摘要
Open AccessCCS ChemistryRESEARCH ARTICLE11 Mar 2022Engineering Exciton–Phonon Interactions for Suppressing Nonradiative Energy Loss in Energy-Transfer-Initiated Photocatalysis Hui Wang†, Sen Jin†, Aolei Wang†, Xiang Jiang, Wenxiu Liu, Xiaodong Zhang, Wensheng Yan, Kuai Yu, Jin Zhao and Yi Xie Hui Wang† Hefei National Laboratory for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, Anhui 230026 Institute of Energy, Hefei Comprehensive National Science Center, Hefei, Anhui 230031 , Sen Jin† Hefei National Laboratory for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, Anhui 230026 , Aolei Wang† Department of Physics, CAS Key Laboratory of Strongly-Coupled Quantum Matter Physics, International Center for Quantum Design of Functional Materials, University of Science and Technology of China, Hefei, Anhui 230026 , Xiang Jiang Department of Physics, CAS Key Laboratory of Strongly-Coupled Quantum Matter Physics, International Center for Quantum Design of Functional Materials, University of Science and Technology of China, Hefei, Anhui 230026 , Wenxiu Liu Hefei National Laboratory for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, Anhui 230026 , Xiaodong Zhang *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] Hefei National Laboratory for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, Anhui 230026 Institute of Energy, Hefei Comprehensive National Science Center, Hefei, Anhui 230031 , Wensheng Yan National Synchrotron Radiation Laboratory, University of Science and Technology of China, Hefei, Anhui 230029 , Kuai Yu *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] Institute of Microscale Optoelectronics, Shenzhen University, Shenzhen, Guangdong 518060 , Jin Zhao *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] Department of Physics, CAS Key Laboratory of Strongly-Coupled Quantum Matter Physics, International Center for Quantum Design of Functional Materials, University of Science and Technology of China, Hefei, Anhui 230026 Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA 15260 CAS Center for Excellence in Quantum Information and Quantum Physics, University of Science and Technology of China, Hefei, Anhui 230031 and Yi Xie *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] Hefei National Laboratory for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, Anhui 230026 Institute of Energy, Hefei Comprehensive National Science Center, Hefei, Anhui 230031 https://doi.org/10.31635/ccschem.022.202101748 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail The prevalent excitonic effects in low-dimensional semiconductors enable energy-transfer-initiated photocatalytic solar-to-chemical energy conversion. However, the generally strong interactions between excitons and lattice vibrations in these low-dimensional systems lead to robust nonradiative energy loss, which inevitably impedes photocatalytic performance of energy-transfer-initiated reactions. Herein, we highlight the crucial role of engineering exciton–phonon interactions in suppressing nonradiative energy losses in low-dimensional semiconductor-based photocatalysts. By taking bismuth oxybromide (BiOBr) as an example, we demonstrate that phonon engineering could be effectively implemented by introducing Bi–Br vacancy clusters. Based on nonadiabatic molecular dynamics simulations and spectroscopic investigations, we demonstrate that the defective structure can promote exciton–low-frequency phonon coupling and reduce exciton–high-frequency optical phonon coupling. Benefiting from the tailored couplings, nonradiative decay of excitons in defective BiOBr is significantly suppressed, thereby facilitating exciton accumulation and hence energy-transfer-initiated photocatalysis. Download figure Download PowerPoint Introduction Over the past decades, low-dimensional semiconductors have been emerging as competitive candidates for photocatalytic solar energy conversion.1–5 Compared with their bulk counterparts, low-dimensional semiconductors tend to possess promoted excitonic effects owing to their reduced dielectric screening. Accordingly, excitons serving as primary photoinduced species dominate photoexcitation properties of these low-dimensional systems, while exciton-involved energy transfer opens up an alternative pathway for realizing versatile photocatalytic