XRD patterns of g-C3N4, Bi4Ti3O12 and 10%[email protected]%Au@BTO were recorded to determine their crystalline structures, as displayed in Fig. 2. On the XRD pattern of g-C3N4, two diffraction peaks are observed at 13.17° and 27.40°, which correspond to the in-plane structural packing motif of tri-s-triazine units (i.e., (100) facet) and the inter-layer stacking of conjugated aromatic system (i.e., (002) facet), respectively [47]. This diffraction feature suggests that g-C3N4 nanosheets are obtained. The diffraction peaks of Bi4Ti3O12, matching well with the diffraction data of PDF#35–0795, imply the formation of pure Bi4Ti3O12 orthorhombic phase (cell: 0.545 × 3.282 × 0.541 nm3). The XRD pattern of 10%[email protected]%Au@BTO is very similar to that of bare Bi4Ti3O12, indicating no structural change of the orthorhombic Bi4Ti3O12 in the composite. The XRD pattern of the composite presents no diffraction peaks assignable to g-C3N4 nanosheets and Au nanoparticles, which is possibly due to the low diffraction intensities of g-C3N4 and low content of Au.
It is necessary to characterize the light-absorption characteristics of nanomaterials because they are highly related to their physical properties [48–50]. UV-vis DRS measurements were carried out to determine the optical absorption properties of Bi4Ti3O12, g-C3N4, 10%CN@BTO and 10%[email protected]%Au@BTO. As shown in Fig. 3(a), Bi4Ti3O12, g-C3N4 and 10%CN@BTO have a poor visible-light absorption in the wavelength region λ > 450 nm. In contrast, the 10%[email protected]%Au@BTO composite with the introduction of Au nanoparticles manifests a relatively higher visible-light absorption, which is attributed to the strong light absorption of Au nanoparticles in the visible-light region. The absorption peak observed at around 550 nm for the 10%[email protected]%Au@BTO composite can be characterized as the plasmon resonance peak of Au nanoparticles [35]. Figure 3(c) shows the apparent colors of the samples, further confirming their visible-light absorption properties. It is observed that Bi4Ti3O12, g-C3N4 and 10%CN@BTO present a white or faint yellow color, implying a weak visible-light absorption of these samples. In contrast, a gray color is observed for 10%[email protected]%Au@BTO, suggesting a relatively stronger visible-light absorption of the ternary composite photocatalyst. Figure 3(b) depicts the differential curves of UV-vis DRS spectra, from which the wavelength of absorption edge (λabs) can be derived [51]). The bandgap of bare Bi4Ti3O12 and g-C3N4 is obtained as 3.13 and 2.83 eV, respectively. For the 10%CN@BTO and 10%[email protected]%Au@BTO composites, the bandgap of Bi4Ti3O12 and g-C3N4 undergoes a slight change possibly due to their interactions.
FTIR analysis was employed to reveal the possible presence of functional groups in the samples, as illustrated in Fig. 4. Figure 4(a) presents the FTIR spectrum of Bi4Ti3O12, where the absorption peaks located at 571 and 472 cm− 1 originate from the stretching vibration of Ti–O, and the peak at 825 cm− 1 is ascribed to the Bi–O stretching vibration, which confirms the crystallization of Bi4Ti3O12 structure [35]. The peaks at 1096 and 1403 cm− 1 could originate from the symmetric and anti-symmetric stretching vibrations of CO32− groups introduced on the surface of Bi4Ti3O12 during the hydrothermal synthesis process, respectively [52]. The absorption peak located at 1642 cm− 1 is induced by the H–O bending vibration of water molecules [53]. On the FTIR spectrum of g-C3N4 (Fig. 4(b)), the characteristic absorption peaks of g-C3N4 nanosheets are observed at 807 cm− 1 (breathing mode of the tri-s-triazine units), 1247/1324 cm− 1 (stretching vibrations of C–NH–C bridges), and 1410–1640 cm− 1 (C–N heterocycles skeletal vibrations of aromatic rings) [47]. For the 10%CN@BTO and 10%[email protected]%Au@BTO composites, the absorption peaks of Bi4Ti3O12 and g-C3N4 are detected on their FTIR spectra (Fig. 4(c) and (d)), indicating that Bi4Ti3O12 and g-C3N4 are included in the composites without structural change. No characteristic peaks from Au nanoparticles are detected for 10%[email protected]%Au@BTO possible due to infrared inactivity of Au nanoparticles. For all the samples, the presence of CO32− groups and water molecules on their surface is confirmed by the observation of the infrared-absorption peaks 1403 and 1642 cm− 1.
