Characterization of The Bone-joint Structure
The morphology of the as-synthesized ZIS was characterized using SEM, AFM, TEM, and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM). As shown in Fig. 1b, ZIS-2 exhibits a hierarchical sheet-like morphology. The typical thicknesses of the ZnIn2S4, Zn2In2S5, ZIS-2, and ZIS-R (random composite samples) were 11.39, 9.16, 6.62, and 7.49 nm (Supplementary Fig. 2), respectively. Based on the Brunauer-Emmett-Teller area method, the specific surface area of ZIS-2 was 31.62 m2 g− 1, which was greater than those of ZIS-R (22.25 m2 g− 1), Zn2In2S5 (14.41 m2 g− 1), and ZnIn2S4 (7.63 m2 g− 1) (Supplementary Fig. 3). This indicates that ultrathin morphologies favor the diffusion of photogenerated carriers and high exposure of the active sites. The TEM elemental mapping (Fig. 1d–g) reveals that Zn, In, and S are uniformly distributed in ZIS, which suggests the successful synthesis of the ZIS heterostructure.
The TEM images confirm that the ZIS was assembled from multiple nanoflakes (Fig. 1h and i). A clear lattice fringe of 0.29 nm can be assigned to the (104) plane of hexagonal ZnIn2S4, and a distinct crystalline fringe of 0.32 nm can be assigned to the (104) plane of hexagonal Zn2In2S5.17,18 However, further lattice distortions and edge dislocations are clearly observed at the interfacial junction, which are attributed to the (006) plane of ZnIn2S4 and the (006) plane of Zn2In2S5.18,19 Based on the molecular models of ZnIn2S4 and Zn2In2S5, the (006) plane in the two materials exhibits the same composition and configuration (Supplementary Fig. 4). This demonstrates that an interlaced interface was formed between the heterostructures, which is called as semicoherent interface. Moreover, an interlaced interface simulation was constructed to test this (detailed simulation method and interpretation are presented in the Supporting Information, see Supplementary Fig. 5). The results revealed that the terminal atoms of the (006) plane of ZnIn2S4 were rotated by an angle of 13° to form the (006) plane of Zn2In2S5, which is produced from the interfacial strain that compelled the distortion of the lattice, suggesting that there are bone-joint links in the heterojunction because of the interatomic distances and different crystal symmetries (Fig. 1j). In addition, other interface inconsistencies were identified (Supplementary Fig. 5). HAADF-STEM was then performed to obtain the detailed information on the continuous semicoherent interfaces. No clear observable phase interface was observed (Fig. 1k–l), which was attributed to the atomic-level continuous interfaces. In addition, some atomic-level distortion exists in the semicoherent interfaces, corroborating the interfacial strain-induced sulfur vacancies (SVs), which was confirmed by the EPR results (Supplementary Fig. 7). These results confirmed the formation of a semicoherent interface bone joint in the ZIS heterojunction.
To further verify this, the crystallographic structures of the as-obtained samples were analyzed using X-ray diffraction (XRD) and Raman spectroscopy. Despite the different XRD patterns of the ZnIn2S4, Zn2In2S5, and ZIS composites, they present similar profiles, which is attributed to a repeating analogous structural unit (Fig. 1m).20 Combining inductively coupled plasma mass spectrometry (ICP-MS) (Supplementary Fig. 9 and Supplementary Table 2) and XPS elemental content analyses (Supplementary Table 3) with the XRD and TEM results, the ZnIn2S4-Zn2In2S5 heterojunction was confirmed (detailed explanation is provided in the Supporting Information). The sharped and shifted (006) XRD peak observed in ZIS-2 indicates that the compact interfacial heterostructure decreased the interplanar spacing and induced structural strain, contributing to the formation of bone-joint structures. This is confirmed by the HAADF-STEM results (Fig. 1l). The clearly distinguishable Raman peaks for the single ZIS are at approximately 128, 247, 302, and 355 cm− 1, with 128 cm− 1 being consistent with the featured layer structure and 247, 302, and 355 cm− 1 corresponding to the LO1, TO2, and LO2 modes of ZIS and associated with V(S-In-S)s (surface), V(S-In-S)i (interior), and V(S-Zn-S)s (surface), respectively (Supplementary Fig. 10).21,22 Moreover, the ZIS-2 Raman peaks shift to a lower wavenumber, which is attributed to the interaction between the semicoherent interface bone joints, suggesting the rapid transition of charge carriers in the bone-joint areas.