solar energy conversion.6–10 In this case, effective exciton accumulation is an essential precondition for gaining advanced energy-transfer-initiated photocatalysis. However, the general robust nonradiative decay of excitons in low-dimensional semiconductors undoubtedly sets limitations to exciton accumulation and hence to energy-transfer-initiated photocatalysis. Therefore, searching effective strategies for suppressing nonradiative decay of excitons in low-dimensional semiconductors is crucial to the relevant photocatalytic solar energy conversion. Due to structural confinements, phonon-mediated scatterings leading to rapid energy exchange between electronic and vibrational subsystems tend to be quite strong in low-dimensional systems.11–15 These scatterings serve as major factors in determining nonradiative decay of excitons in low-dimensional systems. For instance, inelastic exciton–phonon scattering leads to direct energy dissipation via vibrational relaxation, where high-frequency phonon-mediated interactions lead to faster nonradiative decay of excitons compared with low-frequency phonon-mediated ones; in contrast, elastic exciton–phonon interactions dominate the coherent superposition of exciton states, where the loss of quantum coherence (also known as excitonic dephasing) arising from promoted elastic couplings restrain the nonradiative decay rate.14 Moreover, different types (i.e., acoustic and optical) of phonons further complicate the handling of these effects.15 In this regard, suppressing nonradiative decay of excitons by engineering the strength and type of exciton–phonon interactions in low-dimensional semiconductor-based photocatalysts would be promising and meaningful for gaining optimized energy-transfer-initiated catalytic behaviors. To date, several strategies, such as strain engineering, thickness control, and core–shell design, have been proposed to regulate exciton–phonon interactions in low-dimensional systems.16–22 For instance, exciton–phonon interaction in transition metal dichalcogenide monolayers can be effectively regulated by external strain, giving rise to prominent modifications in excitonic properties like energy profile and decay rate16,17; exciton–phonon interaction in organic–inorganic hybrid perovskites is closely associated with the thicknesses of organic/inorganic layers, which can impact exciton dynamics and hence optoelectronic performance of these systems19,20; as for hybrid systems, shell thickness impacted exciton–phonon interaction in CdSe-CdS core–shell structures, and accordingly, the exciton dynamics and hence the photoluminescence quantum yield can be controlled by varying shell thickness.21,22 However, in terms of photocatalysis, the feasibilities of these strategies remain in doubt, and these strategies tend to give rise to some additional, undesirable effects on electronic and surface properties. Here, we deduce that large-size vacancy clusters bringing about ample lattice displacements or even structural distortions would be suitable to tune exciton–phonon coupling in low-dimensional semiconductors. We focused our attention on bismuth oxybromide (BiOBr), whose low electronic dimensionality arising from confined layered structure endows the system with extraordinarily strong excitonic effects,23,24 and the alternating heterogeneous slabs are anticipated to impact the coupling between electronic and lattice subsystems. Herein, by taking BiOBr as a typical example, we highlight that large-size Bi–Br vacancy clusters can effectively suppress nonradiative decay of excitons involved therein. According to spectroscopic analyses and an ab initio nonadiabatic molecular dynamics method based on GW plus real-time propagation of the Bethe–Salpeter equation (GW + rtBSE-NAMD),25 we elucidated that vacancy clusters can significantly promote exciton–low-frequency phonon coupling and reduce exciton–high-frequency optical phonon coupling. Such modifications control the nonradiative decay channel, thereby giving rise to prolonged exciton lifetimes and enhanced exciton accumulation. Consequently, the defective sample with Bi–Br vacancy clusters (denoted BiOBr-V) exhibits prominent performance in energy-transfer-initiated photocatalysis. This work establishes a brand-new perspective on the regulation of excitonic processes