Semimetallic Bismuthene with Edge‐Rich Dangling Bonds: Broad‐Spectrum‐Driven and Edge‐Confined Electron Enhancement Boosting CO2 Hydrogenation Reduction

Broad‐spectrum‐driven high‐performance artificial photosynthesis is quite challenging. Herein, atomically ultrathin bismuthene with semimetallic properties is designed and demonstrated for broad‐spectrum (ultraviolet‐visible‐near infrared light) (UV–vis–NIR)‐driven photocatalytic CO2 hydrogenation. The trap states in the bandgap produced by edge dangling bonds prolong the lifetime of the photogenerated electrons from 90 ps in bulk Bi to 1650 ps in bismuthine, and excited‐state electrons are enriched at the edge of bismuthine. The edge dangling bonds of bismuthene as the active sites for adsorption/activation of CO2 increase the hybridization ability of the Bi 6p orbital and O 2p orbital to significantly reduce the catalytic reaction energy barrier and promote the formation of C─H bonds until the generation of CH4. Under λ ≥ 400 nm and λ ≥ 550 nm irradiation, the utilization ratios of photogenerated electron reduction CO2 hydrogenation to CO and CH4 for bismuthene are 58.24 and 300.50 times higher than those of bulk Bi, respectively. Moreover, bismuthene can extend the CO2 hydrogenation reaction to the near‐infrared region (λ ≥ 700 nm). This pioneering work employs the single semimetal element as an artificial photosynthetic catalyst to produce a broad spectral response.


Introduction
Simulating plant photosynthesis and using sunlight as the driving force to reduce CO 2 into fuels is an effective strategy to alleviate the energy crisis and greenhouse effect. [1,2]Photocatalysts with broad spectral response, high photogenerated electron mobility separation efficiency, as well as efficient CO 2 adsorption/activation capabilities are crucial to the development of photocatalytic CO 2 conversion. [3]Traditional semiconductor photocatalysts only absorb ultraviolet light and a small amount of visible light, thus making the utilization of solar light very low (Scheme 1i-iv), [4,5] while wide bandgap semiconductors are not conducive to efficient migration and separation of photogenerated carriers. [6]In addition, high-purity semiconductor catalysts have fewer active sites for adsorbing and activating CO 2 molecules on the surface. [7]cheme 1. Possible phototransition processes in i) Wide-bandgap semiconductors, ii-iv) Narrow-bandgap semiconductors, and v-viii) Semimetallic materials.
Although the active sites on the surface of semiconductor catalysts can be increased by doping and defect regulation, the dopant and defect sites will become recombination centers for photogenerated carriers, rendering them ineffective in improving the photocatalytic properties. [8]Therefore, it is highly desirable to design novel photocatalysts with "Troikas" running hand in hand.
Between traditional semiconductors and metals, semimetallic materials possess photocatalytic properties with a broad spectral response due to their narrow bandgaps (excited by infrared light), special energy band structures, and high carrier densities. [9]Semimetallic catalysts have a fully electronically occupied full band (B −1 ) and the lowest fully unoccupied empty band (B 1 ) similar to semiconductors, as well as a partially electronically occupied intermediate band (IB) (Scheme 1v-viii). [10,11]The photoexcited electrons on B −1 or IB can generate photogenerated electron-hole pairs through the single transition process (Scheme 1v,vi) or continuous transition process (Scheme 1vii,viii). [12]This means that if the electronic potential excited by any of the above transition mechanisms satisfies the CO 2 reduction potential, the semimetallic materials are expected to achieve photocatalytic CO 2 conversion to carbon-based fuels under broad-spectrum. [13]Some semiconductor, such as metal oxides, [14] sulfides, [15] and nitrides [11,16,17] exhibit semimetallic band structures and properties, thus showing potential broad-spectrum photocatalytic activity in water splitting and CO 2 conversion.However, the direct construction of broadspectrum-excited semimetallic catalysts from metal monomers has not been demonstrated for photocatalytic CO 2 reduction.
Among the various metals, bismuth (Bi) with the largest number of atoms in the VA group is a "green metal". [18]Similar to graphene, MoS 2 , and black phosphorene, there is van der Waals force between the layers of bismuthine with the graphenelike curved hexagonal lattice. [19]Moreover, as the number of bismuthene layers decreases, its bandgap increases gradually, which arises from the metallicity of bulk Bi to the semimetallicity of bismuthene. [20]The bismuthene maintains a high carrier mobility of about 2 × 10 4 cm 2 V −1 s −1 , which is much higher than that of traditional semiconductor materials and compara-ble to that of graphene. [21]However, bulk Bi preferentially exposes the thermodynamically stable and coordination-saturated crystal planes, consequently limiting the exposure of more CO 2 adsorption/activation sites and reducing the CO 2 hydrogenation activity. [22,23]To overcome this disadvantage, two-dimensional ultrathin bismuthene with edge-rich unsaturated coordination dangling bonds is expected to maximize the exposure of catalytic active sites to enhance CO 2 hydrogenation reduction. [24,25]Nevertheless, the rational modulation of bismuthene with edge-rich dangling bonds and the in-depth dissection of the intrinsic mechanism for enhanced CO 2 hydrogenation remain a challenging task.
