dz2 Band Links Frontier Orbitals and Charge Carrier Dynamics of Single-Atom Cocatalyst-Aided Photocatalytic H2 Production

The Cu single-atom catalyst (SAC) supported on TiO2 exhibits outstanding efficacy in photocatalytic hydrogen evolution. The precise operational mechanism remains a subject of ongoing debate. The focus resides with the interplay linking heightened catalytic activity, dynamic valence state alterations of Cu atoms, and their hybridization with H2O orbitals, manifested in catalyst color changes. Taking anatase TiO2 (101) as a prototypical surface, we perform ab initio quantum dynamics simulation to reveal that the high activity of the Cu-SAC is due to the quasi-planar coordination structure of the Cu atom after H2O adsorption, allowing it to trap photoexcited hot electrons and inject them into the hybridized orbital between Cu and H2O. The observed alterations in the valence state and the coloration can be attributed to the H atom released during H2O dissociation and adsorbed onto the lattice O atom neighboring the Cu-SAC. Notably, this adsorption of H atoms puts the Cu-SAC into an inert state, as opposed to an activating effect reported previously. Our work clarifies the relationship between the high photocatalytic activity and the local dynamic atomic coordination structure, providing atomistic insights into the structural changes occurring during photocatalytic reactions on SACs.


■ INTRODUCTION
−11 The reason for the high activity of SACs in photoelectrocatalysis has been studied mainly from the following two aspects: 12 (a) unsaturated active sites enhance adsorption and activation of molecules; 13 (b) isolated metal sites induce electron trap states. 14−19 To investigate the photocatalytic mechanism, we take the example of Cu-loaded anatase TiO 2 .Our findings reveal a connection between the frontier orbital theory and the electron trap states, both associated with the d z 2 orbital of the Cu atom.Adjusting the coordination environment, for instance, by introducing oxygen vacancies or adsorbing small molecules, allows control of the energy and occupancy of the d z 2 orbital, in turn influencing the activity of SACs during chemical reactions.
Despite years of research, the mechanism of the photocatalytic water dissociation on TiO 2 is still under investigation. 20Among the various SACs supported on TiO 2 , Cu, an abundant transition metal, exhibits substantial photocatalytic activity comparable to that of noble metals such as Pt, Au, and Pd.Notably, Cu-SAC/TiO 2 demonstrates intriguingly reversible alterations in color and dynamic changes in valence throughout the photocatalytic reaction. 21,22Gaining insight into the interrelation among the photocatalytic activity, reversible valence state variation, and color transformation can shed light on the factors contributing to the good performance of Cu-SAC/TiO 2 .
The pioneering investigations on Cu-SAC/TiO 2 have primarily focused on the empirical exploration of the correlation between changes in the valence state and heightened activity. 21,22However, in the absence of a comprehensive theoretical analysis, the proposed reaction mechanisms cannot explain all the experimental data.For instance, queries persist concerning the gradual shift of Cu valence from Cu 2+ to Cu + over an extended period, as opposed to an abrupt transition within a brief interval, 22 and the rationale behind the Cu 2 O post-reaction presence indicated by X-ray absorption near-edge structure (XANES) spectra, with the proposed model lacking an explicit explanation for the transformation of the coordination structure from CuO to Cu 2 O. 21 Of significance is the definitive correlation between Cu + attributes and photocatalytic activity, given the conflicting viewpoints delineated within the pioneering literature. 21,22Kim et al. 21proposed that Cu 2+ transforms into Cu + due to H atom adsorption on a neighboring lattice O atom and Cu + works as a metalloenzyme, helping in the activation of adjacent Ti atoms.Previous theoretical investigations 23 have proposed a mechanism rationalizing how Cu + may contribute to the extension of the hot electron lifetime, thereby supporting the role of Cu + as suggested by Kim et al.Subsequently, novel experimental findings have emerged, affording direct substantiation of charge accumulation on Cu atoms during photocatalysis. 22sing ultrafast absorption spectroscopy, Tang et al. 22 unveiled the potential of Cu-SAC to capture electrons, thereby supporting the perspective that Cu + functions directly as an active site.This development has inspired us to reevaluate the preceding analyses and to model the photocatalytic dissociation of H 2 O using advanced theoretical methodologies. 24n this work, reevaluating the previous analysis, 23 we demonstrate that an oxygen vacancy is key for high photocatalytic activity and that the local Cu-SAC coordination structure is important for photocatalysis.We combine experiments and quantum dynamics simulations to identify the active site as the four-coordinated Cu-SAC on the TiO 2 (101) surface with a neighboring oxygen vacancy.The quasiplanar quadrilateral structure of the Cu−O coordination is capable of trapping hot electrons through the Cu d z 2 orbital, leading to high activity.The dynamic changes in the valence state and color can be attributed to adsorption of an H atom onto a bridging O atom neighboring Cu-SAC, with the H atom originating from the Volmer water splitting.We clarify that the ability of Cu-SAC to trap electrons is not harmful.Instead, it is the reason for the high activity.Importantly, the adsorbed H atom passivates Cu-SAC/TiO 2 rather than activates it, indicating that the experimentally detected changes in the valence state and color are indicators of the high photocatalytic activity, rather than the origin.The insights obtained in this work rationalize the experimental results and provide comprehensive atomistic insights into the Cu-SAC/TiO 2 photocatalytic process.