in semiconductor-based photocatalysts via phonon engineering and calls for reevaluation of the potential influence of traditional optimization strategies (like defect engineering) on exciton–phonon interaction in low-dimensional semiconductor-based photocatalysts. Experimental Methods Preparation of BiOBr and BiOBr-V samples BiOBr sample was prepared through a hydrothermal method. In a typical procedure, the stoichiometric precursors (bismuth nitrate pentahydrate, 2 mmol; potassium bromide, 2 mmol) were added into 30 mL of distilled water. After being stirred for 1 h, the above mixture was transferred into an autoclave with Teflon inner lining (50 mL) and heated in an oven at 160 °C for 24 h. After a natural cooling process, the precipitate in the lining was washed with distilled water and ethanol, followed by a drying treatment under ambient conditions. The obtained powders were denoted BiOBr. The defective BiOBr-V was synthesized by annealing the above BiOBr powders in molten alkali-bromide salts. In detail, the mixture of BiOBr (200 mg), potassium bromide (5.5 g), and lithium bromide (4.5 g) were fully ground in an agate mortar using a pestle. The mixture was then transferred into a quartz boat, followed by an annealing treatment (450 °C) under high-purity O2 atmosphere for 12 h. After being cooled, the obtained solid block was dissolved with distilled water, and the precipitate was washed with distilled water and ethanol several times, followed by a drying treatment under ambient conditions. The obtained powders were denoted BiOBr-V. Detecting photocatalytic single oxygen generation 9,10-Diphenylanthracene (DPA) was employed as a molecular probe for detecting photocatalytic singlet oxygen (1O2) generation in the systems. In a typical procedure, a certain amount of catalyst (i.e., 20 mg of BiOBr/BiOBr-V, 0.1 mg of 1- pyrenecarboxylic acid, or their mixture) was added into 50 mL of DPA solution (acetonitrile, 0.06 mM) in a quartz vial equipped with a balloon (to prevent the volatilization of acetonitrile). The mixture was stirred in the dark (30 min) for ensuring adsorption–desorption equilibrium, and then was illuminated by using a visible-light source (xenon lamp, Au-Light CEL-HXF300, equipped with a 420 nm cut-off filter). The DPA consumption was evaluated by monitoring the absorbance evolution of the solution. For atmosphere-dependent tests, the reaction vessel was filled with certain gas (O2 or N2) before illumination. For the scavenger-control test, 5 mg of sodium azide (NaN3) was added into the mixture before illumination. Electron spin resonance (ESR)-trapping tests for detecting reactive oxygen species generation were performed according to our previous report,24 using 2,2,6,6-tetramethylpiperidine (TEMP) as the trapping agent. Blank control groups were also performed, according to the above general procedures in the absence of any photocatalyst (i.e., BiOBr and BiOBr-V) or co-catalyst [i.e., 1-pyrenecarboxylic acid (PCA)], to exclude the interference of photoinstability or impurity of DPA and TEMP. Photocatalytic aerobic oxidation of secondary amines In a typical reaction, 0.25 mmol of amine, 0.4 μmol of PCA, and 5 mg of BiOBr/BiOBr-V were added into 5 mL of acetonitrile in a 20 mL quartz vial equipped with a balloon. A xenon lamp (CEL-HXF300, Au-Light) equipped with a 420 nm cut-off filter was employed as the light source. The performance was evaluated by nuclear magnetic resonance spectroscopy, using 1,1,2,2-tetrachloroethane as the internal standard. For the scavenger-control tests, 10 mg of NaN3 was added into the mixture before illumination. Results and Discussion Structural characterizations of the samples The defective sample (denoted BiOBr-V) was prepared by annealing pristine BiOBr in molten eutectic mixtures (see details in the Supporting Information), during which the release of molecular BiBr3 led to the formation of large-size Bi–Br vacancy clusters. The X-ray diffraction (XRD; Figure 1a) patterns of BiOBr and BiOBr-V match well the reported data (JCPDS card No. 73-2061). The X-ray photoelectron spectra (XPS; Supporting Information Figure S1) verify the components of the samples to be Bi, O, and Br. Note that quantitative analysis of sample composition was challenging due to the complicated surface structures like surface atom