Herein, semimetallic bismuthene with edge-rich dangling bonds is designed and applied to artificial photosynthesis for the first time.Under broad-spectrum excitation, the photogenerated electrons undergo successive intraband and interband transitions in the IB and B 1 bands of bismuthene with unique semimetallic properties.Femtosecond transient absorption spectroscopy verifies that the trap states in the bandgap produced by edge dangling bonds prolong the lifetime of photogenerated electrons enriched at the edge of bismuthine.The abundant dangling bonds at the edge of bismuthene serve as active sites for the adsorption/activation of CO 2 to lower the energy barrier of CO 2 hydrogenation to accomplish high photocatalytic CO 2 conversion activity.Upon exposure to a xenon lamp and visible to nearinfrared light, the utilization ratio of photogenerated electron reduction of CO 2 to CO and CH 4 on bismuthene with the edge confinement effect is exponentially improved compared to bulk Bi.This work opens new horizons for the design of semimetallic artificial photosynthetic catalysts with broad spectral response.

Results and Discussion
Using the solution containing reducing hexamethylenetetramine, bismuthene is prepared by a one-step solvothermal method.Powder X-ray diffraction (XRD) (Figure 1a) reveals that the diffraction peaks from bismuthene arise from the bulk Bi powder (Figure S1, Supporting Information) that can be indexed to the standard card (JCPDS card No. 44-1246).Figure 1b shows two Raman bands at 70.0 and 96.3 cm −1 assigned to the E g and A 1g modes of metallic Bi, respectively. [26]However, XRD and Raman scattering do not show peaks of bismuth-based oxide, indicating that both bismuthene and bulk Bi are pure.The TEM images reveal the ultrathin structure of bismuthene (Figure S2, Supporting Information).And the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images confirm that the exposed bismuthene planes are (012) and (003) with lattice spacings of 0.328 and 0.395 nm, respectively (Figure 1c,d). [27]Bismuthene has two atomic bilayers with a zig-zag structure (Figure 1e) and a thickness of 1.2-1.5 nm equivalent to 2-3 atomic bilayers (Figure S3, Supporting Information). [28]Owing to the ultrathin structure, there are many amorphous bismuth atoms with abundant unsaturated metallic dangling bonds at the edge of bismuthene (Figure 1f).The strong electron paramagnetic resonance (EPR) signal of bismuthene at g = 1.998 further corroborates the edge-rich dangling bonds (Figure S4, Supporting Information). [29,30]Moreover, compared to bulk Bi (0.84 m 2 g −1 ), the ultrathin structure and abundant edge effects produce a larger specific surface area (22.00 m 2 g −1 ) (Figure S5, Supporting Information), which is beneficial to CO 2 adsorption. [31]X-ray absorption fine spectroscopy (XAFS) is performed to determine the chemical structure of bismuthene, referred to as the Bi foil and Bi 2 O 3 .As shown in Figure 1g, the absorption onset of the Bi L 3 -edge X-ray absorption near edge structure (XANES) spectra for the Bi foil, bismuthine, and Bi 2 O 3 shift gradually toward higher energy indicative of increasing valence state of bismuthene compared to the Bi foil.The FT-EXAFS spectra are used to evaluate the density of neighboring atoms as a function of radial distance (Figure 1h and Figure S6, Supporting Information).The measured bond lengths of bulk Bi of 2.6 and 3.1 Å correspond to the intra-and inter-layer Bi-Bi spacings, respectively (Figure 1h blue line), [32][33][34] and the measured bond length of Bi 2 O 3 of 1.6 Å represents the Bi-O distance (Figure 1h yellow line). [35,36]The prepared bismuthine (Figure 1h pink line) shows three characteristic peaks of the two reference samples.According to the EXAFS fitting parameters at the Bi L 3 -edge (Table S1, Supporting Information), the Bi-O, Bi-Bi (intra), Bi-Bi (inter) coordination numbers (CN) of bismuthene are 1.3 ± 0.1, 1.0 ± 0.2, and 3.0 ± 1.3, respectively.The Bi-Bi (inter) coordination numbers of bismuthene are similar to that of metallic Bi (CN Bi-Bi (inter) = 3.0), while the Bi-O, Bi-Bi (intra) coordination numbers of bismuthene are smaller than those of Bi 2 O 3 (CN Bi-O = 3.0) and metallic Bi (CN Bi-Bi (intra) = 3.0).This is mainly because the outer layer of ultrathin bismuthene with abundant dangling bonds can easily bind with oxygen or H 2 O in air, while the inner layer retains the complete crystal structure.The wavelet transform (WT) EXAFS spectra also display the Bi-O, Bi-Bi (intra), and Bi-Bi (inter) coordination in the bismuthine sample (Figure 1i). [35]The X-ray photoelectron spectroscopy (XPS) Bi 4f spectrum after argon sputtering indicates O 2 /H 2 O adsorption on the surface bismuth dangling bonds (Figure S7, Supporting Information). [27,37]In the Bi 4f XPS spectrum, the peaks at 156.8 and 162.2 eV are related to the Bi 0 4f, and those at 158.9 and 164.2 eV are from Bi 3+ 4f. [38]After Ar sputtering, the proportion of metallic state Bi 0 in the inner layer of bismuthene goes up significantly (Figure S7a, Supporting Information), while the proportion of metallic Bi in bulk Bi is less (Figure S7b, Supporting Information), confirming that the bismuthene edge has abundant dangling bonds.
