English

• Oxidation process can be found everywhere in nature, having great influences in chemical changes of cell growing, life activities and industrial production [1]. No matter what kind of chemical changes, reactive oxygen species (ROS) are the core factor of oxidation process [1,2]. ROS directly modulate the redox state of key regulatory proteins, influencing the RNA transcriptional and translational machinery in plant and animals [1,3]. In industrial production, ROS attack chemical reactants with high efficiency, bringing high value-added oxidized products [2,4]. By optimizing reaction conditions and minimizing by-products, ROS allows for more sustainable manufacturing processes, ultimately leading to increased capacity and reduced operational costs [5,6].

  In catalytic reactions, ROS are mainly generated from the activation of surface lattice oxygen and the chemisorbed molecular oxygen on catalyst surfaces [7-9]. Currently, it is widely accepted that ROS consist of monatomic oxygen ions (O^-), superoxide (O₂^-), peroxide (O₂^2-), singlet oxygen (¹O₂) and active surface lattice oxygen species [5,7,10,11]. O^-, O₂^-, and O₂^2- usually generated from the electron transfer processes with O₂ preferentially adsorbed on the surface of catalysts, which can enhance catalytic oxidation activity by directly interacting with adsorbed reactant molecules in various reactions [5,10-13]. Highly reactive ¹O₂ can be produced from energy transfer process between the excited photosensitizer and the adsorbed O₂, which could significantly enhance the photocatalytic efficiency under visible light [14]. While, surface lattice oxygen can be effectively activated through the incorporation of high-valence metal sites [15,16]. These high-valence metal sites can introduce holes into oxygen ligands, thereby facilitating the coupling of intramolecular oxygen and promoting the activation of lattice oxygen [15,16]. However, reports on new ROS other than the aforementioned type are scarce. The discovery of new ROS may show significant potential to enhance oxidation efficiency with a broader range of chemical reactions.

  Herein, we offer direct evidence of a new ROS defined as lattice-superoxide species (LSO), which was generated on oxygen vacancy rich (010) facets of ultrathin CuO nanosheets (CuO-O_v). In situ near-ambient pressure X-ray photoelectron spectroscopy (in situ NAP-XPS), electron paramagnetic resonance (EPR), density functional theory (DFT) calculations and ab initio molecular dynamics (AIMD) results reveal that the LSO possesses both the activity of superoxide species and the stability of lattice oxygen species. Extended X-ray absorption fine structure (EXAFS) measurements and DFT calculations suggest that the LSO has a side-on oxygen adsorption form with a comparable electron (bader charge) distribution between two oxygen atoms (O_I 0.22 eV vs. O_II 0.20 eV). The electron transfer from oxygen vacancies to two oxygen atoms leading to a weak O-O bond. Propene and toluene catalytic oxidation reactions were utilized as probe reactions to investigate the O₂ activation mechanism. The O-O bond in LSO is further identified as new active sites for propene adsorption. The LSO shows an appropriate activation energy in between superoxide species and lattice oxygen, resulting in a considerably improved oxidation activity for C₃H₆ and aromatic VOCs oxidation reactions. The unique charge property of LSO endows it with both high activity and stability, establishing a novel strategy for designing reactive oxygen species. This versatile strategy opens new avenues for efficient oxidation across both chemical and biological contexts.