deficiency, dangling bond (e.g., hydrogen bond and hydroxyl group), and water adsorption. The scanning electron microscopy (SEM; Supporting Information Figures S2a and S2b) images confirmed the similar, plate-like morphology of both the samples. In the BiOBr-V sample, some small dots, which are broken fragments generated during the annealing procedure, attached on the nanoplates were observed ( Supporting Information Figure S2c). The similar morphology was also evidenced by their approximate specific surface areas, while the promoted proportion of nanometer-scale pores would be associated with the presence of vacancy clusters in BiOBr-V ( Supporting Information Figure S3). X-ray absorption spectroscopy (XAS; Figure 1b) was employed to identify the orbital hybridizations in the two samples: the peaks (centered at ∼527.0 and 529.2 eV) originating from O 2p and Bi 6p hybridization exhibited near-identical intensity ratios, which suggests the similar valence states in the two cases; however, BiOBr-V exhibited remarkable spectral broadening (inset of Figure 1b) compared with BiOBr, which would be related to thermal vibrations and static disorder induced by the defective structures.26,27 In addition, ambient-atmosphere ESR measurements confirmed the presence of oxygen vacancy in both samples ( Supporting Information Figure S4). Figure 1 | (a) XRD patterns, (b) XAS spectra (inset displays the magnified version of the dash-lined box), and (c) Raman spectra of BiOBr and BiOBr-V. (d) HAADF–STEM image of BiOBr-V. Download figure Download PowerPoint The Raman spectra for investigating the vibrational scenarios are displayed in Figure 1c. As for BiOBr, typical vibrational bands peaking at ∼91, 111, and 160 cm−1 were assigned to Eg, A1g internal stretching, and A1g external stretching modes, respectively.28 However, quite different vibrational scenarios were found in the BiOBr-V sample: the band centered at 111 cm−1 remained unchanged, whereas the other two bands exhibited obvious redshifts and greater intensities. Since the major contributor to the 111 cm−1 band is the Bi–O bond and the major contributors of the other two bands are the Bi–Br and Br–Br bonds,28 the distinct vibrational scenario in BiOBr-V most likely originated from structural modifications related to Bi and Br atoms. The high-angle annular dark-field scanning transmission electron microscopy (HAADF–STEM) was used to gather direct evidences of defective structures in BiOBr-V. As displayed in Figure 1d, the atomic structure image of BiOBr-V presents a distance of ∼0.28 nm that corresponds to the atomic arrangements of plates with (001) exposed facets. Compared with the uniformly distributed bright spots in BiOBr ( Supporting Information Figure S5), the deficient sites and heterogeneous contrasts (marked by green and red dashed circles in Figure 1d, respectively) in BiOBr-V directly verify the presence of Bi–Br vacancy clusters. In detail, the deficient sites (marked by green dashed circle) suggests the concurrent absences of Bi and Br atoms along the [001] axis; the heterogeneous contrasts (marked by red dashed circle) suggest the inhomogeneous distribution of adjacent Bi and Br atoms (detailed discussions are provided in Supporting Information Note 1 and Figure S6). The above results confirmed that BiOBr plates containing rich Bi–Br vacancy clusters were successfully prepared. Nonadiabatic molecular dynamics simulations In view of the unique structural and vibrational scenarios arising from Bi–Br vacancy clusters, we carried out first-principle calculations to gain theoretical insights of the impacts of Bi–Br vacancy clusters on exciton dynamics. The ground-state calculations suggest that the two systems possess similar electronic structures, and the Bi–Br vacancy clusters do not introduce deep-level trap states (see details in Supporting Information Note 2 and Figures S7 and S8). These features enabled us to exclude the potential effects of defect-state-induced nonradiative recombination of excitons. With regard to the impacts of exciton–phonon interactions, we further interrogated exciton dynamics of the pristine and defective systems at a finite temperature by using the GW + rtBSE-NAMD method (see details in Supporting Information Note 3 and Figure S9). We plotted the time evolutions of the lowest exciton