The band structures and photoelectrochemical properties of bismuthene and bulk Bi are determined by theoretical calculation and photoelectron spectroscopy.The density of states (Figure 2a-d and Figure S8a, Supporting Information) and calculated band structure (Figure 2e-h and Figure S8b, Supporting Information) disclose a wide bandgap between the fully electronically occupied full band (B −1 ) and partially electronically occupied intermediate band (IB), which cannot be excited by light to undergo electronic interband transitions. [9,15,16]There is a narrow bandgap between IB and the lowest fully unoccupied empty band (B 1 ), and the bandgap narrows gradually as the number of bismuthene atomic layers increases. [17,39]The one-layer (1L) bismuthene exhibits a direct bandgap of about 0.518 eV around the Fermi level near the Γ point (Figure 2a,e).In the 2 L bismuthene, the B 1 and IB bands cross the Fermi level near the Γ point and overlap in energy albeit separated by a continuous gap (Figure 2b,f).The bandgap of 2 L bismuthene narrows to 92.1 meV near the Γ point.As the number of bismuthene layers increases, 3 and 5 L bismuthene maintain the band structure and semimetallicity similar to 2 L bismuthene, and the bandgap becomes narrower at the Γ point due to the interlayer interaction (Figure 2c,g and Figure S8, Supporting Information).Calculation of the energy band shows that the direct bandgap of bulk Bi is only 9.5-9.7 meV (Figure 2d,h).
Figure 2i shows that the resistance of bismuthene and bulk Bi decreases with increasing temperature indicating semiconduct-ing/semimetallic properties. [40]The resistance of bismuthene is larger than that of bulk Bi throughout the entire temperature range, indicating a wider bandgap for bismuthene.Furthermore, the valence band XPS spectrum shows that the IB band maximum originating from the Bi 6p states passes through the Fermi level (0 eV), verifying the metallic/semimetallic nature of bismuthene and bulk Bi (Figure 2j). [13,16]The temperature dependence of the resistivity and valence-band XPS spectra further confirms the unique semimetallic properties of bismuthene and bulk Bi.The ultraviolet-visible-near-infrared (UVvis-NIR) absorption spectra in Figure 2k show that both bismuthene and bulk Bi have intense absorption from the UV to the IR regions, and the absorption energy gaps are nearly 0.83 and 0.71 eV, respectively (Figure 2l).Ultraviolet photoelectron spectroscopy (UPS) is performed with a monochromatic He I light source (hv = 21.22 eV) to determine the ionization potential (Figure 2m).The onset (E onset ) and cutoff (E cutoff ) energies are 3.39 and 16.61 eV for bismuthene and 2.92 and 16.65 eV for bulk Bi, respectively.The work function (Φ) (vs vacuum) can be determined by the difference between the photon energy and the binding energies of the secondary cutoff edge and the work functions of bismuthene and bulk Bi are 4.61 and 4.57 eV, respectively.The energy of the highest electron-occupied orbital is the sum of the work function and E onset .In semimetallic materials, the energy level of the highest electron-occupied orbital is approximately the lowest energy level of the excited partially electronically occupied intermediate band (IB) electrons (E IB ).According to the relationship between the potential of the vacuum energy (E abs ) and the normal hydrogen electrode (NHE, E•), E• = −E abs − 4.44 (at 298 K), the E IB values of bismuthene and bulk Bi are 3.56 and 3.05 eV, respectively. [16,41]Since the bandgap between B −1 and IB is too wide, the bandgap derived from the UV-vis-NIR diffuse reflectance spectrum is approximately the bandwidth from the Fermi level to B 1 .Based on the above analysis, the schematic band structures of bismuthene and bulk Bi are shown in Figure 2n.Under broad-spectrum illumination, the electrons in the partially electronically occupied IB transition to the Fermi level through intraband transitions, and the excited electrons at the Fermi level can be further excited to B 1 through the interband transitions prior to participating in the CO 2 reduction reaction.
