the intensity of the peaks decreased with the reduction in the rGO loading. The observed increase in the adhesion of the rGO sheets onto the surface of the ZnO-TiO 2 heterostructures can be assigned to the favorable electrostatic interactions between the amine-modified inorganic surface and the rGO sheets. The APTMS functionalization of the ZnO-TiO 2 core-shell, flower-like heterostructures resulted in the formation of a positively charged inorganic surface onto which the negatively charged rGO sheets could readily attach [34][35][36][37][38]. The high contact degree between the rGO layers and the semiconductor resulted in an effective charge separation during the photocatalytic process, which is highly desirable [22][23][24][25]. Therefore, the APTMS functionalized ZnO-TiO 2 core-shell, flower-like structures were employed in all the experiments discussed below.
The rGO sheets were attached onto the APTMS-modified ZnO-TiO 2 core-shell, flowerlike structures at different GO loadings. The obtained ternary catalyst structures are shown in Figure 4. The attachment of the rGO sheets on the flower-like structures was heavily affected by the rGO content of the composite. The rGO sheets completely wrapped the inorganic structures at 0.5 and 2 wt % graphitic loading (Figure 4a,b and Figure S1c); above this plateau value (Figure 4c), unattached rGO sheets (indicated by white arrows) were also observed.
The rGO-modified core-shell semiconductors were characterized by ATR-FTIR ( Figure 4d). The spectra of the ZT-rGO structures exhibited new peaks at~1560 and 1230 cm −1 , which were attributed to the C=C skeletal vibration mode and C-O-C stretching vibration of rGO, respectively [39,40]. The peak at 1400 cm −1 can be attributed to the C-OH vibration mode of the graphitic material. As expected, the intensity of the peaks decreased with the reduction in the rGO loading.
Raman spectroscopy was also employed to verify the wrapping of the rGO sheets on the surface of the ZnO-TiO 2 core-shell, flower-like structures ( Figure 5). The peaks at 142, 391, 517, and 630 cm −1 observed for TiO 2 can be attributed to the Eg, B 1g , A 1g + B 1g , and E g modes of the anatase phase, respectively [41]. Two peaks of low intensity can be observed in the Raman spectrum of the ZnO flowers at 435 and 1145 cm −1 , corresponding to the E 2 high and 2LO modes of wurtzite ZnO [42][43][44]. The peaks were also detected in the spectrum of the ZnO-TiO 2 core-shell, flower-like structures (Figure 5a), whereas the ZT-rGO ternary material exhibited two additional Raman lines at 1355 and 1590 cm −1 , attributed to the D and G peaks of rGO, respectively.  Raman spectroscopy was also employed to verify the wrapping of the rGO sheets on the surface of the ZnO-TiO2 core-shell, flower-like structures ( Figure 5). The peaks at 142, 391, 517, and 630 cm −1 observed for TiO2 can be attributed to the Eg, B1g, A1g+ B1g, and Eg modes of the anatase phase, respectively [41]. Two peaks of low intensity can be observed in the Raman spectrum of the ZnO flowers at 435 and 1145 cm −1 , corresponding to the E2 high and 2LO modes of wurtzite ZnO [42][43][44]. The peaks were also detected in the spectrum of the ZnO-TiO2 core-shell, flower-like structures (Figure 5a), whereas the ZT-rGO ternary material exhibited two additional Raman lines at 1355 and 1590 cm −1 , attributed to the D and G peaks of rGO, respectively.  Raman spectroscopy was also employed to verify the wrapping of the rGO sheets on the surface of the ZnO-TiO2 core-shell, flower-like structures ( Figure 5). The peaks at 142, 391, 517, and 630 cm −1 observed for TiO2 can be attributed to the Eg, B1g, A1g+ B1g, and Eg modes of the anatase phase, respectively [41]. Two peaks of low intensity can be observed in the Raman spectrum of the ZnO flowers at 435 and 1145 cm −1 , corresponding to the E2 high and 2LO modes of wurtzite ZnO [42][43][44]. The peaks were also detected in the spectrum of the ZnO-TiO2 core-shell, flower-like structures (Figure 5a), whereas the ZT-rGO ternary material exhibited two additional Raman lines at 1355 and 1590 cm −1 , attributed to the D and G peaks of