■ RESULTS
Before discussing the details of our analysis, we present an overview of the complete process of H 2 O dissociation on Cu-SAC/TiO 2 .
Figure 1 illustrates schematically the reaction mechanism stemming from our analysis of photocatalytic H 2 O dissociation on the Cu-SAC/TiO 2 (101) surface.Before the reaction, Cu-SAC exhibits a tetrahedral coordination configuration (Figure 1a,b).After H 2 O molecule adsorption onto the Cu-SAC, the d z 2 orbital of the Cu-SAC hybridizes with H 2 O, creating an electron trap state below the conduction band minimum (CBM) of TiO 2 (Figure 1b).The hybridization, together with hydrogen bonding between H atoms and lattice O atoms, produces a quasi-planar quadrilateral CuO structure (Figure  of the Pt atom, and the hybridized orbitals are located below the TiO 2 CBM.Importantly, these findings can be extrapolated to other transition metal SACs, given the universality of the ligand-field theory in the transition metal context.In order to mitigate potential inaccuracies arising from the use of DFT in describing the hybridized orbital of H 2 O adsorbed on the SACs, we compare DFT with the DFT+U method, in which a Hubbard U correction is added on the transition metal atom, as shown in Figure S3.It is observed that the relative energy difference between the hybridized orbital involving H 2 O and the TiO 2 CBM remains relatively consistent in the two methods (see Supporting Information for details).
−27 The structure involving a transition metal SAC substituting the Ti 5c position (M-Ti 5c ) is employed as an example (Figure 2a).The energy levels of the hybridized orbitals are calculated, showing a positive correlation with the energies of the d orbitals of neutral metal atoms, as illustrated in Figure 2b.The d orbital energies of different transition metals, Figure S4a, are drawn from literature sources based on photoelectron spectroscopy experiments. 28To compare whether different data sources consistently reflect the positive correlation between the energies of the hybridized orbitals involving H 2 O and the d orbitals of the adsorbed SACs, we utilized photoelectron spectroscopy data from NIST, atomic electronic structure calculation reference data from NIST, and the current VASP calculations.The results are shown in Figure S4b− Analysis of the photocatalytic activity of various metal SACs, Figure 2c and Figure 2d, illustrates a general trend wherein the activity diminishes as the energy level of the hybridized orbital of H 2 O decreases, with Cu-SAC being an exception.To understand this anomaly, we investigate the electronic structure of different SACs (Figure S5).The orbital hybridized between the Cu-SAC and of H 2 O lies close to the valence band maximum (VBM) of Cu-SAC/TiO 2 , which should impede hot electron transfer into H 2 O.Moreover, the hot electrons captured by the hybridized orbital fail to reach the threshold energy needed for an effective dissociation, owing to their relatively low energy levels.Curiously, despite these considerations, the photocatalytic experiments underscore Cu-SAC/ TiO 2 as an efficient catalyst for water-splitting hydrogen production.This observation signifies that the increase in the energy of the hybridized orbital during the reaction prevails over the previously stated hindrances.Notably, the increase in energy of the hybridized orbital can be realized through introduction of an oxygen vacancy around the Cu atom, which can introduce extra electrons and increase the energy of the empty d orbital of the Cu atom.Indeed, existing literature substantiates that a Cu atom on TiO 2 can induce formation of lattice oxygen vacancies in Cu-SAC/TiO 2.