In view of the suitable energy band structure of bismuthene, the photocatalytic CO 2 reduction properties are determined.As shown in Figure 3a, as the Xe lamp illumination time goes up, the CO and CH 4 yields from CO 2 reduction by bismuthene are significantly higher than those by bulk Bi.After Xe lamp illumination for 5 h, the CO yield of bismuthene is 14.32 μmol h −1 g −1 , which is 4.6 times larger than that of bulk Bi (Figure 3b).Bismuthene can further hydrogenate CO 2 to CH 4 (4.69 μmol h −1 g −1 ), whereas bulk Bi cannot reduce CO 2 to CH 4 .The utilization ratio of photogenerated electron reduction CO 2 hydrogenation (formation of CO and CH 4 are 2e and 8e, respectively) for bismuthene with the edge confinement effect is 10.64 times higher than that of bulk Bi.Moreover, bismuthene can convert CO 2 to CO and CH 4 upon irradiation with light in different wavelength ranges ( ≥ 400, 550, or 700 nm) (Figure 3b).Upon exposure to light with wavelengths greater than 400 and 550 nm, the utilization rates of photogenerated electrons of bismuthene are 58.24 and 300.50 times that of bulk Bi, respectively.Under irradiation with near-infrared  light with a wavelength greater than 700 nm, bismuthene can still reduce CO 2 to generate CO and CH 4 , but bulk Bi fails.The CO 2 conversion efficiency of bismuthene exceeds that of many photocatalysts reported in the literature (Table S2, Supporting Information).
Exploration of the photothermal properties of these catalysts reveals that the photothermal temperature of bismuthene is about 131 °C under xenon light irradiation, while the photothermal temperature of bulk Bi is about 118 °C (Figure S9, Supporting Information).The higher photothermal temperature is favorable to the thermally assisted promotion of photocatalytic CO 2 reduction.When the reaction temperature is controlled at 40 °C, the yield of CO 2 reduction to CO decreases to 2.44 μmol h −1 g −1 for bismuthene, while the yield of CH 4 is quite small (Figure 3c-ii).Further analysis reveals that bismuthene cannot catalyze CO 2 reduction when heated to different temperatures without light irradiation (Figure 3c-iii-v) and without the catalyst, the CO 2 reduction reaction does not occur in spite of light irradiation (Figure 3c-iv).These experiments suggest that the driving force for CO 2 reduction is the photogenerated electrons generated by photo-excited bismuthene.In addition, when no H 2 O (Figure 3c-vii) or CO 2 (Figure 3c-viii) is added to the reactor, CO and CH 4 products cannot be generated, indicating that photocatalytic CO 2 reduction requires H 2 O to provide protons for hydrogenation, and the carbon source of the product comes from CO 2 .Interestingly, as the reaction time increases, the O 2 concentration in the reactor increases initially and then decreases (Figure S10, Supporting Information).The initial O 2 in the reactor comes from the input of high-purity CO 2 .In the first hour of photocatalytic CO 2 reduction, the catalyst rapidly oxidizes H 2 O adsorbed on the surface of the materials to form O 2 . [42]Electron spin resonance (ESR) (Figure S11, Supporting Information) performed in an oxygenated environment shows that the generated oxygen or a trace amount of oxygen will react with electrons to generate superoxide radicals (O 2 + e → •O 2 − ) as a competitive reaction to CO 2 reduction (Figure S11a, Supporting Information). [43]With a longer illumination time, the effectiveness of bismuthene photogenerated electrons reacting with O 2 to generate •O 2 − is better than that of bulk Bi.The •O 2 − radicals react with holes to generate singlet oxygen ( 1 O 2 ) and react with H + to generate H 2 O 2 and •OH (Figure S11b-d, Supporting Information). [44]Therefore, the photocatalytic mechanism is unique and complex in the artificial photosynthesis process.It is further verified that the photogenerated holes in IB of the Bibased sample possess the to oxidize H 2 O to generate O 2 .The oxygen production rate of bismuthene (28.05 μmol h −1 g −1 ) is also larger than that of bulk Bi (3.25 μmol h −1 g −1 ) in photocatalytic water splitting oxygen production using AgNO 3 as the photogenerated electron capturing agent (Figure S12, Supporting Information).To further determine the source of the CO 2 photoreduction products, the photocatalytic reaction with isotope-labeled 13 CO 2 and H 2 O 18 is carried out (Figure 3d).The characteristic peaks at 2.2, 2.7, 3.3, and 4.9 min detected by gas chromatography-mass spectrometry (GC-MS) correspond to O 2 , N 2 , CH 4 , and CO.The MS peaks at m/z = 17, 29, and 36 belong to 13 CH 4 , 13 CO, and 18 O 2 , confirming that the 13 CO and 13 CH 4 products coming from 13 CO 2 , and H 2 O 18 can be oxidized to 18 O 2 .The stability of bismuthene is investigated using four (5 h each) consecutive exposures (Figure S13a, Supporting Information).Bismuthene exhibits good photostability, thus showing good potential in long-term photocatalytic applications.The stable XRD patterns of bismuthene after the cyclic photocatalytic experiments corroborate the stability (Figure S13b, Supporting Information).