Due to the geometry of the semicoherent interface bone joints, the results of the XPS can determine the electronic interactions. The Zn 2p, In 3d, and S 2p doublets were investigated in ZnIn2S4, Zn2In2S5, ZIS-R (random composite heterojunction), and ZIS-2 (Supplementary Fig. 11). This demonstrated the existence of Zn2+, In3+, and S2− in both samples.23,24 Notably, the binding energies of S 2p (161.9 and 163.1 eV), In 3d (445.2 and 452.8 eV), and Zn 2p (1022.2 and 1045.2 eV) in ZIS-2 are negatively shifted compared to those of pristine ZnIn2S4. ZIS-R is in a similar scenario. However, they are still positive compared to Zn2In2S5, suggesting the migration of electrons from ZnIn2S4 to Zn2In2S5 following the contact between them, which is a typical characteristic of S-scheme heterojunctions.25 Compared to ZIS-R, ZIS-2 exhibits a positive shift, suggesting that the semicoherent interface promotes internal electronic interactions. These results clearly show that the semicoherent interface bone joints introduce large interfacial strain and regulate the electron density of Zn2In2S5 and ZnIn2S4, resulting in electronic interaction between the two components.26 Therefore, the semicoherent interface bone joints adjust the inherent chemical and energetic interactions between the two components.
Formation of S-scheme Heterojunctions
Photon absorption is a thermodynamic factor of charge transfer. Therefore, the UV-Vis diffuse reflectance spectra of the prepared samples were determined (Supplementary Fig. 12).27 As shown in Supplementary Fig. 13, the calculated CB locations for ZnIn2S4 and Zn2In2S5 were − 0.66 and − 0.88 V, respectively. Thus, the VB potentials of ZnIn2S4 and Zn2In2S5 were determined to be 1.48 and 1.44 V, respectively.28 These results indicate the possibility of establishing an S-scheme heterojunction in ZIS and satisfying the thermodynamic requirements for H2O2 generation (Supplementary Fig. 14). To gain insight into the energy band structure, density functional theory (DFT) simulations were performed. Intriguingly, the band gap of ZIS was considerably decreased, suggesting that further charge carriers are produced (Supplementary Fig. 15). As shown in Supplementary Fig. 16, unlike ZnIn2S4, Zn2In2S5, and ZIS-R, the PDOS of ZIS-2 overlaps Zn and In, implying strong electronic interactions in the bone-joint structure, which is conducive to the directional transfer of internal charges.
To confirm the S-scheme heterojunction in ZIS, several experiments were performed. First, DFT calculations were performed to predict the charge effective mass in the as-prepared samples (Supplementary Fig. 17). According to effective mass calculations, ZIS exhibits the smallest charge effective mass, indicating that the highest charge transfer rate can be expected in the heterojunction, suggesting that the tendency of charge carrier transfer follows the S-scheme mechanism.29 To further demonstrate that photogenerated electrons at the interface of a ZnIn2S4-Zn2In2S5 homojunction follow an S-scheme transfer mechanism, light-assisted KPFM was used to measure the change in the local potential of the homojunctions before and after illumination (Fig. 2a–f). Under dark conditions, ZIS-2 exhibited the largest surface potential (Supplementary Fig. 18). However, when integrated with Zeta potential results (Supplementary Fig. 19), ZIS-2 exhibited the strongest BEF (Supplementary Fig. 20), which can substantially encourage carrier migration and exciton dissociation. In addition, the surface potential of ZnIn2S4 increased from 53.86 to 65.22 mV after illumination, whereas that of Zn2In2S5 decreased from 10.03 to 1.66 mV, which is consistent with the ZIS-R trend. This demonstrates that ZnIn2S4 loses electrons and the surface potential rises, whereas Zn2In2S5 acquires electrons and the surface potential falls, further proving that the internal carrier migration follows the S-type heterojunction.30
Intrinsic charge transfer dynamics
Despite the surface BEF charge transfer pathways being clearly observable, the complexation sites and internal driving force of the BEF are undefined. Based on the KPFM results, the surface potential of ZIS-2 increased by 19.73 eV under the light condition compared to the dark condition, which is more pronounced than the ZIS-R value (3.15 eV). This increase in surface potential can be explained by the intermediate-crystal plane hastening internal electron transport.31 Such a swinging carrier screening effect occurred in ZIS-2, indicating that the (006) transition layer of the bone-joint structure plays an important role in inward charge complexation. To determine the intrinsic charge transfer dynamics, first-principle calculations based on DFT were performed. As shown in Fig. 2g, the work functions of ZnIn2S4 and Zn2In2S5 are 5.306 and 4.053 eV, respectively. This difference in work functions causes the transfer of electrons from Zn2In2S5 to ZnIn2S4 to form a BEF. However, when the bone-joint structure is modified by vacancies, the work functions of ZnIn2S4-(006) and Zn2In2S5-(006) are calculated as 4.405 and 4.227 eV, respectively. Thus, electrons are transferred from ZnIn2S4 to (006), while holes are transferred from Zn2In2S5 to the (006) crystal plane, and the (006) crystal plane becomes a charge complexation site. In addition, the charge transfer path was determined using differential electron density (Supplementary Fig. 21) and surface charge distribution mapping (Supplementary Fig. 22). These results demonstrate that the (006) crystal plane accumulates a massive charge compared to the ZIS-R interface, which is derived from interface-vacancy-induced internal electronic structural changes that concentrate electrons and holes. This suggests that the semicoherent interface acts as the source of BEFs.
