Tuning the Electronic Bandgap of Graphdiyne by H-Substitution to Promote Interfacial Charge Carrier Separation for Enhanced Photocatalytic Hydrogen Production

: Graphdiyne (GDY), which features a highly π -conjugated structure, direct bandgap and high charge carrier mobility , presents the major requirements for photocatalytic applications. Up to now, all photocatalytic studies were performed without paying too much attention on the GDY band gap (1.1 eV at the G 0 W 0 many-body theory level). Such a narrow bandgap is not suitable for the band alignment between GDY and other semiconductors, making it difficult to achieve efficient photogenerated charge carrier separation. Herein,


Introduction
2][23][24][25][26] However, all these photocatalytic systems were developed without any optimization of the GDY bandgap.The narrow bandgap around 1.1 eV of pristine GDY neither meets the requirement to overcome the theoretical endothermic change in the process of water splitting (i.e., 1.23 eV), [27][28] nor achieve the interfacial separation of photogenerated charge carriers to the maximum extent, because of the mismatch between the conduction/valence band edges of GDY and the donating/accepting energy levels of the other semiconductor.Tuning the bandgap of GDY in a hybrid system would provide a photosynthetic platform to mimic the "natural photosynthesis" for efficient hydrogen evolution.0] Thus, tuning the electronic bandgap of GDY, to fabricate such a type II heterostructure represents a great potential for developing highperformance photocatalysts.
2][33][34] The traditional "top-down" doping method requires high temperature treatment under a specific atmosphere containing the heteroatom source, which is hard to control in term of doping sites (at either the acetylene moieties or the benzene rings) to achieve the targeted properties.Owing to the unique solution-synthesis method of GDY, [35][36][37][38][39][40] a controllable doping/substitution method using a monomer design strategy provides an ideal bottom-up solution for adjusting the electronic bandgap of GDY.Hydrogen-substituted graphdiyne (H-GDY), prepared by the coupling reaction of 1,3,5-triethynylbenzene monomer, appears to be a good candidate.On the one hand, the introduction of hydrogen atoms on the meta-positions of benzene rings maintains the large π-conjugated structure and excellent charge carrier mobility. 41On the other hand, the introduction of H atoms has been demonstrated to enable enlarging the bandgap of carbonbased materials, [42][43] which is beneficial to target the construction of type II heterostructured photocatalysts.
Toward this end, H-GDY was prepared through the polymerization of 1,3,5-triethynylbenzene monomer using CuCl as a catalyst.Experiments and theoretical calculations indicated H-GDY possessed a bandgap energy of ~2.5 eV and exploitable conduction band minimum (CBM) and valance band maximum (VBM) edges.As a proof of concept, TiO2 was combined with H-GDY to fabricate a type II heterostructured photocatalyst.The H-GDY/TiO2 heterojunction showed remarkable photocatalytic activity for H2 generation, under both UV-visible and visible light excitation.DFT calculations attribute this enhancement to the strong interaction and the suitable band alignment between H-GDY and TiO2, which promotes the interfacial separation of the photogenerated electrons and holes and the photosensitization of TiO2 under visible light.
TRMC measurements confirmed a longer charge carriers lifetime for H-GDY/TiO2 (~0.33 μs), compared to GDY/TiO2 (~0.24 μs), further evidencing an efficient interfacial charge separation between H-GDY and TiO2.The concept of tuning the electronic band gap of GDY to promote interfacial charge separation opens a new avenue for GDY-based photocatalysts.