29−33 To corroborate the above arguments, we calculate the free energy of H 2 O adsorption (Figure S6a).The results indicate that Cu-Ti 5c fails to stably adsorb H 2 O in the absence of an O vacancy.Additionally, microcanonical ab initio molecular dynamics (AIMD) simulations indicate that an H 2 O molecule adsorbed on Cu-Ti 5c desorbs within 500 fs (Figure S6b).
To elucidate the precise coordination structure of the Cu-SAC, we conducted microcanonical AIMD simulations with the initial temperature of 300 K to assess the Cu-SAC capability in adsorbing H 2 O molecules.We consider an oxygen vacancy within the first coordination shell surrounding the Cu atom.Potential surface vacancy sites are denoted as 1′−4′ in Figure 2a.Since VO2′ and VO4′ are structurally equivalent, we focus on VO2′ only.As illustrated in Figure 2e,f, only VO1′ can stably adsorb H 2 O, forming a quasi-planar configuration, as depicted in Figure 1c.Similarly, a quasi-planar arrangement between the Cu atom and its surrounding lattice O atoms is established in the VO3′ configuration (see Figure S7).While the VO3′ configuration shows good stability, it also demonstrates an insufficient capacity for H 2 O adsorption, attributed to the coordination saturation of the Cu atom.The VO2′ configuration displays lattice rigidity in comparison to VO1′.Consequently, the VO2′ configuration lacks the structural flexibility exhibited by VO1′, making the H 2 O adsorption unstable.
In summary, since the active catalytic site needs to adsorb the H 2 O molecule stably, we choose the VO1′ configuration for the subsequent analysis.
Dynamic Structure Induced by H 2 O Adsorption on the Cu-SAC.As depicted in Figure 3a, the transition state search suggests that the adsorbed H 2 O exists in a molecular state, wherein one H atom forms a hydrogen bond with the O2′ atom surrounding the Cu-SAC (refer to Figure 1c for the configuration).The energy barrier for H 2 O dissociation is 0.34 eV, which can be overcome through thermal catalysis.However, the energy barrier for the recombination of H and OH into H 2 O is smaller, suggesting a higher likelihood of the molecular form of H 2 O. Furthermore, a relatively small dissociation energy barrier indicates an intrinsic activity of the Cu-SAC structure with lattice O vacancies.However, to enhance the dissociation of H 2 O molecules, additional energy and charge must be provided through photocatalysis.In comparison, in the absence of an injected electron, thermal catalysis would promote H and OH recombination, as discussed further below.
To validate the dynamic adsorption configuration of the H 2 O molecule, we conducted a 5 ps microcanonical AIMD simulation.Figure 3b and 3c illustrate that within the initial 3 ps, the Cu atom is displaced from its original position due to hydrogen bond formation, creating a quasi-planar quadrilateral structure involving the four neighboring O atoms (O3′, O4′, O5′, and O in H 2 O).Over the subsequent 2 ps, the Cu−O4′ bond is replaced by the Cu−O2 bond.Given the structural equivalence of the O2′ and O4′ atoms, a comparable quasiplanar quadrilateral configuration emerges.This local structural transformation is further validated through a 30 ps AIMD simulation (refer to Figure S8).
To investigate the structure after H 2 O adsorption, we calculate the statistical distribution of the H1−O3′ bond length (Figure S9 4b and Figure S5c indicates that oxygen vacancies introduce additional electrons that occupy d x 2 −y 2 orbitals.In Figure S10a, the time-dependent electron populations on various energy levels during the electron transition from the CBM back to the VBM are presented.Notably, the energy level corresponding to the highest empty hybridized orbital of H 2 O with Cu-SAC (Trap 5) can capture hot electrons, highlighting the ability of the Cu d z 2 orbital to transfer a hot electron to the adsorbed H 2 O molecule.
Since the energies of the d orbitals of the Cu atom are lower than those of the Ti atoms, the orbitals formed by hybridization of the d orbitals of the Cu atoms and the H 2 O molecules form bandgap states, corresponding to the trap states in Figure 4b.The hybridization process is illustrated according to the ligand-field theory in Figure S1.After an electron is captured, the antibonding hybridized orbital, corresponding to the d z 2 orbital closest to the CBM (Trap 5), can transfer the electron to the H 2 O molecule.This can be seen from the local charge distribution of Trap 5 shown in Figure S12.In addition, because the orbital is antibonding, the captured hot electron pushes the H 2 O molecule away from the surface, preventing the O atom formed after H 2 O dissociation from filling the original oxygen vacancy, thus avoiding deactivation of the single atom site.