SEM observation was carried out to reveal the morphologies of Bi4Ti3O12 and 10%[email protected]%Au@BTO. Figure 5(a) shows the SEM image of Bi4Ti3O12, implying that Bi4Ti3O12 is crystallized into nanosheets with thickness of 45–80 nm (average thickness: ~60 nm). The surface of Bi4Ti3O12 nanosheets appears smooth and clean. Figure 5(b) shows the SEM image of 10%[email protected]%Au@BTO, demonstrating the formation of composite nanosheets with thickness of 150–210 nm (average thickness: ~170 nm). Compared to that of bare Bi4Ti3O12 nanosheets, the thickness of the composite nanosheets becomes much increased, suggesting that they are constructed by g-C3N4 and Bi4Ti3O12 nanosheets with face-to-face contact. Moreover, the surface of the composite nanosheets becomes rough possibly due to the decoration of Au nanoparticles.
To reveal the microstructure of the 10%[email protected]%Au@BTO composite, TEM investigation was further performed. Figure 6(a) and (b) show the TEM images of the composite, demonstrating that g-C3N4 nanosheets and Bi4Ti3O12 nanosheets are coupled face-to-face and Au nanoparticles are possibly sandwiched between g-C3N4 and Bi4Ti3O12 nanosheets. On the selected area electron diffraction (SAED) pattern (Fig. 6(c)), one can see that the diffraction spots are periodically arranged and can be indexed into the [010] zone axis of Bi4Ti3O12 orthorhombic phase. This implies that Bi4Ti3O12 nanosheets are featured by a single-crystalline nature with highly exposed (010) facet. No diffraction spots or rings from g-C3N4 nanosheets and Au nanoparticles are detected on the SAED pattern, which is possibly due to the amorphous feature of g-C3N4 nanosheets and absence of Au nanoparticles in the selected area. Figure 6(d) and (e) illustrate the high-resolution TEM (HRTEM) images of the composite, further elucidating the construction of 2D-g-C3N4@0D-Au@2D-Bi4Ti3O12 heterojunctions with face-to-face contact. The clear lattice fringes with 2d = 0.384 nm, corresponding to the (202) facet of the orthorhombic Bi4Ti3O12, confirm the single-crystalline nature of Bi4Ti3O12 nanosheets with highly exposed (010) facet.
Energy-dispersive X-ray spectroscopy (EDS) spectrum was collected from the 10%[email protected]%Au@BTO composite (Fig. 7(a)), which clearly shows the inclusion of C/N/O/Bi/Ti/O/Au species in the composite. Except for the Cu signals, possibly induced by the TEM microgrid holder [54], no other impurity elements are detected on the EDS spectrum. Figure 7(b) displays the dark-field scanning TEM (DF-STEM) image recorded from 10%[email protected]%Au@BTO, and the corresponding EDS elemental mapping images of the area are presented in Fig. 7(c)‒(h). It is obvious that the composite nanosheets manifest the uniform distribution of the C/N/O/Bi/Ti/O elements, whereas Au element is dispersedly decorated on the composite nanosheets. The elemental mapping analysis gives support to the construction of 2D/0D/2D face-to-face contact g-C3N4@Au@Bi4Ti3O12 heterojunctions.