Results and Discussion
We have first characterized the electronic structure of an isolated sheet of H-GDY versus GDY at the Density Functional Theory (DFT) level using periodic boundary conditions and the Perdew−Burke−Ernzerhof (PBE) functional within the Generalized Gradient Approximation (GGA) in order to describe on a relative basis the changes in the electronic structure when going from GDY to H-GDY. Figure 1 displays the band structure and the corresponding Density of States (DOS) together with the shape of the HOMO and LUMO orbital of the unit cell.8] As expected, the bandgap calculated at the PBE level is lower than the quasi-particle bandgap of 1.1 eV found at the GW many-body theory level due to the well-established underestimation of electronic bandgaps with GGA functionals.
Nevertheless, comparing at the GGA level the relative values of the bandgap of two different materials and the relative alignment of their valence or conduction band edges is meaningful, which thus will not affect the interpretation of the experimental results based on our calculations.
The theoretical results reveal that the hydrogenation of graphdiyne (H-GDY) induces the opening of the band gap up to 2.40 eV, while the width of valence band is reduced to 0.82 eV.
Those trends are rationalized by the fact that all benzene rings are connected in meta positions in H-GDY, thus limiting the degree of conjugation and hence increasing the gap.Such change is also expected to impact the electronic coupling between the unit cells and hence the mobility  H-GDY was then synthesized via the coupling reaction of 1,3,5-triethynylbenzene monomer.
In brief, 1,3,5-triethynylbenzene and CuCl were firstly added in pyridine and then heated at 60 °C for 24 h (Figure 2a).The presence of oxygen stimulated the formation of Cu I/II species, which catalysed the coupling reactions of the monomer to yield a yellowish powder (H-GDY).
Raman spectroscopy, a useful technique for carbon materials, was used to study the bonding structure of the prepared H-GDY.Compared to the terminal C≡C stretching mode at 2123 cm −1 in the monomer (Figure S1), the shifted peak at 2224 cm −1 in H-GDY confirmed the successful formation of conjugated diacetylenic linkages (Figure . 2b).The peaks located at 997 and 1589 cm -1 are attributed to the ring breathing and ring stretching of aromatic moieties, respectively. 44ansmission electron microscopy (TEM) displayed that the prepared H-GDY was composed of dispersed thin layers, which promote the adsorption of solvent and fast charge transfer (Figure S2).X-ray photoelectron spectroscopy (XPS) revealed that the prepared H-GDY powder mainly contained elemental carbon and oxygen while no Cu species were detected, indicating that they were successfully removed (Figure 2c).Deconvolution of the highresolution C 1s spectra displayed the major fractions of sp and sp 2 hybridized carbons with binding energies at 284.5 and 285.1 eV, respectively (Figure 2d).The additional oxygen atoms could originate from the partial oxidation of terminal alkyne units.In comparison to the monomer, an obvious bathochromic shift was observed in the UV−visible spectrum (Figure 2e), suggesting enlarged electron delocalization by the extended conjugated π-system.The optical bandgap (Eg) estimated from the Tauc plot is 2.50 eV (Figure 2f), which is in agreement with the DFT calculations.Such a bandgap is large enough to overcome the theoretical endothermic change associated to the process of water splitting (i.e., 1.23 eV).The valence band edge of H-GDY was determined by high resolution XPS to be 1.24 V vs NHE (Figure 2g).Combined to the bandgap of 2.50 eV, the conduction band and valence band edges of H-GDY are calculated at -1.26 V and 1.24 V vs NHE (-3.24 eV and -5.74 eV vs vacuum level), respectively.The DFT calculations and the experimental determination of the bandgap, CB and VB energy levels are in good agreement.In order to demonstrate the potential of tuning the bandgap of GDY by H-substitution to fabricate type II heterojunction, TiO2 was chosen as a model semiconductor to be interfaced with H-GDY.The microstructure of H-GDY/TiO2 composite was firstly characterized by transmission electron microscopy (TEM) and the results were shown in Figures 3a and 3b.The images indicated that the TiO2 nanoparticles were dispersed with particle size of about 20 nm (Figure 3a).The high resolution TEM (HRTEM) micrograph taken from one of them in Figure 3b revealed that this nanoparticle had a crystal phase in agreement with TiO2 Anatase (space group = I41/AMDS) with a=3.7850Å b=3.7850Å and c= 9.5140 Å (Figure 3b).The electron energy loss spectroscopy (EELS) compositional mapping of the selected region displayed a homogeneous distribution of both Ti and O.The spatial distribution of C was different from that of Ti and O elements, indicating that the sample had two composite phases (Figure 3c).Xray diffraction (XRD), Raman spectra, and XPS were performed to investigate the composition and chemical state of H-GDY/TiO2 composite.The XRD of TiO2 exhibited a mixture of anatase and rutile crystalline phase, typical of P25, with the anatase (101) preferential orientation.No typical diffraction patterns for H-GDY were observed, probably due to their relatively low amount and crystallinity (Figure 3d).Compared to pristine TiO2, the H-GDY/TiO2 composites exhibit enhanced characteristic peaks of H-GDY at 1587 cm -1 and 2223 cm -1 with the increase of H-GDY content, confirming the successful blending (Figure 3e).The high-resolution Ti 2p peak for H-GDY/TiO2 was resolved at the same binding energy, indicating that the valence band states of Ti have not changed upon coupling (Figure S3).UV-visible diffuse reflectance spectra demonstrated that the introduction of H-GDY redshifted the absorption of TiO2 (Figure S4).
The hybridization of GDY and TiO2 was also characterized and the Raman spectra and TEM images of 5% GDY/TiO2 were presented in Figure S5 and S6.The photocatalytic activity of the H-GDY/TiO2 nanocomposite was evaluated towards H2 generation in methanol/water solution (1:3 v/v) under Xenon lamp illumination.The photocatalytic activity of pristine TiO2, which showed negligible H2 generation, was improved remarkably when mixed with H-GDY.The photocatalytic activity increased with the content of H-GDY and reached an optimal efficiency for 5% H-GDY/TiO2 (Figure 4a).The total amount of H2 generated is up to 370 μmol after 6 hours, corresponding to ~6200 μmol h -1 g -1 H2 production rate (Figure 4b).A further increase in the H-GDY content reduces the photocatalytic performance.This result suggests that a higher coverage of the TiO2 surface reduces the accessibility to the active sites on the TiO2 surface.In order to assess the stability of the photocatalysts, the optimal 5% H-GDY/TiO2 sample was subjected to four consecutive test cycles of hydrogen evolution (Figure 4c).The 5% H-GDY/TiO2 sustained a stable photoactivity for hydrogen evolution after four cycles under identical experimental conditions.For comparison, GDY was synthesized and hybridized with TiO2 under the same conditions.
Both H-GDY/TiO2 and GDY/TiO2 composites displayed much higher photocatalytic efficiencies than those of pure TiO2 (Figure 4d, 4e), which is attributed to the synergetic charge transfer processes in the composites.Notably, the H2 generated rate of the 5% H-GDY/TiO2 were 1.35 times and 4 times higher than those of the 5% GDY/TiO2 under UV-visible and visible light, respectively.Such results point to the more efficient interfacial separation of photogenerated carriers between H-GDY and TiO2.Excitingly, the H-GDY/TiO2 even generate nearly 11 μmol H2 from pure water after 8 h illumination, nearly 1.6 times that of GDY/TiO2 (Figure 4f), further indicating the great potential of bandgap engineering of GDY for enhancing the photocatalytic efficiency.DFT calculations were performed to shine light on the effect of H-substitution of the GDY on the properties of the hybrid systems made with TiO2.For the sake of comparison with our experimental observations, we considered electron transfer processes only over the main exposed surface (101) of TiO2, which is non polar.The optimized structures of the built TiO2/GDY (H-GDY) interfaces are given in Figure S7 and the projected band structure of the hybrid system in Figure S8.We have calculated at DFT/PBE level with van der Waals corrections the interaction energy between the two components by the following equation: Where   ,   2 and  (/−) are the total energies of the heterostructure, individual TiO2 and individual GDY (H-GDY) layers, respectively, in the geometry of the fully relaxed heterostructure.
The GDY (H-GDY) monolayer and the top layer of the TiO2 surface undergo only a negligible buckling at the interfaces.The distance between H-GDY and TiO2 (2.57Å) is smaller than that computed between GDY and TiO2 (2.75 Å).The calculated binding energies are found to be systematically twice larger in TiO2/GDY compared to TiO2/H-GDY heterostructures (-7.32 eV versus -4.20 eV, respectively).This points to the stronger interactions between H-GDY and TiO2, which is expected to be beneficial for the interfacial charge transfer.The amount of interfacial charge transfer in the two different systems are next quantified by using a Bader charge analysis (Figure 5c).In both cases, there is a charge transfer in the ground state between the two components, with the positive charge systematically localized in the H-GDY or GDY monolayer while the negative charge is localized in the TiO2 slab, acting thus as the acceptor layer.This trend is rationalized by the small energy difference between the VB edge of GDY/H-GDY with respect to the CB edge of TiO2.Interestingly, although the size of unit cell is smaller when considering GDY, the hole accumulation in GDY in the ground state is larger than that in H-GDY, most probably due to the larger carbon areal density in GDY.The pronounced hole accumulation in GDY also explains the large downward shifts of the VB and CB edges when going from the individual components to the blend (Figure 5d).It is also of prime interest to analyse the alignment of the band edges, which are key parameters that govern the charge transfer processes at the interfaces.As shown in Figure 5d, there is a large valence band offset ΔEVB in the GDY/TiO2 heterostructure while the conduction band minimum (CBM) of GDY is lying very close to that TiO2.Under UV-visible light excitation of TiO2, charges can be generated following an efficient photoinduced hole transfer between the valence bands of the two components whereas electron transfer processes between the conduction bands are expected to be much less efficient due to the absence of a significant driving force.For H-GDY/TiO2 interfaces, there is a large offset between both the CBM and VBM of the two layers, leading to a type-II band alignment; photogenerated charge carriers are thus expected to be observed, originating from both efficient photoinduced hole transfer processes between the valence bands (from TiO2 to H-GDY) upon photoexcitation of TiO2 and efficient photoinduced electron transfer processes between the conduction bands (from H-GDY to TiO2) upon excitation of H-GDY.In turn, these two available pathways are expected to make the hydrogen evolution reaction more efficient.
To further elucidate and confirm the interfacial charge transfer between TiO2 and H-GDY (GDY), a Time Resolved Microwave Conductivity (TRMC) analysis was performed under laser-pulsed excitation at 360 nm, which enables exciton generation in both materials.
6] Upon laser excitation of the samples, the concentration of the photogenerated charge carriers increased during the pulse, and then a subsequent decay is observed due to the decrease in the amount of excess charges, either by recombination, trapping or surface reaction processes.It is worth noting that no TRMC signal was observed for pristine H-GDY (GDY), probably due to the low sensitivity of the change in the conductivity of the carbon-based material.As shown in Figure 5e, pure TiO2 shows higher maximum value of the signal than those of GDY/TiO2 and H-GDY/TiO2, which is probably attributed to the shield effect.However, these effects, usually harmful for the photoactivity, can be compensated by the influence on the charge carriers' lifetime (decay), a liming factor in the photocatalytic reaction.
In contrast to pure TiO2, GDY/TiO2 and H-GDY/TiO2 exhibited much slower global decay, leading to a higher number of charge carriers exploitable for photocatalytic reactions.Notably, a lifetime of charge carriers of ~0.33 μs was observed in H-GDY/TiO2, much longer than that for GDY/TiO2 (~0.24 μs), thus indicating that the heterojunction between H-GDY and TiO2 is more efficient for the charge separation than that between GDY and TiO2, in full consistency with the theoretical analysis and experimentally observed photocatalytic activity.All in all, the charge carriers' separation in the H-GDY/TiO2 and GDY/TiO2 photocatalysts can be rationalized: GDY acts solely as a hole transport layer, while the H-substitution enables to markedly increase the bandgap and provide well-disposed CB and VB bands to construct an efficient type II heterojunction with TiO2, fully promoting a spatial charge carrier separation (Figure 5f), which subsequently improves the photocatalytic activity towards H2 generation.