state's energies and exciton binding energies in the two systems at 300 K (Figure 2a). Because of the comparable exciton energies and exciton binding energies, we confirmed that Bi–Br vacancy clusters do not impact the intrinsically strong excitonic effects in the BiOBr structure. The calculated exciton energies are larger than the values estimated by photoluminescence measurements in our previous reports,23,24 which might be associated with the different dielectric environments. Additionally, the lowest-energy exciton states exhibit notable time-dependent energy fluctuations, which reveal remarkable exciton–phonon interactions in both the pristine and defective systems. Figure 2 | (a) Time evolutions of the lowest exciton state's energies (upper panel) and exciton binding energies (lower panel). (b) Fourier transforms of the relevant exciton-energy fluctuations. Time-dependent exciton energy changes of (c) pristine (BiOBr) and (d) defective (BiOBr-V) systems at 300 K. Download figure Download PowerPoint To further understand the exciton–phonon interactions, we carried out Fourier transform analyses of the relevant exciton-energy fluctuations (Figure 2b). The Fourier transform of the pristine BiOBr case displays a dominant peak centered at ∼117 cm–1, which was assigned to the A1g internal stretching mode. Moreover, there is another prominent, high-frequency phonon mode peak around ∼800 cm−1, which might be associated with the superposition of Eg internal shear and B1g internal stretching modes.28 As for the defective system, a quite different scenario was observed: the highest peak was still located at ∼117 cm−1, whereas there was no obvious peak around ∼800 cm−1; furthermore, two distinct peaks centered at ∼67 and 433 cm−1 emerged, which correspond to two different Eg modes that were mainly related to the shear vibrations of intralayer oxygen atoms and interlayer bismuth atoms, respectively. Clearly, Bi–Br vacancy clusters notably modulated the couplings between excitons and phonons, promoting exciton–low-frequency phonon coupling but suppressing exciton–high-frequency phonon coupling. With respect to exciton dynamics, the vanishing of exciton–high-frequency phonon coupling would effectively slow down exciton depopulation.29,30 Time-dependent exciton energy changes deduced from the population of certain excitons are provided in Figures 2c and 2d. The average exciton lifetimes were calculated to be ∼249 and 415 ps for the pristine and defective systems, respectively, which is consistent with the averaged non-adiabatic coupling relationships in Supporting Information Figure S10. Based on the GW + rtBSE-NAMD simulations, we deduced that prolonged exciton lifetime and enhance exciton accumulation arising from the promoted exciton–low-frequency optical phonon coupling and suppressed exciton–high-frequency optical phonon coupling could be expected in the defective BiOBr-V system. Photoluminescence properties of the samples Photoluminescence measurements were performed to gain experimental insights into the impacts of Bi–Br vacancy clusters on exciton–phonon coupling and hence on exciton dynamics. Excitation wavelength in all the photoluminescence measurements was set at 350 nm, based on the UV–vis spectra ( Supporting Information Figure S11). Figure 3a displays the steady-state fluorescence spectra at 10 K. The pristine BiOBr sample exhibited a prominent sub-band emission peaking at ∼480 nm, which is in accordance with the excitonic emission of BiOBr system.23,24 The weak shoulder observed at ∼565 nm was associated with the slight oxygen vacancies in the sample. In comparison, the defective BiOBr-V sample exhibited a high-intensity, broad emission centered at ∼492 nm. Notably, the ∼12 nm redshift (i.e., 480 vs 492 nm) was ascribed to the distinct variation tendencies (in both energy level and proportion) of various emissions; details are provided in Supporting Information Note 4 and Figure S12. The broadening spectral profile could be related to the promoted exciton–phonon coupling in the defective sample.31,32 In addition, the potential impacts of broken fragments generated during the annealing procedure on the excited-state properties of BiOBr-V sample were excluded (see details in Supporting Information Note 5 and Figure S13). Figure 3 | (a) Steady-state