  The ultrathin CuO nanosheets were prepared through a mild solution method. The construction of oxygen vacancy (CuO-O_v) was achieved by reducing the as-prepared CuO nanosheets in NaBH₄ solution (Fig. S1 in Supporting information). As shown in X-ray diffraction (XRD) results (Fig. S2 in Supporting information), both CuO and CuO-O_v exhibit the characteristic diffraction lines of CuO structure (JCPDS 48–1548). The morphology of the catalysts was examined by scanning electron microscopy (SEM) and transmission electron microscope (TEM) as shown in Figs. S3 and S4 (Supporting information). Both CuO and CuO-O_v display the nanosheets structure with clear edges which width ranges from 300 nm to 500 nm. Atomic force microscopic (AFM) was performed to characterize the thickness of the CuO and CuO-O_v nanosheets, which is both about 4.0 nm (Fig. S5 in Supporting information). We observed from angular bright-field STEM (ABF-STEM) and high-resolution transmission electron microscope (HRTEM), the (010) facets are the most exposed crystal facets of CuO and CuO-O_v (Fig. 1a and Fig. S7 in Supporting information). In addition, there are substantial uneven area appears on CuO-O_v surface, indicating the existence of defective structures (pink circles in Fig. S6d in Supporting information) [17]. The deficient structure of CuO-O_v can be further described by the intensity of oxygen atoms (Figs. 1a-c and Fig. S7) [18]. As interpreted from Fig. 1a (the green dash line), the consecutively equivalent intensity of oxygen atoms indicates there are no oxygen vacancies (Fig. 1b). In contrast, as shown in Fig. 1c, the absence of oxygen atom is differentiated by the various intensities between oxygen atoms (interpreted from yellow dash line in Fig. 1a).

  Figure 1

  

  Figure 1.  (a) ABF-STEM image of CuO-O_v. (b, c) Intensity profile corresponding to the green and yellow dashed line in (a), respectively. (d) EPR pattern under vacuum. (e) Structure model of CuO-O_v nanosheet. (f) O₂-TPD profiles of CuO and CuO-O_v. (g) Oxygen K-edge spectra of the CuO and CuO-O_v.

  DownLoad: Full-Size Img PowerPoint

  Surface defects on CuO-O_v were also investigated by electron paramagnetic resonance (EPR), X-ray absorption near edge structure (XANES) spectra and extended X-ray absorption fine structure (EXAFS). As shown in Fig. 1d, CuO-O_v exhibits a typical oxygen vacancy signal with g = 2.003 [19]. The much stronger intensity of EPR signal suggests much larger amount of oxygen vacancies on CuO-O_v than pristine CuO nanosheets. Fig. S8 (Supporting information) presents the XANES spectra at the Cu K-edge of the CuO and CuO-O_v samples. The pre-edge feature of CuO-O_v sample shifted to lower energy, which is between Cu₂O and CuO standards. This result indicates the coordination state of Cu in CuO-O_v sample has a minor change due to the generation of oxygen vacancies during NaBH₄ reduction, which is supported by the results in normalized Cu L-edge XANES spectra (Fig. S9 in Supporting information) and XPS results (Fig. S10 in Supporting information) [4,20,21]. Simultaneously, as depicted in the Cu K-edges EXAFS spectra, the peaks at 1.95 Å and 1.93 Å corresponding to the Cu-O scattering path of CuO and CuO-O_v, respectively (Fig. S11a and Table S1 in Supporting information) [4,22,23]. The shift of the peak position and the decrease of the coordination number in Cu-O path indicates more unsaturated Cu atoms and the absence of O atoms in CuO-O_v than CuO sample. In addition, as shown in Fig. S11b (Supporting information), the reduced amplitude of the peak in Cu-O confirm again that oxygen deficient CuO nanosheets were successfully fabricated (CuO-O_v) [24]. On the basis of experimental observations (Fig. 1a and Fig. S6d in Supporting information), the theoretical model of CuO (010) and CuO-O_v (010) facets were established (Fig. 1e and Fig. S12 in Supporting information).