Conclusions
In conclusion, the electronic bandgap of GDY was adjusted by H-substitution and the obtained H-GDY was hybridized with TiO2 to form a type II heterojunction.The prepared H-GDY/TiO2 displayed excellent photocatalytic activity with H2 production rate of ~6200 μmol h -1 g -1 under UV-visible illumination and ~670 μmol h -1 g -1 under visible illumination, which were 1.35 times higher and 4 times higher than that of GDY/TiO2 composite, respectively.DFT calculations revealed that the superior activity of H-GDY is attributed to the suitable band alignment between H-GDY and TiO2, which promote photogenerated electron and hole separation via different pathways compared to GDY/TiO2 heterojunction.TRMC evidenced the longest lifetime of charge carriers in H-GDY/TiO2, further evidencing the successful fabrication of type II heterojunction.The concept of tuning the electronic bandgap of GDY to construct efficient heterojunctions enhancing interfacial carrier separation is certainly not confined to Hsubstitution and TiO2 but should be extended to other element doping and semiconductors for photocatalytic applications.Related works are in progress in our laboratory.
of the charge carriers, as evidenced by the flattening of the CB and VB edges in Figure 1b.The positions of the VB and CB edges of GDY (H-GDY) are calculated to be -4.92 and -5.37 eV (-3.50 and -5.83 eV), respectively, which translates into a downward shift by 1.42 eV of the VB edge and an upward shift by 0.46 eV of the CB edge when going from GDY to H-GDY.