fluorescence spectra (10 K). (b) Temperature-dependent fluorescence spectra of BiOBr-V. (c) Peak energies of the extracted short-lived excitonic emissions of BiOBr and BiOBr-V under different temperatures, where fitting results using Eq. 1 are presented in dash lines. (d) Time-dependent fluorescence spectra (monitoring at 470 nm). Download figure Download PowerPoint The remarkable changes in exciton–phonon coupling induced by Bi–Br vacancy clusters were also evidenced by temperature-dependent photoluminescence tests (from 10 to 100 K, with a temperature interval of 10 K). As for the pristine BiOBr case ( Supporting Information Figure S14), the intensity of the 480 nm excitonic emission dramatically decreased with increasing temperature, which resulted in a high proportion of oxygen-vacancy-related emission (i.e., 565 nm emission) under high-temperature conditions. We attributed remarkable suppression of the 480 nm excitonic emission to the promoted phonon-mediated nonradiative decay of excitons under a high In comparison, the emission peak of BiOBr-V (Figure exhibited a notable redshift with increasing temperature, which suggests the different exciton–phonon coupling scenarios in the To a quantitative on the of exciton–phonon coupling in the two we plotted the peak energy of short-lived excitonic emission extracted from the spectral for both BiOBr and BiOBr-V (Figure and Supporting Information Note 4 and Figure A was employed to the temperature-dependent peak ( 1 + 2 ( 1 , is the emission peak energy at temperature is to the strength of exciton–phonon and is the effective phonon energy on temperature Compared with BiOBr, BiOBr-V a (i.e., vs in the promoted exciton–phonon coupling strength in the defective system. However, a is that the temperature-dependent of emission intensity of BiOBr-V was than that of BiOBr promoted exciton–phonon couplings lead to faster in other Moreover, photoluminescence tests also fluorescence spectra monitoring at 470 nm to the short-lived excitonic emission Figure that BiOBr-V a excitonic decay than BiOBr. Time-dependent measurements also confirmed the lifetime of excitonic emission in BiOBr-V ( Supporting Information Figure with the theoretical simulations, the features are as Bi–Br vacancy clusters the couplings between excitons and low-frequency optical phonons and suppress the couplings between excitons and high-frequency optical the be for the broadening and temperature-dependent peak of excitonic while the is for the changes in intensity and lifetimes of excitonic The above results confirmed that that Bi–Br vacancy clusters effectively the exciton–phonon couplings and the exciton dynamics in the system. Photocatalytic performance the of the optimized exciton dynamics, the defective BiOBr-V sample was expected to be to photocatalysis. To verify we evaluated the of photocatalytic effective of both samples. Due to the energy absorption and hence in the light (Figure However, the transfer of excitons from semiconductors to ground-state generation of Therefore, we visible-light photocatalytic generation in the system. Here, we used DPA to probe generation in the presence of different which were by monitoring the absorbance evolution of DPA solution. Figure displays the absorbance evolutions of DPA solution in the presence of and exhibited in generation than Note that generation was also for BiOBr and BiOBr-V (i.e., in the absence of Figure whereas the energy between excitonic states of the systems and states of molecular oxygen were for the tests under and O2 (inset of Figure confirmed that the reactive species for DPA consumption were from molecular The of NaN3 of can effectively suppress DPA the generation in the system. Moreover, ESR measurements also confirmed the relevant using as the spin agent. As displayed in Figure both the and exhibit typical structures to whereas the intensity the enhanced generation in The above results the energy-transfer-initiated photocatalytic performance of the defective BiOBr-V sample, which the role of Bi–Br vacancy clusters in suppressing nonradiative decay of excitons involved therein. Figure 4 | (a) UV–vis and photoluminescence spectra of and BiOBr-V. (b) of the absorption spectra of DPA solution in the presence of and in (c) Photocatalytic DPA under different conditions. (d) ESR spectra of