The photogenerated holes accumulate at the VBM and can recombine with hot electrons in Trap 5 by transitions through Traps 1−4.We perform NAMD simulations to investigate this process.Figure 4c displays the time-dependent populations of the VBM, CBM and trap states.While the hole can efficiently transfer stepwise from the VBM to Trap 4 within approximately 1 ps, the transition to Trap 5 is significantly prolonged, taking around 10 ps, allowing buildup of the hot electron population in Trap 5. Thus, Trap 5 can effectively accumulate electrons over several picoseconds and shield them from recombination with holes, in agreement with the reported experimental results obtained by transient absorption spectroscopy. 22A closer examination of the electron trapping and recombination steps is shown in Figure 4d.The existence of a substantial hot electron population in the d z 2 orbital of Cu hybridized with H 2 O over 15 ps creates the condition needed for electron transfer to H 2 O, initiating the photocatalytic reaction.
The non-adiabatic coupling (NAC) matrix elements between different energy levels in Figure S11 indicate a relatively higher transition probability between the VBM and Traps 1−4, while a smaller transition probability between Traps 1−4 and Trap 5. Additionally, as shown in Figure S12, the wavefunction of Trap 4 primarily distributes parallel to the catalyst surface, corresponding to the d x 2 −y 2 orbital of Cu atoms, whereas the wavefunction of Trap 5 is predominantly oriented vertically to the surface, corresponding to the d z 2 orbital of Cu atoms.Such arrangement, and in particular the distribution of the Trap 5 wavefunction, relies on the quasiplanar quadrilateral structure demonstrated in Figure 1c, underscoring the important role of the quasi-planar structure induced by the H 2 O adsorption in the photocatalytic hydrogen evolution on Cu-SAC/TiO 2 .
The electron-induced dissociation of H 2 O is simulated using the impulsive two-state (I2S) method (see Supporting Information and Figure S13 for details).Figure 4e and 4f depict H 2 O dissociation subsequent to the injection of an electron into the orbital hybridized between the Cu-SAC and H 2 O, starting from the two configurations shown in Figure 3d.In configuration 1, the H1 atom freed during the water dissociation is directly captured by the O3′ atom (bridging O atom) at t = 10 fs.In comparison, configuration 2 leads to capture of the H1 atom by the O2′ atom at t = 10 fs, followed by transfer to the O3′ atom at t = 38 fs.Application of the Hubbard U correction to the Cu atoms does not significantly alter the wavefunction distribution of the Trap 5 state (see Figure S14 for details), but rather elevates its energy level.Consequently, the energy gap between Trap 4 and Trap 5 is increased, slowing electron−hole recombination, while the energy of the trapped hot electron in the Trap 5 state grows, increasing the driving force of the photochemical reaction.Both factors facilitate a more efficient H 2 O dissociation process.
Inert State of the Cu-SAC Induced by H Adsorption. The H 2 O dissociation results presented in Figure 4 indicate that the released H atoms can be captured by the lattice O atoms.An adsorbed H atom introduces an additional electron into the d z 2 orbital of the Cu atom, as illustrated in Figure S15.The introduction of the extra electron prompts the Cu atom to repel the lattice O atoms (specifically O3′ and O5′) that lie within the d z 2 orbital plane.Ultimately, this repulsion results in breaking of corresponding Cu−O bonds and the consequent transformation leading to the formation of a two-coordinated Cu atom displaying a linear configuration.This configuration aligns with the characteristics of Cu 2 O or Cu + , as indicated in the Mulliken charge analysis (Table S1).It is important to note that photogenerated electrons directly captured by the Cu atom can also induce the Cu atom to display the linear configuration.However, this process is expected to occur rapidly due to the ultrafast electron transition.In contrast, the experimental observations reveal a gradual enhancement of the Cu + signal throughout the photocatalytic reaction.This phenomenon strongly suggests that the predominant cause of the Cu + signal is the capture of H atoms by the lattice O atoms.