The intrinsic mechanism and active sites of bismuthene with high photocatalytic CO 2 reduction activity are explored in detail by experiments and theoretical simulation.The CO 2 adsorptiondesorption isotherms of the samples reveal that the adsorptionspecific surface area of bismuthene (8.94 m 2 g −1 ) for CO 2 is larger than that of bulk Bi (3.48 m 2 g −1 ) (Figure 3e).Adsorption of more CO 2 reactants is beneficial to the photocatalytic CO 2 reaction. [45]he transient photocurrent spectra of bismuthene and bulk Bi for different light irradiation ranges show that the photocurrent intensity of bismuthene is stable and significantly stronger than that of bulk Bi (Figure 3f and Figure S14, Supporting Information), meaning that bismuthene can generate more photogenerated electrons to participate in the CO 2 reduction reaction. [46]The photogenerated electron lifetime of these samples is determined on the microscopic scale by single-particle temporally resolved photoluminescence (PL) (Figure 3g-j and Figure S15, Supporting Information).The PL intensity of bismuthene is generally lower than that of bulk Bi (Figure S15, Supporting Information), indicating that the recombination probability of photogenerated electrons in bulk Bi is higher than that in bismuthene.The PL decay profiles (dots) observed at sites 1-4 on bismuthene (Figure 3g) and bulk Bi (Figure 3h) from Figure 3g,h reveal that the photogenerated electron lifetime of bismuthene (6.5-9.3 μs) is significantly longer than that of bulk Bi (4.7-6.1 μs).The PL results are consistent with the photocurrents.
The number of electrons in the outermost shell of a bismuth atom is 5, which is composed of electrons in the 6s2 and 6p3 orbitals.According to the theoretical calculation (Figure 4a) for the charge density contour plots and Figure 4b showing the distributions of charge density of two-layer bismuthene, the electron cloud of the outer layer for bismuth atoms is larger than that of the inner layer.Moreover, the number of outermost elec-trons for the outer bismuth atoms is greater than 5, and the number of outermost electrons for the inner bismuth atoms is less than 5. Hence, the ultrathin bismuthene enriches the surface with more free electrons.To fathom the amplified electric field and charge transport induced by localized surface plasmon resonance, three-dimensional finite-difference time-domain (3D-FDTD) simulations are performed to calculate the spatial electric field distributions of bismuthene surface plasmon resonance under optical excitation (Figure 4c,d and Figure S16, Supporting Information). [47,48]Upon light irradiation (: 400-1000 nm), the photogenerated electrons are mainly concentrated at the edges of the bismuthene nanosheets with a length of 60 nm and thickness of 1.2 nm (red area) (Figure 4c and Figure S16a, Supporting Information).The electric field vector shows that the photogenerated excited state electrons migrate efficiently on bismuthene and accumulate at the edge of the nanosheets due to the edge effect (Figure 4d and Figure S16b, Supporting Information).
The ultrafast femtosecond laser pulse technique is employed to further investigate the dynamics of the photogenerated carriers of bismuthene.According to the 3D plot (wavelength vs time vs intensity) (Figure 4e) and transient absorption (TA) spectra acquired at different time delays (Figure 4f) from bismuthene, the TA spectra of bismuthene in the visible light region exhibit the following features: photoinduced absorption at nearly 470 nm and photoinduced bleaching at nearly 600 nm. [49]The positive signal around the probe wavelength of 470 nm is attributed to the band-filling effect of excited electrons at the B 1 level.The photo-induced excited absorption signals are observed after about 4 ps, indicating that under 400 nm pump excitation, the photogenerated electrons of bismuthene can migrate to the B 1 level by successive intra-band transitions and inter-band transitions.As shown in Figure 4g,h, after excitation by the 400 nm pump, the transient absorption intensity decays exponentially according to the apparent oscillation, exhibiting a coherent phonon behavior within a few picoseconds.The oscillation time of coherent phonons in bulk Bi (Figure 4h) and bismuthine (Figure 4g) decreases gradually from >6 to < 4 ps due to the stronger spinorbit coupling effect of bismuthine. [50]In addition, the oscillation frequencies obtained by fast Fourier transform (FFT) for bismuthene at pumping wavelengths of 400 and 2400 nm in the visible light region are 2.874 and 2.796 THz (Figure S17, Supporting Information), respectively.Compared to the A 1g optical phonon mode of bulk Bi for 2.9-2.96THz, [51,52] the oscillation frequency of bismuthene exhibits a significant redshift.This is mainly due to the reduced size of bismuthene to 2-3 atomic layers.As a result, the interlayer force weakens, the interplanar spacing widens, and the number of crystal interfaces and defects increases. [51]his is consistent with the previous structural characterization results.