  As an anion defect, oxygen vacancies can effectively activate O₂ by transferring electrons to O₂ antibonding π* orbital. The possible activated oxygen species formed in these vacancy sites will be further validated by O₂-temperature programmed desorption (O₂-TPD) and the O K-edge X-ray absorption near-edge structure (XANES) spectra. As shown in O₂-TPD (Fig. 1f), peaks at 207 ℃ and 227 ℃ can be assigned to the desorption of surface adsorbed oxygen in CuO and CuO-O_v, respectively [10,19]. Desorption peaks were observed around 150 ℃ (142 and 151 ℃) for both samples, which may correspond to chemisorbed oxygen species at oxygen vacancies that inevitably generated from preparation process. An extra desorption peak is noticed of CuO-O_v catalyst appears at 115 ℃, which may be attributed to a new chemisorbed oxygen species at oxygen vacancy site. To further confirm the new ROS formed on CuO-O_v, the O K-edge XANES spectra of CuO-O_v was employed. As displayed in Fig. 1g, the peaks at 530.0 eV are due to the hybridization of O 2p and Cu 3d orbitals. The peaks at 534.6 and 538.9 eV are formed by the hybridized orbitals between O 2p and Cu 4 s, O 2p and Cu 4p, respectively [22,23]. All these peaks are formed by the hybridized orbitals of lattice oxygen and Cu atoms on CuO surface (Cu-O bond). However, as for the CuO-O_v sample, a distinctly new peak at 532.0 eV appears, which does not match either O or Cu orbital transition compared to CuO and CuO₂ control sample [22,23]. It is deduced that a new adsorbed oxygen species generated on CuO-O_v, which is similar to that of superoxide species (O₂^-) [2,25-27].

  DFT calculations were performed to elucidate the formation of new ROS on CuO-O_v. Simulation results elucidate that O₂ adsorption on oxygen vacancy of CuO-O_v (010) may undergo side-on and end-on paths (Fig. 2a and Fig. S13 in Supporting information). As shown in Fig. S13a (Supporting information), end-on adsorbed oxygen species are formed on the CuO-O_v surface with formation energy of -1.57 eV and O-O bond length of 1.29 Å, which is consistent with the value of superoxide species in the previous publication [28,29]. As shown in Fig. 2a, side-on adsorbed oxygen species is formed with formation energy of -1.83 eV and the O-O bond length of 1.30 Å, which is also belong to superoxide species [28,29]. Comparing the formation energy (-1.57 eV vs. -1.83 eV), the generation of O₂ adsorption with side-on mode is preferable, which is the primary active oxygen species generated on CuO-O_v. In this mode, when O₂ molecule approaching the CuO-O_v surface (Fig. 2a), the oxygen atoms (O_I and O_II) of adsorbed O₂ would bind with two Cu atoms close to oxygen vacancy site. Then, the Cu-O-O-Cu hybridization structure is generated. Bader charge analysis suggests that O_I and O_II gain 0.22 and 0.20 electrons from Cu atoms, respectively. As shown in Fig. 2b, the O₂ adsorb on CuO surface without oxygen vacancies, only obtain 0.11 and 0.09 electrons from lattice oxygen atom. The O-O bond length is only 1.25 Å. Above results suggest that the oxygen vacancy can induce more electrons transfer from Cu atoms to adsorbed O₂. Based on the DFT calculations illustrated in Figs. 2a and b), the superoxide species on CuO-O_v surface is located upon the oxygen vacancy site, taking the place of one surface lattice oxygen atom. Consequently, it is rational to speculate that the superoxide species on CuO-O_v surface may possesses the properties of both superoxide and lattice oxygen, which were provisionally labeled as lattice-superoxide (LSO). For a comprehensive understanding of LSO, the chemisorbed oxygen species generated on other facets ((110) and (111) facets) of CuO-O_v was studied by DFT. As shown in Fig. S14 (Supporting information), the oxygen species adsorb on CuO-O_v (110) is an end-on mode with formation energy of -1.56 eV, and the O-O bond length is 1.31 Å, which is a superoxide species. For CuO-O_v (111), the formation energy of chemisorbed oxygen species is -1.99 eV with a side-on mode, which is a peroxide species. The distinct reactive oxygen species on different CuO-Ov facets have been further confirmed by O₂-TPD (Fig. S15 in Supporting information). According to the Bader charge analysis of the different crystal facets (Fig. S14 in Supporting information), the peroxo species (O₂^2-) formed on the (111) face mediates a stronger electron transfer, compared to the superoxide species (O₂^-) formed on the (110) and (010) faces. Based on the above results, LSO can only be generated on the CuO-O_v (010), which is a new type of ROS for CuO. Forthermore, in situ near-ambient pressure XPS (NAP-XPS), EPR, integral crystal orbital Hamilton populations (-iCOHP), electronic local function (ELF) and ab initio molecular dynamics (AIMD) simulation were employed to fully elucidate the unique properties and behaviors of this proposed LSO.