The simulated absorption spectra of Cu-SAC/TiO 2 before and after the H adsorption (Figure S16) clarify the origin of the color alteration during the reaction process.Namely, H adsorption is the reason the Cu-SAC turns black.An additional, crucial aspect arises at this stage: the Cu-SAC becomes passivated because of the repulsive force to adsorption of a H 2 O molecule (Figure 5a,b).The COHP analysis, displayed in Figure 5a and detailed in Table S2, reveals a significant reduction in the strength of the chemical bond between the Cu atom and the adjacent H 2 O molecule as a result of adsorption of the extra H atom.A comparison of the 3 ps microcanonical AIMD simulations conducted before and after the adsorption of the H atom underscores that H adsorption leads the system to maintain a relatively large distance between the H 2 O molecule and the catalyst surface.This is demonstrated in Figure 5c,d, as well as in Figures S17  and S18.The analysis underscores that the H atom adsorbed onto the neighboring bridge O atom prompts a transformation of the Cu atom into an inert state.Only after the H atom desorbs from the catalyst surface can a new reaction cycle be initiated.
Experimental Verification.Given that the desorption of hydrogen from the TiO 2 surface necessitates surpassing an  energy barrier of at least 1 eV, 37 it is reasonable to infer that the hydrogen desorption may serve as the rate-determining step in the process of hydrogen production on the Cu-SAC.If this is indeed the case, altering the conditions of the photocatalytic process, such as lowering the pressure, can enhance the reaction rate.
To experimentally validate the analysis, we synthesize a highly active Cu-SAC/TiO 2 composite (Figures S19 and S20).The homogeneous incorporation of individual Cu atoms within Ti vacancies is confirmed through atomic-resolution high-angle annular dark-field (HAADF) STEM imaging, as illustrated in Figure 5e.Subsequently, a series of controlled experiments aimed at the photocatalytic hydrogen evolution are conducted.Initially, the gradual conversion of Cu 2+ to Cu + during the reaction is confirmed using in situ electron paramagnetic resonance (EPR) (Figure S21).We then proceeded to explore the impact of pressure.As depicted in Figure S22, an increase in pressure corresponds to a rapid decline in the photocatalytic activity of Cu-SAC/TiO 2 , while the pure TiO 2 catalyst exhibits negligible variations under various pressures, as shown in Figure S23.This observation underscores the adverse influence of slow H atom desorption on hydrogen evolution over Cu-SAC/TiO 2 under high pressure.The experiment involving dynamic pressure changes, as presented in Figure 5f, also reaffirms this effect.
Even after illumination stops, the production of the H 2 gas (depicted in Figure 5g) continues at a slow rate and eventually stops after several hours.Simultaneously, the initially black catalyst reverts to its original white color (Figure S24).These phenomena collectively indicate the formation of passivated Cu atoms during the photocatalytic reaction, emphasizing the pivotal role of H desorption from neighboring O atoms in maintaining reaction continuity.Hence, it is imperative to enhance the desorption rate of H atoms, a factor that can be modulated by altering the system pressure or temperature. 38,39ne of the major difficulties in the application of SACs is their insufficient lifetime and stability.For instance, in SACs supported on C 3 N 4 , the M−C or M−N bonds are easily affected by heat or photoelectric effects.The energies of the M−C and M−N bonds are relatively low, and it is easy for isolated atoms to migrate and form clusters, leading to deactivation.In comparison, single atoms are anchored on metal oxide substrates by M−O bonds, which have high energy and strongly resist deactivation.In addition, Cu initially exists in a four-coordinate state in the Cu-SAC studied here.However, after one reaction cycle, the H adsorption induces the Cu atom to enter the two-coordinate state.Although the coordination number decreases, a protective electronic layer is formed around Cu, preventing further contact with H 2 O and capture of more charge, thus avoiding a further decrease in the coordination number.Therefore, single atoms supported on metal oxides have good stability.In our experiments, no significant deactivation was observed after 20 h of changing pressure and irradiation conditions (Figure 5f).