rGO, respectively. The band gap energy values (E g ) of the synthesized ZT-rGO materials were quantified by UV-Vis diffuse reflectance measurements, employing the Tauc equation: where hν is the photon energy, A is an energy-independent constant, and n = 2, as the indirect band gap, TiO 2 is the outer material in the core-shell heterostructures [1]. The normalized and transformed Kubelka-Munk plots as a function of light energy are shown in Figure 6, while the calculated E g values are presented in Table 1. A narrowing in the band-gap values and a redshift in the absorbance edge were observed in the ZT-rGO structures compared with their parent ZnO-TiO 2 analogue, in agreement with the literature [45][46][47]. An E g of 3.08 eV was calculated for the ZnO-TiO 2 heterostructures, lower than the values reported for the bare ZnO and TiO 2 (~3.2-3.3 eV) nanoparticles [2]. This phenomenon can be assigned to the generation of new energy states at the ZnO and TiO 2 heterojunction [48]. The E g values of the ternary photocatalysts decreased monotonically with the increase in the rGO content due to the generation of Ti-O-C bonds during the thermal treatment, as we reported in our earlier work [6].
TiO2 core-shell flowers (blue solid line) and (b) ZT-rGO with 0.5 (red dotted line), 2 (green dashed line), and 5 wt % (blue solid line) rGO (the TiO2 peaks are indicated with a rhombus, the ZnO peaks with an asterisk and the rGO peaks with a circle).
The band gap energy values (Eg) of the synthesized ZT-rGO materials were quantified by UV-Vis diffuse reflectance measurements, employing the Tauc equation: where hν is the photon energy, A is an energy-independent constant, and n = 2, as the indirect band gap, TiO2 is the outer material in the core-shell heterostructures [1]. The normalized and transformed Kubelka-Munk plots as a function of light energy are shown in Figure 6, while the calculated Eg values are presented in Table 1. A narrowing in the band-gap values and a redshift in the absorbance edge were observed in the ZT-rGO structures compared with their parent ZnO-TiO2 analogue, in agreement with the literature [45][46][47]. An Eg of 3.08 eV was calculated for the ZnO-TiO2 heterostructures, lower than the values reported for the bare ZnO and TiO2 (~3.2-3.3 eV) nanoparticles [2]. This phenomenon can be assigned to the generation of new energy states at the ZnO and TiO2 heterojunction [48]. The Eg values of the ternary photocatalysts decreased monotonically with the increase in the rGO content due to the generation of Ti-O-C bonds during the thermal treatment, as we reported in our earlier work [6].  Next, the photoactivity of the parent core-shell and the ZT-rGO structures was investigated in the decoloration of aqueous MB solutions under visible-light irradiation (Figure 7).  Next, the photoactivity of the parent core-shell and the ZT-rGO structures was investigated in the decoloration of aqueous MB solutions under visible-light irradiation (Figure 7). The kinetic rate constants, k, for the decoloration process of MB were calculated from the first-order equation: The regression coefficient of the linear fits, R 2 , was greater than 0.99 for all studied samples.
The photocatalytic activity of the rGO-modified ZnO-TiO2 core-shell, flower-like photocatalysts was, for all studied samples, superior compared to that of the parent ZnO-TiO2 core-shell semiconductor. A graphitic-content-dependent performance was observed, which was optimum for the ZT-rGO sample with 2 wt % rGO loading, and was reduced upon further increase in the rGO loading to 5 wt %.
The parent ZnO-TiO2 photocatalyst exhibited a ~51% dye decoloration (rate constant k = 0.0056 min −1 ) after a 120 min irradiation time. The decrease in the dye concentration for the ZT-rGO sample with 0.5 wt % rGO was 76% (k = 0.011 min −1 ), while the ZT-rGO photocatalyst with 2 wt % rGO exhibited the highest photocatalytic performance with an 86% dye removal (k = 0.015 min −1 ). Lastly, upon increasing the rGO loading to 5 wt % a decrease in the performance of the ZT-rGO photocatalyst to a 60% dye decoloration, with a rate constant k = 0.007 min −1 , was found.