  Figure 2

  

  Figure 2.  (a) The configuration of O₂ and the difference charge density of O₂ adsorbed on CuO-O_v surface through theoretical calculation. (b) The configuration of O₂ and the difference charge density of O₂ adsorbed on CuO surface through theoretical calculation. (c) NAP-XPS of O 1s spectra from CuO-O_v with the Ar etching. The arrows represent the order in which the catalyst is treated with O₂ or Ar etching. (d) NAP-XPS of O 1s spectra from CuO with the Ar etching. (e) Crystal orbital hamilton population (COHP) of O-O bond in LSO and O₂^- species. (f) O K-edge XANES of CuO-O_v with different crystal facets. (g) The electronic local function (ELF) of LSO. (h) The electronic local function (ELF) of oxygen adsorbed on CuO. (i) COHP of Cu-O bond (Cu and O_I atom in LSO, Cu and lattice oxygen in CuO). (j) Ab initio molecular dynamics (AIMD) simulation for LSO desorption on CuO at 473 K.

  DownLoad: Full-Size Img PowerPoint

  Figs. 2c and d show the NAP-XPS under different atmospheric conditions, which was conducted to study the properties of the LSO on CuO-O_v surface. Before the NAP-XPS test, all the samples were pretreated at 300 ℃ under N₂ to eliminate the influence of external oxygen. Then, all the synthesized catalysts were treated by O₂ atmosphere and switch to argon-ion-cluster etching under UHV in turn for two more times. For both samples after pretreatment, the peaks located at 529.9 eV are assigned to lattice oxygen (O_latt), while the peaks at 531.4 eV are attributed to surface-adsorbed oxygen (O_ads) [30,31]. Subsequently, when switched to O₂ atmosphere for the first time, a new peak appears at 530.7 eV between peaks of O_ads and O_latt, which can be assigned to the LSO on CuO-O_v. The peak of LSO is different from normal adsorbed oxygen (531.4-532.8 eV) or lattice oxygen (529.3-530.5 eV) in previous literatures [31-33]. After argon-ion-cluster etching under UHV for the first time (Fig. 2c and Tables S2 and S3 in Supporting information), the content of O_ads decreased and no peak at 530.7 eV appears, indicating that the adsorption capacity to oxygen mainly possessed by the surface vacancies of CuO-O_v sample. No LSO could be formed in the subsurface or bulk lattice of the CuO-O_v catalyst. As for CuO catalyst (Fig. 2d), no LSO can be observed during all test conditions, indicating that oxygen vacancy is the key factor for the formation of the LSO. Therefore, only the surface oxygen vacancy of CuO-O_v catalyst is able to induce the formation of LSO, which are different from reported ordinary adsorbed oxygen species. In addition, as shown in Table S4 (Supporting information), the binding energy (BE) of adsorbed oxygen is generally at about 531.4–531.6 eV. However, the BE of LSO are at 530.7 eV, which indicate that the LSO is a special ROS.