■ DISCUSSION
The comprehensive process of photocatalytic hydrogen evolution through water splitting on Cu-SAC is outlined in Figure 1 21 and EPR 22 characterizations.In contrast to the proposed beneficial role of Cu + , 21,22 we illustrate that Cu + is an inert state of the Cu-SAC appearing during the reaction.To enhance the efficiency of the photocatalytic hydrogen evolution, it becomes essential to accelerate the rate of desorption of H atoms, a task achievable by reducing system pressure or elevating reaction temperature. 38,39egarding transition metal cocatalysts present in the bulk phase: small molecules can interact with a cluster of metal surface atoms.−42 Accounting for the energy splitting of d orbitals through the ligand-field theory, the frontier orbitals of a metal SAC loaded onto TiO 2 or other metal oxides with an octahedral structure include d x 2 −y 2 and d z 2 orbitals.NAMD calculations for Cu-SAC/TiO 2 reveal that the hot electron is trapped by the d z 2 orbital of the transition metal atom, thus creating an efficient bridge for charge transfer to the adsorbed molecule. 12ccordingly, we propose the following formulation to estimate the activity of SAC loaded on an octahedral metal oxide: a subset of d orbitals of a single transition metal atom ≈ an electron trap state ≈ a charge transfer bridge.Cu-SAC/TiO 2 is the case in point.If the Cu atom substitutes Ti 5c , the Cu d z 2 energy level becomes excessively low to effectively capture high-energy hot electrons.Consequently, the trapped hot electrons lack the energy necessary to initiate H 2 O dissociation.However, the presence of a neighboring oxygen vacancy results in the release of two extra electrons, which occupy the d x 2 −y 2 orbital, regardless of whether the d x 2 −y 2 orbital is initially higher or lower in energy than the d z 2 orbital.The preferred position of the d z 2 orbital closer to the CBM of the substrate material endows Cu-SAC/TiO 2 with an excellent photocatalytic activity.The notion of the d z 2 orbital of SAC/ TiO 2 functioning as a descriptor for the hot electron trap state driving the photochemistry can be extended to other SACs involving different substrate catalysts.
It is important to recognize the synergistic effect of the quasi-planar Cu−O coordination structure and the d z 2 orbital as electron transport bridge.Considering the potential of the d z 2 orbital to capture electrons, it is natural to expect that H 2 O can be adsorbed onto Cu atoms through the d z 2 orbital to form a hybridized molecular orbital, thereby effectively utilizing the charge captured by d z 2 .However, because the d z 2 orbital has an antibonding property, five-coordinated Cu atoms cannot facilitate effective adsorption.At the same time, when there is an oxygen vacancy around the Cu atom, the fourcoordinated Cu−O structure gains sufficient flexibility.Although the d z 2 orbital still has an antibonding property, H 2 O can stably form a quasi-planar structure with the help of hydrogen bonding and adsorb onto the Cu atom through the d z 2 orbital.Considering that SACs supported on metal oxide substrates typically exhibit an octahedral metal−oxygen structure, SACs that replace cations of metal oxides are often unable to generate sufficient adsorption capacity in the absence of oxygen vacancies.Therefore, even if they can capture charges, the charge cannot be effectively used to drive reactions.The effects of the local structure of "oxygen vacancies + single atoms" on adsorption and charge capture studied in this paper can be extended to other similar catalysts.

■ CONCLUSION
We investigated the reversible dynamic coordination changes in Cu-SAC/TiO 2 and their correlation with different steps of the photocatalytic hydrogen evolution mechanism.The reported findings underscore the pivotal role of the local Cu−O coordination structure in the heightened photocatalytic activity of Cu-SAC/TiO 2 .The simulations show that the Cu d z 2 orbital effectively captures photogenerated electrons and injects them into the H 2 O molecule, promoting the watersplitting reaction.
The H atom generated during the Volmer process adsorbs to a bridging O atom neighboring the Cu-SAC, resulting in changes in the valence state and color of Cu.The H atom adsorption does not activate Cu-SAC/TiO 2 .Conversely, it induces an inert state in Cu by introducing an additional electron into the Cu d z 2 orbital.Consequently, the corresponding color and valence state changes serve as indicators of the photocatalytic process rather than the origin of hydrogen evolution.The study provides a novel perspective on the mechanism of water splitting on SACs and offers valuable guidelines for the design of SACs for photocatalytic reactions.

Figure 1 .