The ultrathin edge-rich size effect of bismuthene causes phonon confinement and bond softening.In order to understand the underlying mechanism, fitting of the representative decay curves obtained at 600 nm reveals two components (Figure 4i,j).The lifetime components of  1 and  2 correspond to carrier trapping and recombination, respectively.The lifetime components of bismuthene ( 1 = 150 ps,  2 = 1500 ps) are much longer than that of bulk Bi ( 1 = 10 ps,  2 = 80 ps).The ultrathin edge-rich dangling bond structure of bismuthene extends the lifetime of photogenerated electrons by introducing more defect levels into (the numerical value represents the number of electrons in the outermost orbit, and the unit is e).c) Spatial distribution of the LSPR-induced enhancement of the electric field intensity from FDTD simulations and d) Electric-field vector for bismuthine at 600 nm excitation wavelengths.Femtosecond transient absorption spectra of bismuthine at a pumping wavelength of 400 nm: e) 3D plots; f) Spectra for different time delays; Representative decay curves recorded at g) 590 nm and i) 600 nm; Femtosecond transient absorption spectra of bulk Bi at a pumping wavelength of 400 nm; Decay curves recorded at h) 590 nm and j) 600 nm; Femtosecond transient absorption spectra of bismuthine at a pumping wavelength of 2400 nm: k) 3D plots, l) Spectra for different time delays; m) Decay curves recorded at 650 nm.
the bandgap. [49,51,53]In bulk Bi with very few crystal defects, most of the photogenerated carriers recombine in the bulk phase and few photogenerated electrons migrate to the surface.This is consistent with the photocurrents and photocatalytic performance.
The intraband electronic transition mechanism of the unique partially electronically occupied IB of bismuthene is explored by infrared light (2400 nm) pumping excitation (Figure 4k-m).Compared to the 400 nm pumping excitation (Figure 4e,f), pho-toinduced absorption at nearly 470 nm disappears (Figure 4k,l), indicating the 2400 nm infrared pump does not excite the photogenerated electrons of bismuthene for the interband transition from IB to B 1 .Meanwhile, the photoinduced bleaching region undergoes a redshift from 600 to 650 nm.The decay curves with two components:  1 = 5 ps and  2 = 200 ps are recorded at 650 nm (Figure 4m), revealing that the electrons in the IB of bismuthene can undergo intraband excited transitions under infrared light irradiation, rendering them near the Fermi level ready to receive suitable light energy for further interband transitions.The unique energy band structure and ultra-thin edge-rich confinement effect prolong the relaxation lifetime of bismuthene photogenerated electrons, providing a prerequisite for efficient photocatalytic reactions.
Bismuthene with the ultrathin structure produces bond softening at the edges to form a large number of dangling bonds.[56] With/without light irradiation and heating, the EPR spectra of bulk Bi do not change significantly, indicating that dangling bonds are very few and not affected by light and heat (Figure S18, Supporting Information).Figure 5a,b shows the in situ EPR spectra and corresponding EPR intensity gap of bismuthine, respectively.In the dark, the spin magnetic behavior of bismuthene originates from the unpassivated dangling bonds of edge atoms (Figure 5a-i).Upon light irradiation and heating (Figure 5a-ii-iv), the EPR signals of bismuthine are enhanced indicating that light and heat can activate the edge Bi atoms and enrich free electrons on the dangling bonds.When CO 2 gas is introduced to the device (Figure 5a-v, vi), the intensity of EPR signals diminishes, suggesting that CO 2 adsorbs on the edge bismuth atoms containing dangling bonds, consequently reducing the spin magnetic behavior of dangling bonds.When CH 4 (Figure 5a-vii) and CO (Figure 5a-viii) are respectively introduced into the device, the EPR signal intensity decreases due to the weak adsorption capacity of CH 4 on the edge atoms of bismuthene, while the adsorption capacity of CO on the edge atoms of bismuthene is higher.This is in line with the experimental results that bismuthene can generate CH 4 during photocatalytic CO 2 reduction, but bulk Bi cannot generate CH 4 .