  More specifically, EPR result was carried out with 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) as an element spin-trapping agent to elucidate the properties of the LSO generated on CuO-O_v [29]. In Fig. S16 (Supporting information), six characteristic peaks with different intensities correspond to DMPO-OOH is resulting from trapping superoxide [34]. Appreciably, the signal intensity of CuO-O_v is much higher than CuO, indicating the LSO is easier to generate on CuO-O_v surface, which is consistent with the DFT result (Fig. 2a). The -iCOHP (integral crystal orbital Hamilton populations) of the LSO adsorbed on CuO-O_v and other typical ROS were also calculated (Fig. 2e and Fig. S17 in Supporting information) to uncover the strength of O–O bond. Fig. 2e is the -iCOHP of O-O bond in LSO and ordinary superoxide species (O₂^-), which are 12.45 eV and 15.37 eV, respectively. The smaller value indicates that the O-O bond in LSO is easier to be activated compared to ordinary O₂^- species in catalytic oxidation reaction [4]. In addition, the -iCOHP of O-O bond in gas phase O₂ (16.10 eV) and O₂^2- (11.17 eV) were also calculated (Fig. S17 in Supporting information). The O-O bond in LSO is weaker than O₂ or O₂^-, but a little higher than end-on adsorbed superoxide (12.44 eV) and O₂^2-. A large -iCOHP is not conducive to the break of O-O bond. While if the -iCOHP is too small, the oxygen species is unstable to participate in the reaction [35]. Moreover, the O-O bond of the LSO has been elongated to 1.30 Å, which indicates that the LSO on CuO-O_v is easier to break compare to the adsorbed O₂ on CuO. To further verify the uniqueness of the adsorbed oxygen species generated on the (010) crystal facet, CuO-O_v (111) and CuO-O_v (110) comparative facets were successfully prepared as control samples (Fig. S18 in Supporting information). O K-edge XANES was employed to probe the electronic structure of adsorbed oxygen species on each facet. As shown in Fig. 2f, no discernible peak was observed on CuO-O_v (111) in the range of 531.0–533.0 eV. A peak belongs to superoxide (O₂^-, 532.2 eV) can be observed over CuO-O_v (110). Whereas a 0.2 eV negative shift compare to (110) facet with concomitantly increased intensity can be observed over CuO-O_v (010). These spectral features demonstrate that the lattice superoxide (LSO) generated exclusively on CuO-O_v (010), fundamentally differs from conventional peroxo/superoxide species (Table S5 in Supporting information), exhibiting enhanced bond cleavage and electronic excitation. The bonding of LSO has also been calculated by electron localization function (ELF charge). As assembled in Figs. 2g and h, compared with the adsorbed oxygen species on CuO surface, the electron range of O atom in LSO on CuO-O_v is larger, and the ELF value of O atom in LSO is about 0.5, indicating the weak electron localization and weak covalent feature of O-O interaction in LSO [36]. This result manifests the electrons in LSO are metastable due to free electron gas state, which suggests that the O-O bond in LSO on CuO-O_v is easier to be activated. The ELF is consistent with the -iCOHP discussed above. In order to investigate the specific properties of the LSO on CuO-O_v, the electronic properties and other characteristics were compared between LSO and other reported ROS species. As shown in Table 1, the bond length, adsorbed energy, Barge charge, and electron distribution of the LSO in our work are quite different from other ROS [28,37-40]. To summarize, all the values of the LSO are in relatively moderate state than other ROS. The comparable electric charge of O_I and O_II in LSO, rendering a relatively stable electronic structure between it and catalyst surface, which is similar to the properties of lattice oxygen.

  Table 1

  

  Table 1.  The calculated change in Bader charge on CuO and CuO-O_v. And the comparison of different ROS in references.

  DownLoad: CSV

  ROS	Bond length (Å)	Adsorbed energy (eV)	Barder charge (eV)	Ref.
  LSO	1.30	-1.83	OI: 0.22 OII: 0.20	This work
  Peroxide O22-	1.36	-1.99	OI: 0.30 OII: 0.39	This work
  Superoxide O2-	1.31	-1.56	OI: 0.14 OII: 0.39	This work
  Superoxide O2-	1.32	-1.52	-	[37]
  Peroxide O22-	1.43	-2.80	-	[28]
  O-	1.41	-2.71	0.646	[38]
  O2-Pt-H2WO4	1.37	-1.74	0.22	[39]
  Fe(O=O)Fe	-	-2.00	OI: 0.06 OII: 0.37	[40]