Figure 1 .
Figure 1.Band structure and projected density of states on the carbon orbitals of GDY (a) and H-GDY (b) calculated at the DFT level and shape of the HOMO and LUMO levels of the unit cell.

Figure 2 .
Figure 2. The synthesis and characterization of H-GDY.(a) Schematic illustration of the synthesis of H-GDY powder; (b) Raman spectrum of H-GDY; (c) XPS survey spectrum and (d) high-resolution C 1s spectrum of H-GDY; (e) UV-vis absorption spectrum and (f) Tauc plot of H-GDY; (g) highresolution XPS valence band spectrum of H-GDY.

Figure 3 .
Figure 3. Characterization of H-GDY/TiO2; (a) TEM image of H-GDY/TiO2; (b) HRTEM micrograph, detail of the orange squared region and its corresponding power spectrum; (c) EELS mapping of selected region in H-GDY/TiO2; (d) XRD patterns and (e) Raman spectra of TiO2 and H-GDY/TiO2 with different content of H-GDY.

Figure 5 :
Figure 5: Side views of the 3D charge density difference in the (a) TiO2(101)/GDY and (b) TiO2(101)/H-GDY heterostructures; isosurface values for 3D charge density difference plots are 0.09 e/Å 3 ; the accumulation and depletion of electrons are represented in cyan and yellow, respectively; (c) calculated Bader charge accumulated on the GDY (H-GDY) monolayer in the heterostructure; (d) relative band alignment of VB and CB band edges before and after the formation of the heterostructure.(e) TRMC measurements of TiO2, GDY/TiO2 and H-GDY/TiO2 at 360 nm; inside graph: the fitting of the delay lifetime of the three samples; (f) schematic illustration of the interfacial charge transfer between TiO2 and H-GDY/GDY.