In the photocatalytic CO 2 conversion process, efficient migration and separation of photogenerated carriers are crucial to the photocatalytic activity.Meanwhile, adsorption/activation of CO 2 on the catalyst surface is an important driving force to enhance the CO 2 conversion efficiency.In situ Fourier transform infrared spectroscopy (FTIR), in situ Raman, and density functional theory (DFT) theoretical calculations are performed to study the adsorption/activation of CO 2 molecules and intermediates as well as product desorption.In the in situ FTIR spectra of bismuthene for simulated solar-driven CO 2 reduction (Figure 5c), as the illumination time goes up, the peaks at 1200-1600 cm −1 increase gradually, while those of the intermediate products at 1000-1200 cm −1 weaken gradually.The absorption peaks of bidentate carbonate (b-CO 3 2− ) and monodentate carbonate (m-CO 3 2− ) are 1280 and 1303 cm −1 , respectively. [57]The absorption peaks at 1390 and 1437 cm −1 are assigned to bicarbonate (HCO 3 − ) [45] and those at 1334 and 1556 cm −1 are ascribed to *COOH species, [46] which are important intermediates in photocatalytic reduction of CO 2 to CO. CO can be produced by hydrodehydration of *COOH.The absorption peaks at 1076 and 1140 cm −1 are assigned to *OCH and *OCH 3 , respectively, which are the intermediate products in CO 2 reduction to CH 4 . [58]In the in situ FTIR spectra of bismuthene, as the illumination time increases, the intensity of the *COOH intermediates increases gradually, but those of *OCH and *OCH 3 weaken gradually.Excessive accumulation of *COOH thus inhibits the formation of CO, while the rapid transformation of *CHO and *CH 3 O is ben-eficial to the reaction generating CH 4 .As for bulk Bi (Figure S19, Supporting Information), there are only weak *COOH peaks in the in situ FTIR spectra and it is consistent with the depressed photocatalytic CO 2 conversion.The photocatalytic CO 2 conversion process of bismuthene is further explored by in situ Raman scattering.With the continuous injection of CO 2 gas, the totally symmetric A 1g mode and doubly degenerate E g Raman-active modes at the center of the Brillouin zone of bismuthene redshift gradually (Figure 5d) because of efficient adsorption of CO 2 molecules on the surface of bismuthene.Meanwhile, a series of characteristic peaks of intermediates appear gradually in 1000-3000 cm −1 (Figure 5e).The peak at 2270 cm −1 can be assigned to symmetric stretching of CO 2 (*CO 2 ) of adsorbed CO 2 . [59]The peak at 1590 cm −1 represents the asymmetric stretching of CO 2 − ( as CO 2 − ) of the functional group in the *COOH intermediate. [60]he peaks at 1912 and 2120 cm −1 originate from bridge-bound CO (*CO* bridge ) and atop-bound CO (*CO atop ), respectively. [61]he bridge-bound CO (*CO* bridge ) occupies more active sites and is not conducive to the desorption of CO products.The (CH) vibration band appears at 1300 cm −1 with the bidentate signal, [61] and the peak at 2855 cm −1 stems from C-H stretching of the (C-H) peak of the *OCHO intermediates. [59]The *OCHO intermediates are the first hydrogenation product of CO 2 to CH 4 .The peak at 1444 cm −1 can be assigned to H-C-H bending of the (HCH) peak of the *OCH 2 intermediates, [62] which is the second hydrogenation step of CO 2 to CH 4 .In situ FTIR and Raman scattering show that the intermediates are captured effectively in the photocatalytic CO 2 reduction process.
Bismuthene is a two-dimensional layered material, and its catalytically active sites are mainly at the edge or surface of the layer.The photocatalytic CO 2 reduction mechanism on the surface and edge sites of bismuthene is simulated and determined by theoretical calculation (Figure 5f,g, Tables S3-S6, Supporting Information).The conversion of CO 2 to CO is a 2-electron reduction reaction, and the key step is the formation of the intermediate product *COOH.As shown in Figure 5f-g-iii, the energy barrier of *COOH generation on the bismuthene surface (∆E = 2.575 eV) is higher than that on the bismuthene edge (∆E = 0.446 eV).The rate-determining step of reduction of CO 2 to CO at the edge of bismuthene is desorption of *CO, which needs to overcome the energy barrier of 0.479 eV (Figure 5g-iv) and belongs to the inhalation reaction.Desorption of *CO on the bismuthene surface is an exothermic reaction (Figure 5f-iv).It has been reported that the first hydrogenation step is from *CO to C-H when generating CH 4 . [15,58]In this system, theoretical calculations reveal that the reaction energy barrier for hydrogenation from *CO 2 to the formation of C─H bonds of *OCHO is smaller than that for hydrogenation to form O─H bonds of *COOH (Figure 5f,g-iii).Among them, the energy barrier for hydrogenation of *CO 2 to *OCHO at the edge of bismuthene is only 0.098 eV, which is also smaller than the energy barrier of the generation of *OCHO on the bismuthene surface.During the reduction of CO 2 to CH 4 , the rate-determining steps at the surface and edge of bismuthene are the generation of *OCHO (∆E = 1.584 eV) and *OCHOH (∆E = 0.443 eV), respectively.The abundant unsaturated coordination dangling bonds at the edge of bismuthene increase the hybridization ability of Bi 6p orbital and O 2p orbital, effectively adsorb/activate CO 2 molecules, and adjust and significantly reduce the CO 2 hydrogenation energy barrier.Hence, the CO 2 reduction capability of the bismuthene materials with a rich edge confinement effect is better than that of bulk Bi, and bismuthene can realize CO 2 hydrogenation to generate higher value-added CH 4 products.The conversion of CO 2 to CH 4 is an 8-electron reduction reaction and involves four hydrogenation and two dehydration reactions.Therefore, although the energy barrier of the rate-determining step to generate CH 4 is slightly smaller than that for CO production, the reduction of CO 2 to generate CH 4 and CO proceeds simultaneously.