Figure 1.Diagram of photocatalytic hydrogen evolution on the Cu-SAC/TiO 2 surface.Schematics showing (a) the water splitting on the Cu-SAC/TiO 2 (101) surface and (b−f) the steps of hydrogen evolution with the corresponding local coordination structures.The process of H 2 O splitting on Cu-SAC starts with the adsorption of H 2 O molecules onto the Cu atom.Subsequently, (b, c) a quasi-planar quadrilateral Cu−O coordination structure is formed due to hybridization of the Cu and H 2 O orbitals below the CBM of TiO 2 and hydrogen bonding between H 2 O and lattice O atoms.(d) Under illumination, photoexcited electrons accumulate at the CBM before transition to the hybridized bandgap state formed by the hybridized d orbital of the Cu atom and p orbital of the O atom.The hot electrons trapped in the bandgap state facilitate the transfer of charge to H 2 O through the Cu−O bond, leading to H 2 O dissociation.(e) The H atoms released from the H 2 O dissociation are captured by lattice O atoms, causing a transformation of the Cu−O coordination structure into a linear arrangement.The adsorption of H atoms induces a repulsive force on subsequent H 2 O molecules, rendering the Cu-SAC inert.(f) Following H atom desorption, the Cu−O coordination structure can revert to its initial state, initiating a new cycle of photocatalysis.

Figure 2 .
Figure 2. Determination of the local coordination structure of the metal-SAC/TiO 2 .(a) Single transition metal atom substituting Ti 5c on the TiO 2 anatase (101) surface.The H atoms in the H 2 O molecule are marked as 1−2, the lattice oxygen atoms around the transition metal atom are marked as 1′−5′, and the lattice Ti atoms around the transition metal atom are marked as 1″−3″.(b) Hybridized orbital energy level of adsorbed H 2 O vs d orbital energy of different metal atoms.(c, d) Photocatalytic hydrogen evolution activity of different Metal-SACs/TiO 2 .The hydrogen evolution rates are from the experiments reported in refs 22 and 21, respectively.(e) Relative formation energies of different O atom vacancy configurations with H 2 O adsorbed on Cu-SAC.(f) Lengths of bonds between the Cu atom and the O atom in the adsorbed H 2 O molecule.
d.All data show a positive correlation between the energies of the hybridized orbital and the d orbital, indicating that if the energy of the d orbital of a SAC is inside the TiO 2 bandgap (FigureS1) the hybridized orbital involving H 2 O is also inside the bandgap, facilitating H 2 O adsorption and hot electron capture and transfer from TiO 2 to H 2 O.

Figure 3 .
Figure 3. Dynamic coordination structure of Cu-SAC after H 2 O adsorption.(a) Free energy diagrams of the H 2 O adsorption and dissociation, including the molecular state, dissociative state, and transition state.(b, c) Evolution of the key bond lengths in canonical ab initio molecular dynamics simulation with the initial temperature of 300 K.Note that O2′ and O4′ are equivalent, and the switching of the Cu−O2′ and Cu−O4′ bond lengths at 3 ps creates a structure equivalent to that before 3 ps.(d) Two adsorption configurations, chosen prior to 3 ps in (b), and used as the initial structure for the H 2 O photodissociation dynamics shown in Figure 4e,f.The periodic transformation of the Cu−O coordination structure is illustrated by the changes in the bond length at 3 ps, as depicted in (b).(d) Statistical analysis of the H1−O3′ bond lengths during the 0−3 ps time frame reveals the existence of two predominant local structures preceding the periodic Cu−O structural changes.The statistical data for the H1−O3′ bond lengths is presented in Figure S9.

Figure 4 .
Figure 4. Charge carrier dynamics and photoinduced H 2 O dissociation.(a) Crystal orbital Hamilton population (COHP) between the Cu atom and the O atom of H 2 O.(b) Projected density of states (PDOS) of H 2 O adsorbed Cu-SAC/TiO 2 .(c) Hot electrons relax to the TiO 2 CBM, hop to and accumulate in the Trap 5 state, and then recombine with holes that hop from the VBM to Traps 1−4, as illustrated in the inset.The time-dependent populations of different states provide insights into the dynamics of electron and hole trapping and recombination within the spindown channel.(d) Starting with electrons in the CBM, the inset demonstrates electron trapping by Trap 5 and electron recombination with a hole in Trap 4. The time-dependent populations of various states elucidate the dynamics of electron and hole trapping and recombination, with emphasis on the transition involving Traps 4 and 5 and the CBM.(e, f) Following H 2 O adsorption and subsequent acquisition of hot electrons, H 2 O undergoes desorption from the catalyst surface and subsequent dissociation.The evolution of the Cu−O and H−O bond lengths after electron transfer to H 2 O is depicted for two representative initial configurations, as shown in Figure 3d.