TA spectroscopy was used to determine the S-scheme internal complex site charge transfer mechanism (Fig. 2h–k). The electron delay of pure ZIS-R can be fitted by two exponential functions, and the time constants are τ1 = 56.10 ps and τ2 = 575.26 ps (Fig. 2j), where τ1 reflects the stage where the electron goes from the CB to the trap state and τ2 represents the migration to the VB.32,33 While the ZIS-2 values τ1 = 9.51 ps, τ2 = 10.14 ps, and τ3 = 550.72 ps (Fig. 2k) are assigned to the stage where the electron goes from the CB to the trap state, the charge complexation on the semicoherent interface, and migration to the VB, respectively. These results demonstrate that the intermediate transition state provides a composite platform for electrons and holes. To intuitively determine the complexation sites, a metal photodeposition experiment was performed. As shown in Supplementary Fig. 24, Pt and Mn3O4 nanoparticles are deposited on the edges of the (006) lattice transition layers, confirming that the (006) transition layer acts as the charge complexation site. This can be explained by the interfacial electric field being rooted in the bone-joint structure.
Due to the directional charge complexation platform, the high efficiency of carrier migration and exciton dissociation can be predicted. As shown in Supplementary Fig. 25a, the steady-state photoluminescence (PL) spectroscopy intensity of the ZIS-2 was considerably lower than that of the control sample, indicating that the development of the bone-joint structure is essential for the radiative complexation of charge carriers. As shown in Supplementary Fig. 25b, ZIS-2 exhibits the fastest PL decay, with an expected duration of 1.02 ns compared to ZnIn2S4 (1.65 ns) and Zn2In2S5 (1.27 ns), indicating that the bone-joint structure is conductive to exciton dissociation. The Arrhenius equation [IT = I0/(1 + AExp(− Eb/KBT))] can be used to obtain the exciton binding energy (Eb).33 The calculated Eb of the ZIS-2 catalyst is 0.1 eV (Supplementary Fig. 26), which is much lower than that of ZIS-R (0.16 eV), ZnIn2S4 (0.37 eV), and Zn2In2S5 (0.20 eV). This indicates a more straightforward and rapid disintegration of the excitons into a free charge in ZIS-2. To further evaluate the effectiveness of the photoexcited separation, photocurrent and electrochemical impedance spectroscopy were undertaken. According to Supplementary Fig. 27a, the decreasing photocurrent density intensity is in the order of ZIS-2 > ZIS-1 > ZIS-3 > ZIS-4 > ZIS-5 > ZIS-R > Zn2In2S5 > ZnIn2S4, indicating that the "bone-joint" S-scheme heterojunctions enhances carrier separation. In addition, ZIS-2 has the shortest arc radius compared to the other samples (Supplementary Fig. 27b), suggesting a more effective capacity for charge transfer.
Above all, the semicoherent interface bone joint provides a platform for internal charge complexation and a powerful driving force for the BEF, which prevents the random migration of interface charges and thus enhances the carrier utilization rate. As expected, bone-joint S-scheme heterojunctions exhibit excellent photocatalytic activity.
Photocatalytic activity and mechanism
The photocatalytic conversion of H2O and O2 into H2O2 (PHH) is promising for the industrial production of H2O2. However, its use is often confined to the employment of sacrificial reagents. Here, the photocatalytic performance of the catalysts was evaluated using PHH under visible light without sacrificial reagents or cocatalysts. First, no product was produced in parallel studies conducted without light or O2, suggesting that the PHH reaction is photocatalytic and requires O2.