Conclusion
In conclusion, photogenerated electrons in bismuthene with unique semimetallic properties are transferred from the IB band to the B 1 band under broad-spectrum light irradiation.The bismuthene-rich edge unsaturated coordination dangling bonds introduce trap states in the bandgap to prolong the lifetime of photogenerated electrons and form more excited electrons at the edge.Meanwhile, the bismuthene edge serves as the main adsorption/activation site for CO 2 reduction to reduce the reaction energy barrier for converting CO 2 to CO and CH 4 .Bismuthene with a rich edge-confined structure achieves "Troikas" running hand in hand.It can be excited by broad-spectrum light, prolongs the lifetime of photogenerated carriers, and actively adsorbs/activates CO 2 molecules to realize efficient photocatalytic conversion of CO 2 into high-value carbon-based fuels.This work opens a new avenue for the application of high-performance twodimensional semimetallic materials in the field of photocatalysis.

Figure 1 .
Figure 1.a) XRD patterns and b) Raman scattering spectra of bismuthene and bulk Bi; c-f) HAADF-STEM images of bismuthine; g) XANES spectra and h) Fourier transform (FT) EXAFS spectra of the Bi L 3 -edge of bismuthene, Bi 2 O 3 , and Bi foil; i) Wavelet-transformed (WT) EXAFS spectra of k 3 -weighted of bismuthene.

Figure 3 .
Figure 3. a) Time courses of photocatalytic CO and CH 4 evolutions; b) CO and CH 4 production rates of bismuthene and bulk Bi for 5 h under different light irradiation conditions; c) CO 2 reduction performance of bismuthene under different conditions; d) GC-MS spectra of the products after 13 CO 2 and H 2 O 18 photoreduction for the bismuthine; e) CO 2 adsorption-desorption isotherms of bismuthene and bulk Bi. f) Transient photocurrent spectra of bismuthine (pink) and bulk Bi (blue) under different light irradiation conditions.PL lifetime mapping of g) Bismuthene and h) Bulk Bi with the colors standing for different lifetimes from the single-particle PL measurements during 405 nm pulsed laser irradiation (scale bars in the figure are 4 μm); PL decay profiles (dots) observed at sites 1-4 from g) Bismuthene and h) Bulk Bi shown in (g, h).

Figure 4 .
Figure 4. a) Calculated charge density contour plots, and b) Distributions of charge density at the crystal plane pointed by the arrow in (a) for bismuthine (the numerical value represents the number of electrons in the outermost orbit, and the unit is e).c) Spatial distribution of the LSPR-induced enhancement of the electric field intensity from FDTD simulations and d) Electric-field vector for bismuthine at 600 nm excitation wavelengths.Femtosecond transient absorption spectra of bismuthine at a pumping wavelength of 400 nm: e) 3D plots; f) Spectra for different time delays; Representative decay curves recorded at g) 590 nm and i) 600 nm; Femtosecond transient absorption spectra of bulk Bi at a pumping wavelength of 400 nm; Decay curves recorded at h) 590 nm and j) 600 nm; Femtosecond transient absorption spectra of bismuthine at a pumping wavelength of 2400 nm: k) 3D plots, l) Spectra for different time delays; m) Decay curves recorded at 650 nm.

Figure 5 .
Figure 5. a) In situ EPR spectra and b) EPR intensity gap of bismuthene.c) In situ FTIR spectra of bismuthene for simulated solar-driven CO 2 reduction.d,e) In situ Raman scattering spectra of bismuthene for CO 2 reduction adsorption.Free energy diagrams of CO 2 photoreduction to CO and CH 4 on bismuthine: f) surface and g) edge.