The reusability of the ZT-rGO catalyst with 2 wt % rGO, which exhibited the highest photocatalytic activity, was evaluated for three photocatalytic cycles. As shown in Figure  8, the photocatalyst exhibited excellent reusability, with its performance decreasing only by ~8% after three photocatalytic cycles. This is indirect evidence that the intimate contact between the inorganic component and the rGO sheets is maintained during photocatalysis. Arguably, the rGO content would decrease since each photocatalytic cycle is followed by centrifugation and washing with water; thus, a rapid decrease in the photocatalytic performance after each cycle would be observed. The kinetic rate constants, k, for the decoloration process of MB were calculated from the first-order equation: The regression coefficient of the linear fits, R 2 , was greater than 0.99 for all studied samples. The photocatalytic activity of the rGO-modified ZnO-TiO 2 core-shell, flower-like photocatalysts was, for all studied samples, superior compared to that of the parent ZnO-TiO 2 core-shell semiconductor. A graphitic-content-dependent performance was observed, which was optimum for the ZT-rGO sample with 2 wt % rGO loading, and was reduced upon further increase in the rGO loading to 5 wt %.
The parent ZnO-TiO 2 photocatalyst exhibited a~51% dye decoloration (rate constant k = 0.0056 min −1 ) after a 120 min irradiation time. The decrease in the dye concentration for the ZT-rGO sample with 0.5 wt % rGO was 76% (k = 0.011 min −1 ), while the ZT-rGO photocatalyst with 2 wt % rGO exhibited the highest photocatalytic performance with an 86% dye removal (k = 0.015 min −1 ). Lastly, upon increasing the rGO loading to 5 wt % a decrease in the performance of the ZT-rGO photocatalyst to a 60% dye decoloration, with a rate constant k = 0.007 min −1 , was found.
The reusability of the ZT-rGO catalyst with 2 wt % rGO, which exhibited the highest photocatalytic activity, was evaluated for three photocatalytic cycles. As shown in Figure 8, the photocatalyst exhibited excellent reusability, with its performance decreasing only bỹ 8% after three photocatalytic cycles. This is indirect evidence that the intimate contact between the inorganic component and the rGO sheets is maintained during photocatalysis. Arguably, the rGO content would decrease since each photocatalytic cycle is followed by centrifugation and washing with water; thus, a rapid decrease in the photocatalytic performance after each cycle would be observed.
As presented above, the unmodified ZnO-TiO 2 core-shell photocatalyst exhibited a low, nonnegligible photoactivity in the visible light range. As mentioned in the experimental section, the lamp used in all photocatalytic experiments exhibited emission lines between 390 and 600 nm. The observed photoactivity of the ZnO-TiO 2 core-shell photocatalyst can be attributed to its band-gap value, which was measured as 3.08 eV (Table 1), corresponding to~400 nm. Therefore, the ZnO-TiO 2 core-shell photocatalyst could be effectively activated in the 390-400 nm emission range of the lamp. The photocatalytic decoloration of MB by the ZT-rGO photocatalysts was significantly promoted compared to that of the parent core-shell photocatalyst. The observed superior performance of the ZT-rGO photocatalysts can be assigned to the rGO sheets acting as sensitizers in the visible As presented above, the unmodified ZnO-TiO2 core-shell photocatalyst exhibited a low, nonnegligible photoactivity in the visible light range. As mentioned in the experimental section, the lamp used in all photocatalytic experiments exhibited emission lines between 390 and 600 nm. The observed photoactivity of the ZnO-TiO2 core-shell photocatalyst can be attributed to its band-gap value, which was measured as 3.08 eV (Table 1), corresponding to ~400 nm. Therefore, the ZnO-TiO2 core-shell photocatalyst could be effectively activated in the 390-400 nm emission range of the lamp. The photocatalytic decoloration of MB by the ZT-rGO photocatalysts was significantly promoted compared to that of the parent core-shell photocatalyst. The observed superior performance of the ZT-rGO photocatalysts can be assigned to the rGO sheets acting as sensitizers in the visible light, inducing band-gap narrowing and a shift in the absorption edge of the ZT-rGO materials toward the visible light range [49,50].