) during the initial 3 ps of the canonical AIMD simulation of Figure 3b.The analysis reveals two dominant adsorption configurations based on the length of the hydrogen bond between the H 2 O molecule and the Cu SAC surface.The two adsorption configurations are shown in Figure 3d.In conclusion, the AIMD simulation underscores a periodic reconstruction of local Cu−O coordination.Hot Electron Injection into the Hybridized Cu-H 2 O Orbital. Figure 4a illustrates the crystal orbital Hamilton population (COHP) between Cu and O in H 2 O 34−36 after the adsorption of the H 2 O molecule onto the Cu atom.The corresponding PDOS is presented in Figure 4b.The results indicate that the hybridized orbital of H 2 O is in the spin-down channel and exhibits antibonding characteristic.The Cu atominduced localized electronic states, labeled as Traps 1−5, span the bandgap from the VBM to the CBM.Thus, the hot electron in the CBM can transition to the midgap orbital hybridized between the Cu-SAC and H 2 O.Comparison of the data in Figure

Figure 5 .
Figure 5. Cu-SAC/TiO 2 passivation by H adsorption. (a) Crystal orbital Hamilton population between the Cu atom and the O atom of H 2 O with an extra H atom adsorbed onto the neighboring bridging O atom of Cu-SAC.(b) PDOS of Cu-SAC/TiO 2 with the adsorbed H atom. (c, d) Distributions of the Cu−O bond lengths before and after H adsorption at 300 K.The observed increase in the Cu−O bond lengths following the H adsorption suggests that the additional electron released from the H atom is injected into the d orbital of the Cu atom, thereby generating a repulsive force between the Cu atom and the H 2 O molecule.(e) High-resolution STEM and STEM-EDS mapping images of Cu-SAC/TiO 2 .(f) Photocatalytic H 2 evolution experiments of Cu-SAC/TiO 2 under dynamic pressures with Ar atmosphere protection.0 kPa represents the standard atmospheric pressure.(g) H 2 evolution with light on and off.The sensitivity of the photocatalytic performance of Cu-SAC to pressure suggests that the rate-determining step in photocatalytic hydrogen production involves H diffusion and desorption.Comparative data for Cu-SAC/TiO 2 and TiO 2 are presented in Figures S22 and S23.
Figure 5. Cu-SAC/TiO 2 passivation by H adsorption. (a) Crystal orbital Hamilton population between the Cu atom and the O atom of H 2 O with an extra H atom adsorbed onto the neighboring bridging O atom of Cu-SAC.(b) PDOS of Cu-SAC/TiO 2 with the adsorbed H atom. (c, d) Distributions of the Cu−O bond lengths before and after H adsorption at 300 K.The observed increase in the Cu−O bond lengths following the H adsorption suggests that the additional electron released from the H atom is injected into the d orbital of the Cu atom, thereby generating a repulsive force between the Cu atom and the H 2 O molecule.(e) High-resolution STEM and STEM-EDS mapping images of Cu-SAC/TiO 2 .(f) Photocatalytic H 2 evolution experiments of Cu-SAC/TiO 2 under dynamic pressures with Ar atmosphere protection.0 kPa represents the standard atmospheric pressure.(g) H 2 evolution with light on and off.The sensitivity of the photocatalytic performance of Cu-SAC to pressure suggests that the rate-determining step in photocatalytic hydrogen production involves H diffusion and desorption.Comparative data for Cu-SAC/TiO 2 and TiO 2 are presented in Figures S22 and S23.
. Our findings demonstrate that the flexible interaction between Cu-SAC and lattice O atoms establishes an active site for H 2 O molecule adsorption.Facilitated by a hydrogen bond between H 2 O and lattice O atoms, this interaction results in the formation of a quasi-planar structure, creating a hybridized Cu/H 2 O orbital below the TiO 2 CBM.The photogenerated hot electrons captured by the Cu d z 2 orbital can be injected efficiently into the H 2 O molecule.Additionally, H atoms released during the H 2 O dissociation can be adsorbed onto lattice O atoms, inducing a transformation of the coordination configuration of the Cu atom from CuO to Cu 2 O.These findings align with the XANES