As shown in Fig. 3a, higher yield rates were observed in the ZIS-based S-scheme heterojunction. More importantly, ZIS-2 had the highest H2O2 production rate (5939 µmol g− 1 h− 1), almost 10 times greater than that of ZnIn2S4 (580 µmol g− 1 h− 1). Additionally, among the reported catalysts, it had the highest formation rates (Fig. 3b) (Supplementary Table 4).34–51 As illustrated in Fig. 3c, the AQY of ZIS-2 at 420 nm reached 7.08% and continued to function even at wavelengths less than 600 nm. Furthermore, the stability of ZIS-2 over the course of eight subsequent runs indicated no loss of photoactivity or structural integrity (Fig. 3d), demonstrating the excellent stability of ZIS-2. According to the XRD (Supplementary Fig. 28a) and XPS (Supplementary Fig. 28b) spectra, no obvious damage to the ZIS-2 crystal surface was observed before or after the recycling test.
The bone-joint S-scheme heterojunctions had the highest charge carrier transfer capacity for enhancing photocatalytic activity. However, more importantly, surface properties can also affect photocatalytic performance. As shown in Supplementary Fig. 29, ZIS-2 exhibits the smallest H2O contact angles, suggesting that it has a good affinity for H2O. Moreover, H2O2 has a substantially larger contact angle with ZIS than it does with H2O (Supplementary Fig. 30), indicating that the H2O2 molecules are difficult to break down in aqueous solutions. TPD studies on O2 were subsequently undertaken. As shown in Supplementary Fig. 31, an obviously enhanced chemical O2 activation capacity was observed in ZIS-2, suggesting that the bone-joint structure adjusts the inherent chemical and energetic interactions that enhance the activation of H2O and O2.
To further clarify the involvement of active species, free radical trapping studies were conducted. As shown in Supplementary Fig. 32, electrons play important roles in the photocatalytic production of H2O2. Despite the pathway of H2O2 production being a two-step 2e− progress,52 it also consists of two typical steps (i.e., superoxide radicals as the intermediates or a one-step direct 2e− process). In the system used herein, the generation of H2O2 over ZIS-2 is mainly dependent on the one-step direct 2e− process. To better understand the mechanism, in situ FT-IR tests were carried out (Fig. 4a). Absorption bands at 910 cm− 1, 1184.3–1413.2 cm− 1, and 3200–3600 cm− 1 were assigned to the adsorbed states of O2, H2O2/–OH, and H2O/–OH, respectively.8,53 Surprisingly, a peak at 1115–1295 cm− 1 belonging to Yeager-type –O–O–53 indicates that the initial adsorption structure of O2 on ZIS is primarily "bridged." Moreover, a (*OOH) signal derived from the superoxide radical was not found at 1258 cm− 1.7 These results demonstrate that the reaction follows the double bond adsorption of the O2 path, with the direct production of H2O2 without the production of intermediate *OOH.
To understand the underlying mechanisms, DFT calculations were conducted. As shown in Fig. 4b, the chemical bond lengths of –O–O– over ZIS are longer than those of –O–O, demonstrating that the activation of O2 on bone-joint S-scheme heterojunctions leans toward the –O–O– configuration. Moreover, the more negative surface of the –O–O– molecule on the ZIS suggests that it is easier for bone-joint S-scheme heterojunctions to transfer electrons via the –O–O– configuration. Furthermore, the more efficient polarization of –O–O– adsorbed on ZIS proves that electronic interaction of O2 with –O–O– models are likelier to appear. As shown in Fig. 4c, the free energy of the "bridge connection" is smaller, indicating that the O2 adsorption follows the –O–O– type path.
The reaction pathways of the different catalysts were then further investigated. As shown in Supplementary Fig. 34, the chemical molecular charge polarization of –O–O– changed more prominently over ZIS-2. This was attributed to the semicoherent interface bone joint adjusting to the energetic interaction between the two components and guiding the electron transfer directly, resulting in a more rapid electron transfer from the surface to the absorbed O2 molecule. Moreover, the smallest free energy of H2O2 production over ZIS, demonstrating a bone-joint configuration, is conducive to H2O2 generation (Fig. 4d). Based on this, the ZIS-2 photocatalytic H2O2 production mechanism is shown in Fig. 4e. When exposed to light, ZIS-2 is activated and generates relevant charge carriers. Following the bone-joint S-scheme heterojunction transfer mechanism, internal charges migrate to the "energy platform" of the (006) transition layer and generate a large BEF, while external electrons act on the Yeager-type –O–O– to produce H2O2.