  The -iCOHP values of Cu-O bonds were further calculated. As shown in Fig. 2i, the strength of Cu-O bond in LSO (-iCOHP = 1.90) is similar to Cu-O bond that between Cu atom and lattice oxygen (iCOHP = 2.20) in CuO-O_v sample. This result indicates that the O atom in LSO not only have the similar stability to lattice oxygen, but also have the active properties similar to adsorbed oxygen. Ab initio molecular dynamics (AIMD) simulation was used to observe the LSO behavior on CuO-O_v surface. Fig. 2j is the snapshots trajectory of the LSO on CuO-O_v during 1000 fs at 473 K. After 1000 fs simulations, the LSO remain adsorbed with the initial state on CuO-O_v surface, which reveals that the LSO are relatively stable. The LSO on CuO-O_v surface occupy the position that previously belongs to surface lattice oxygen. After accepting extra electrons from two neighboring Cu atoms, the LSO exhibit both the high activity of superoxide species and the stability of lattice oxygen, which may play versatile roles in the catalytic oxidation reactions.

  The catalytic oxidation of C₃H₆ was first used as a probe reaction to study the superiority of LSO species. As shown in Fig. 3a, CuO-O_v nanosheets exhibit a decent activity with T_90% = 229 ℃ (the temperature of 90% conversion), which is 90 ℃ lower than that of CuO catalysts (312 ℃) and commercial CuO (Fig. S19 in Supporting information). It also shows good activity compared to other metal oxides in literature (Table S6 in Supporting information). The corresponding Arrhenius plot of CuO and CuO-O_v catalysts were also evaluated (Fig. 3b). The activation energy for CuO-O_v catalyst determined from the slope of the curve is about 39 kJ/mol, which is much lower than CuO (120 kJ/mol). This result indicates that the active sites and reaction mechanism between CuO and CuO-O_v are quite different. The reaction rates of C₃H₆ oxidation over CuO and CuO-O_v catalysts at various temperatures were calculated (Fig. 3c). It can be found that the reaction rate of CuO-O_v is obviously higher than CuO at all temperature. The apparent activation energy of CuO-O_v (111) and (110) is also calculated (Fig. S30 in Supporting information). The apparent activation energy of (010) facet (39 kJ/mol) is lower than the (111) facet (57 kJ/mol) and the (110) facet (56 kJ/mol), making (010) easier to be activated.

  Figure 3

  

  Figure 3.  (a) Light-off curve for as-prepared CuO, CuO-O_v and Commercial CuO (CuO-Com) catalysts for C₃H₆ oxidation. (b) Arrhenius plots for the catalysts for C₃H₆ oxidation. (c) Calculated reaction rate for C₃H₆ oxidation. (d) Upper: C₃H₆ oxidation catalyzed by CuO and CuO-O_v for 100 h stability test. Lower: C₃H₆ oxidation with the presence of 10 vol% H₂O for 24 h stability test. The reaction conditions: [C₃H₆] = 1 vol%, [O₂] = 20 vol%, N₂ balanced. (e) Light-off curve for as-prepared CuO, CuO-O_v and CuO-Com catalysts for toluene oxidation. (f) Comparison of T_50% and T_90% between CuO, CuO-O_v and CuO-Com catalysts. Reaction conditions: [toluene] = 100 ppm, [O₂] = 20 vol%, N₂ balanced.