The superior photocatalytic performance of the ZT-rGO core-shell architectures can also be attributed to the appropriate energy level differences between ZnO, TiO2, and rGO. These differences result in a more efficient separation and migration of the photogenerated electrons and holes, significantly suppressing its recombination and promoting the photocatalytic rates. According to the literature, the conduction and valence bands of ZnO are more negatively charged (~0.48 eV) and are located above those of TiO2 [1,17,51,52], while rGO has a work function of −4.42 eV [38,[53][54][55][56]. Therefore, rGO can act as a trap for the electrons transferred to the conduction band of TiO2 (−4.2 eV) from ZnO and hinder the charge recombination, which enhances the overall performance of the ternary photocatalyst [50,57,58]. However, a very high rGO loading (above the optimum value of 2 wt %) decreases the active sites on the surface of the photocatalyst, which become adversely blocked by the excessive rGO, resulting in the observed reduction of the photoactivity of the ternary photocatalyst [59,60].
To elucidate the radical species that are responsible for the photocatalytic removal of MB, radical trapping experiments were conducted by the addition of tert-butanol, (2,2,6,6-Tetramethylpiperidin-1-yl)oxyl (TEMPO), and ethylenediaminetetraacetic acid (EDTA) as •OH, •O2 − , and h + traps, respectively, during the photocatalytic experiments ( Figure 9). The superior photocatalytic performance of the ZT-rGO core-shell architectures can also be attributed to the appropriate energy level differences between ZnO, TiO 2 , and rGO. These differences result in a more efficient separation and migration of the photogenerated electrons and holes, significantly suppressing its recombination and promoting the photocatalytic rates. According to the literature, the conduction and valence bands of ZnO are more negatively charged (~0.48 eV) and are located above those of TiO 2 [1,17,51,52], while rGO has a work function of −4.42 eV [38,[53][54][55][56]. Therefore, rGO can act as a trap for the electrons transferred to the conduction band of TiO 2 (−4.2 eV) from ZnO and hinder the charge recombination, which enhances the overall performance of the ternary photocatalyst [50,57,58]. However, a very high rGO loading (above the optimum value of 2 wt %) decreases the active sites on the surface of the photocatalyst, which become adversely blocked by the excessive rGO, resulting in the observed reduction of the photoactivity of the ternary photocatalyst [59,60].
To elucidate the radical species that are responsible for the photocatalytic removal of MB, radical trapping experiments were conducted by the addition of tert-butanol, (2,2,6,6-Tetramethylpiperidin-1-yl)oxyl (TEMPO), and ethylenediaminetetraacetic acid (EDTA) as •OH, •O 2 − , and h + traps, respectively, during the photocatalytic experiments ( Figure 9). The addition of EDTA, and thus the quenching of photogenerated holes, had an insignificant effect on the photocatalytic rates induced by the ZT-rGO photocatalyst with 2 wt % rGO. However, when quenching the OH radical with the alcohol, the MB removal efficiency decreased from 86% to 58% after 120 min of irradiation. This suggested that the highly reactive hydroxyl radicals produced by the photogenerated holes participate in the redox reactions. Moreover, an even more pronounced suppression of the photocatalytic performance (from 86% down to 40%) was observed upon scavenging the O 2 − species with TEMPO, which indicated that the superoxide anion radicals were the main active species in the removal of MB, in agreement with the literature [16,19,61]. As discussed above, the modification of ZnO-TiO 2 with the rGO sheets increased the separation of electrons and holes, resulting in an increased migration of electrons on the graphene sheets, which reacted further with dissolved O 2 present in the reaction solution to generate the highly responsive O 2 − species. The addition of EDTA, and thus the quenching of photogenerated holes, had an insignificant effect on the photocatalytic rates induced by the ZT-rGO photocatalyst with 2 wt % rGO. However, when quenching the OH radical with the alcohol, the MB removal efficiency decreased from 86% to 58% after 120 min of irradiation. This suggested that the highly reactive hydroxyl radicals produced by the photogenerated holes participate in the redox reactions. Moreover, an even more pronounced suppression of the photocatalytic performance (from 86% down to 40%) was observed upon scavenging the O2 − species with TEMPO, which indicated that the superoxide anion radicals were the main active species in the removal of MB, in agreement with the literature [16,19,61]. As discussed above, the modification of ZnO-TiO2 with the rGO sheets increased the separation of electrons and holes, resulting in an increased migration of electrons on the graphene sheets, which reacted further with dissolved O2 present in the reaction solution to generate the highly responsive O2 − species.