The 10%[email protected]%Au@BTO composite was analyzed by XPS to reveal the chemical states of the elements Bi, Ti, O, C, N and Au in the composite, as shown in Fig. 8. Two peaks separately with binding energies of 159.1 (Bi-4f7/2) and 164.4 eV (Bi-4f5/2) are detected from the XPS spectrum of Bi-4f core level (Fig. 8(a)), implying the presence of Bi3+ oxidation state [35, 55]. By deconvoluting the Ti-2p XPS core-level spectrum (Fig. 8(b)), three peaks at 458.1, 463.6 and 466.0 eV are recognized. The former two peaks correspond to the binding energies of Ti4+-2p3/2 and Ti4+-2p1/2, respectively, whereas the third peak can be ascribed to Bi-4d3/2 binding energy [35, 56]. The O-1 s XPS spectrum (Fig. 8(c)) reveals two kinds of oxygen species, i.e., the crystal lattice oxygen of Bi4Ti3O12 (529.9 eV) and chemisorbed oxygen species on the sample (531.9 eV) [35, 57, 58]. The C-1 s XPS spectrum (Fig. 8(d)) presents three peaks located at 284.8, 288.0 and 282.6 eV, which are characterized as the carbon existing in the instrument, sp2-hybridized carbon in g-C3N4, and metal carbides, respectively [47]. On the N-1 s XPS spectrum (Fig. 8(e)), the binding energy peak at 397.5 eV is ascribed to sp2-hybridized nitrogen (C = N − C) resulting from g-C3N4 [47]. The strong peak at 392.2 eV could be ascribed to the formation of metal nitrides. The observed binding energies at 84.1 (Au-4f7/2) and 87.8 eV (Au-4f5/2) on the Au-4f XPS spectrum (Fig. 8(f)) are indicative of the existence of metallic Au nanoparticles in the composite [35].
Figure 9(a) shows the time-dependent degradation curves of RhB photocatalyzed by Bi4Ti3O12, g-C3N4, CN@BTO composites and 10%[email protected]%Au@BTO composite. Before photocatalysis, the dye adsorption onto the samples is determined to be 5.4%‒9.9%. The blank (photolysis) experiment shows that RhB exhibits good stability under simulated sunlight irradiation in the absence of photocatalysts [59]. Under irradiation for 120 min, bare Bi4Ti3O12 and g-C3N4 photocatalyze 71.0% and 44.8% degradation of RhB, respectively. When Bi4Ti3O12 nanosheets and g-C3N4 nanosheets are coupled face-to-face together, the constructed CN@BTO composite nanosheets exhibit improved photodecomposition performances. The 10%CN@BTO composite with a g-C3N4 mass fraction of 10% is demonstrated to be the optimal composite photocatalyst, over which the degradation percentage of RhB reaches 85.3% after 120 min photoreaction. Furthermore, by sandwiching Au nanoparticles between Bi4Ti3O12 and g-C3N4 nanosheets, a more promising ternary 10%[email protected]%Au@BTO composite photocatalyst is achieved, which causes 94.4% of RhB to be photodegraded. As organic dyes are generally “tricky” in their photocatalytic degradation process, the exact photodegradation mechanism needs to be further investigated [60].
The photodegradation performances of the samples are further elucidated from the kinetic viewpoint. As displayed in Fig. 9(b), the degradation kinetic plots of RhB conform perfectly to the pseudo-first-order kinetic equation Ln(Ct/C0) = − kappt [61] due to their good linear behavior with R2 larger than 0.99. The apparent first-order reaction rate constant kapp can be employed for the quantitative comparison between the photodegradation performances of the photocatalysts. According to the values of kapp as displayed in Fig. 9(b), it is concluded that the 10%CN@BTO composite possesses a photodegradation activity ~ 1.8 and ~ 3.9 times over that of bare Bi4Ti3O12 and g-C3N4, respectively; whereas the photodegradation activity of the ternary 10%[email protected]%Au@BTO is increased by 2.3 and 5.0 times compared with that of bare Bi4Ti3O12 and g-C3N4, respectively.
To examine the reusability of 10%[email protected]%Au@BTO for photocatalytic degradation of RhB, the photocatalyst was collected by centrifugation after the photodegradation experiment and recovered with deionized water rinsing. The next photodegradation experiment was carried out under the same procedure by loading the recovered 10%[email protected]%Au@BTO in fresh RhB solution. To balance the minor loss of the photocatalyst after each run, fresh photocatalyst was added. As seen in Fig. 9(c), the photodegradation percentage of RhB within 120 min reaction slightly decreases from 94.4% at the 1st cycle to 90.1% at the 4th cycle, implying the degradation percentage of RhB undergoes only a minor loss (3.3%). The recycling photocatalytic experiment clearly demonstrates an excellent stability of the 10%[email protected]%Au@BTO composite photocatalyst for repeatedly degrading organic dyes.