Synthesis of GO
For the synthesis of the GO, graphite powder (1.0 g) was first stirred into a mixture of H2SO4 (23 mL, 98%) and NaNO3 (0.5 g) in an ice bath. Then, KMnO4 (3 g) was slowly added over a period of 2 h. After 4 h, the reaction mixture was heated to 35 °C for 30 min and was added into H2O (50 mL), followed by heating at 70 °C for 15 min. The mixture was then poured into 250 mL H2O and the unreacted KMnO4 was removed by the addition of 3% H2O2. The GO was purified by several centrifugation/redispersion cycles in H2O

Synthesis of GO
For the synthesis of the GO, graphite powder (1.0 g) was first stirred into a mixture of H 2 SO 4 (23 mL, 98%) and NaNO 3 (0.5 g) in an ice bath. Then, KMnO 4 (3 g) was slowly added over a period of 2 h. After 4 h, the reaction mixture was heated to 35 • C for 30 min and was added into H 2 O (50 mL), followed by heating at 70 • C for 15 min. The mixture was then poured into 250 mL H 2 O and the unreacted KMnO 4 was removed by the addition of 3% H 2 O 2 . The GO was purified by several centrifugation/redispersion cycles in H 2 O and was dried under a vacuum overnight. An aqueous GO suspension (0.5 mg GO/mL H 2 O) was prepared by ultrasonication for 1 h.

Synthesis of ZnO Flower-Like Structures and ZnO-TiO 2 Core-shell Heterostructures
The synthesis of the ZnO flower-like structures, as well as the ZnO-TiO 2 core-shell, flower-like heterostructures, has been described in detail in our previous work [2]. Briefly, ZnO, with a flower-like morphology, was prepared via the reaction of 40 mL of a 0.035 M Zn(CH 3 COO) 2 ·2H 2 O and 0.055 M NaOH aqueous solution at 70 • C for 24 h. For the synthesis of the ZnO-TiO 2 core-shell, flower-like heterostructures, 0.1 g ZnO flowers were first dispersed in 35 mL ethanol by ultrasonication. A solution of 0.25 mL Ti[OCH(CH 3 ) 2 ] 4 in 5 mL ethanol was prepared separately and was added dropwise in the ZnO dispersion. The dispersion was transferred in a preheated oil bath at 80 • C and a 10 mL mixture of water in ethanol (50 mL/L) was added via dropwise. After 2 h, the ZnO-TiO 2 core-shell catalyst was isolated by centrifugation, washed thoroughly with ethanol, and dried under a vacuum.

Synthesis of ZnO-TiO 2 Core-Shell Structures Wrapped with rGO
First, the ZnO-TiO 2 core-shell structures (400 mg) were ultrasonically dispersed in 200 mL dry ethanol. Then 3 mL APTMS was injected in the dispersion, followed by an overnight reflux at 90 • C. Finally, the APTMS-modified ZnO-TiO 2 flowers were rinsed thrice with ethanol and dried under a vacuum.
Next, 300 mg of the APTMS-modified ZnO-TiO 2 heterostructures were dispersed in 100 mL H 2 O by ultrasonication, and an appropriate amount of the GO suspension was added under vigorous stirring to obtain ternary photocatalysts with 0.5, 2, and 5 wt % theoretical GO content. After stirring for 1 h, the GO-modified inorganic heterostructures were washed repeatedly with H 2 O and dried under a vacuum. This process was repeated for the ternary photocatalyst with 2 wt % GO, using core-shell heterostructures without the APTMS functionalization to elucidate the effect of the amine groups on the interactions between the GO sheets and the inorganic core-shell structures.
To induce the reduction of GO to rGO, the GO-modified ZnO-TiO 2 flower-like particles were dispersed in a mixture of H 2 O/EtOH (2:1) and were hydrothermally treated at 120 • C for 24 h in a Teflon-lined, stainless-steel autoclave. Finally, the ternary photocatalysts were calcined at 500 • C for 2 h under N 2 , to further reduce rGO and crystallize TiO 2 .

Characterization
The ternary core-shell structures were characterized by Raman spectroscopy using a Nicolet Almega XR Raman spectrometer (Thermo Scientific, Waltham, MA, USA) with a 473 nm blue laser as the excitation source in high resolution mode. ATR-FTIR spectra were recorded on a Thermo Scientific Nicolet 6700 spectrometer in the 400-4000 cm −1 range. We collected 64 scans at a resolution of 4 cm −1 . XRD patterns were measured on a PANalytical (Almelo, Netherlands) Xpert Pro X-ray diffractometer, using Cu-Kα radiation (45 kV and 40 mA). UV-Vis diffuse reflectance spectra in the 300-800 nm wavelength range were obtained using a Shimadzu UV-2401 PC spectrophotometer with an ISR-240A integrating sphere, with BaSO 4 powder used as a 100% reflectance standard. Finally, the morphology of the samples was investigated by field emission scanning electron microscopy (FE-SEM, JEOL JSM-7000F).

Photocatalytic Study
The photocatalytic performance of the parent and the rGO-modified ZnO-TiO 2 coreshell semiconductors was quantified via the decoloration of MB dye solutions (20 mg/L) under visible-light irradiation at pH 5 (corresponding to the natural pH of the solutions). The photocatalytic reactions were conducted at a photocatalyst loading of 160 ppm, while no catalyst was added for the photolytic experiments. The photocatalyst dispersion in the dye solutions was first stirred in the dark for ca. 40 min to establish the MB adsorption/desorption equilibrium, followed by irradiation using a medium-pressure mercury lamp emitting in the 200-600 nm range. The lamp was placed in a glass jacket that filtered out the UV lines at λ exc < 390 nm (Table S1) [2,6]. All photocatalytic experiments were carried out in duplicate. To evaluate the reusability of the most photoactive photocatalyst, three photocatalytic cycles were conducted by recovering the catalyst after each cycle by centrifugation and washing thrice with H 2 O before the next cycle.

Conclusions
In summary, complex ZnO-TiO 2 core-shell, flower-like structures wrapped with rGO sheets were prepared by a simple and well-controlled synthetic approach. First, core-shell flowers with a wurtzite ZnO core and an anatase TiO 2 shell of 59 ± 7 nm thickness were prepared and their surfaces were functionalized with primary amino groups using APTMS. The surface-modified ZnO-TiO 2 core-shell structures were then hydrothermally wrapped with rGO sheets at 0.5, 2, and 5 wt % loading. The amine functionalities on the inorganic surface induced strong adhesion of the rGO sheets on the TiO 2 shell due to favorable electrostatic interactions. The rGO-modified ZnO-TiO 2 photocatalysts exhibited superior photoactivity in the decoloration of aqueous solutions of MB under visible-light irradiation compared with that of the parent ZnO-TiO 2 core-shell structures, which was dependent on the rGO loading. Specifically, the ternary photocatalyst with a 2 wt % rGO content resulted in~86% dye removal after 120 min of irradiation, which was significantly higher than thẽ 51% decoloration found for the parent ZnO-TiO 2 . Overall, the enhanced photocatalytic response of the ternary photocatalysts was attributed to the increase in the absorption in the visible-light range, as well as the promoted electron and hole separation due to the favorable energy level differences between ZnO, TiO 2 , and rGO. Notably, excessive rGO loading above the optimum value of 2 wt % diminished the photoactivity of the ternary heterostructures to~60% MB removal because of the decrease in free active sites on the surface of the photocatalyst.