In the 10%[email protected]%Au@BTO photodegradation system, the active species including hydroxyl (•OH) radicals, superoxide (•O2−) radicals and photoexcited holes were determined by active species trapping experiments as described in the literature [62]. Ethanol (scavenger for •OH, 5 mL), benzoquinone (BQ, scavenger for •O2−, 0.1 mmol) and ammonium oxalate (AO, scavenger for h+, 0.1 mmol) were separately added in the photoreaction solution to examine their effects on the RhB degradation. As shown in Fig. 9(d), the addition of ethanol has a minor effect on the photodegradation of RhB, suggesting a very small role of •OH in the photodegradation process. The dye degradation is obviously inhibited by the introduction of BQ or AO, confirming that •O2− and h+ are the main reactive species causing the dye degradation. In particular, the photoexcited h+ plays the largest role in the photocatalysis due to the highest suppression efficiency by AO. To quantitatively determine the role of the reactive species in the photodegradation process, the trapping experiments using more quenchers are necessary [63].
It is noted that Bi4Ti3O12 is an intrinsic n-type semiconductor (Eg = 3.13 eV) and g-C3N4 behaves as an intrinsic p-type semiconductor (Eg = 2.83 eV) [35]. The CB/VB potentials of Bi4Ti3O12 and g-C3N4 are theoretically estimated, by using the method elaborated in the literature [64, 65], to be − 0.19/+2.94 and − 1.19/+1.64 V vs NHE (normal hydrogen electrode), respectively, as schematically depicted in Fig. 10. When Bi4Ti3O12 nanosheets, Au nanoparticles and g-C3N4 nanosheets are coupled to form 2D/0D/2D face-to-face contact CN@Au@BTO heterojunctions, electrons will diffuse from n-type Bi4Ti3O12 to p-type g-C3N4, and conversely, holes will diffuse from p-type g-C3N4 to n-type Bi4Ti3O12. The role of Au nanoparticles sandwiched between Bi4Ti3O12 and g-C3N4 nanosheets is to act as the “bridge” to facilitate the transfer of photoexcited carriers. This carrier diffusion process leads to the creation of negative charge centers at the g-C3N4 interface and positive charge centers at the Bi4Ti3O12 interface, and simultaneous formation of internal electric field (pointing from Bi4Ti3O12 to g-C3N4). The created internal electric field will prevent the continuous diffusion of the charge carriers, and finally a thermal equilibrium state is reached in the CN@Au@BTO heterojunctions. Under irradiated from simulated sunlight, both Bi4Ti3O12 and g-C3N4 are photoexcited to generate CB electrons and VB holes. The internal electric field drives the electron migration from the g-C3N4 CB to the Bi4Ti3O12 CB and conversely the hole migration from the Bi4Ti3O12 VB to the g-C3N4 VB by using Au nanoparticles as the “bridge”. Consequently it results in the efficient separation of the photoproduced electrons and holes, which is confirmed by the photocurrent and photoelectrochemical impedance spectroscopy (EIS) analyses (Fig. S1 and S2). As a result, more electrons accumulated in the CB of Bi4Ti3O12 and holes accumulated in the VB of g-C3N4 are expected to participate in the photoreactions. This is the major factor resulting in the enhanced photocatalysis capacities of the ternary CN@Au@BTO heterojunction composite photocatalysts. Moreover, other secondary factors could also cause the dye degradation. For example, the LSPR of Au nanoparticles could cause the local electromagnetic field enhancement and thus stimulate the production of additional electrons/holes in Bi4Ti3O12 and g-C3N4, and the LSPR-induced electrons in Au nanoparticles could also take part in the photoreactions.
As the main reactive species confirmed in the CN@Au@BTO photocatalytic system, •O2− radicals can be thermodynamically generated via the reaction between the CB electrons in Bi4Ti3O12 with adsorbed O2 species since the Bi4Ti3O12 CB potential (− 0.19 V vs NHE) is negative to the O2/•O2− redox potential (− 0.13 V vs NHE [66]). From the thermodynamic viewpoint, the holes in theg-C3N4 VB cannot combine with OH− or H2O species to form •OH radicals, because the g-C3N4 VB potential (+ 1.64 V vs NHE) is not sufficiently positive when compared with E0(OH–/•OH) = + 1.99 V and E0(H2O/•OH) = + 2.38 V vs NHE [66]. Direct oxidation by the photoexcited h+ is suggested to be another important mechanism causing the dye degradation, which agrees with the active species trapping experiment results.