  DownLoad: Full-Size Img PowerPoint

  Moreover, the time-on-stream test on CuO-O_v catalysts (top part in Fig. 3d) shows there is no obvious decrease of CuO-O_v in conversion after 100 h reaction. The used CuO-O_v catalyst was characterized by XRD, XPS and EPR (Fig. S26 in Supporting information). The XRD pattern of the used catalyst shows the bulk crystal structure remains intact. The Cu 2p XPS spectrum reveals a stable Cu⁺ ratio (39.6% vs. 40.8%) compared to the fresh catalyst, indicating no significant sintering or valence change of Cu. More importantly, the EPR signal at g = 2.00, although slightly attenuated, remained pronounced, confirming that a high density of O_v was retained after 100 h of reaction in the presence of oxygen and C₃H₆. This robust stability of O_v is a key factor underpinning the excellent long-term performance of the catalyst. In addition, H₂O is a nonnegligible factor which can deactivate the catalysts, the influence of H₂O had been considered for all samples [41]. A time-on-stream test with 10 vol% H₂O were evaluated (bottom part in Fig. 3d). Under an ongoing test of 24 h, almost no decrease can be observed with 10 vol% H₂O over CuO-O_v sample. Activity evaluation of aromatic VOCs oxidation (such as toluene) over CuO and CuO-O_v was also performed. As shown in Fig. 3e, the CuO-O_v exhibits better catalytic activity for toluene oxidation than that of CuO. The T_90% of CuO-O_v is 246 ℃ for toluene oxidation, which is 52 ℃ and 112 ℃ lower than that of CuO (298 ℃) and commercial CuO (358 ℃), respectively (Fig. 3f). Activity evaluation results suggest the LSO in CuO-O_v may pave a universal pathway for the C-H bond activation in VOCs oxidation process.

  In situ DRIFT spectra were performed to further determine the reactive intermediate species and the possible reaction pathways over CuO and CuO-O_v catalysts. In Fig. 4a, under C₃H₆ + O₂ atmosphere on CuO surface, the intensive bands at 1629, 1415 and 1368 cm^-1 can be assigned to propylene glycol species [42,43]. At 200 ℃, the band at 1542 cm^-1 (C-O stretching) attributed to tridentate carbonate [42,43]. The band at 2358 and 2315 cm^-1 are assigned to gaseous CO₂ [10]. With the temperature increased to 320 ℃, propylene glycol and tridentate carbonate species gradually disappeared. Afterwards, the band at 1063 cm^-1 (symmetric O-C-O stretching) assigned to bidentate carbonate becoming more intensive with the temperature raised to 380 ℃. The change of band intensities reveals the propylene glycol species are the pivotal intermediates during C₃H₆ oxidation process on CuO surface. For CuO-O_v, the band position and intensity are quite different with CuO sample (Fig. 4b). The band at 1283 cm^-1 is attributed to ν_{as(CH2=CH-O-)} mode of enolate species [20]. The band at 1432 cm^-1 is attributed to ethylene species. Besides, the bands at 1350 and 1571 cm^-1 are assigned to asymmetric and symmetric stretching modes of carboxylate (COO^-) groups, respectively [44]. With the increase of temperature, the band intensity of carboxylate becomes stronger obviously, while the intensity of ethylene (1432 cm^-1) becomes weak, indicating the ethylene species may transform to carboxylate species on CuO-O_v surface, which is a vital active intermediate in catalytic oxidation of C₃H₆.

  Figure 4

  

  Figure 4.  In situ DRIFTS spectra of C₃H₆ + O₂ chemisorption on (a) CuO, (b) CuO-O_v. (c) DFT computed reaction mechanism of propene oxidation on CuO and CuO-O_v (010) surface.

  DownLoad: Full-Size Img PowerPoint

  Comparing Figs. 4a and b, the disparate intermediates strongly suggesting there are different reaction pathways of CuO and CuO-O_v in C₃H₆ catalytic oxidation. For CuO, the propylene glycol species can transform to carbonate species on surface. While for CuO-O_v, molecular oxygen is preferably activated at the oxygen vacancy to generate LSO species, which provides adsorption sites and promote propene oxidize to ethylene intermediates and further shift to carboxylate. The C₃H₆ in situ DRIFT spectra in Fig. S21 (Supporting information) also indicate that the LSO species are the key adsorbed sites for C₃H₆. Therefore, CuO follows the propylene glycol pathway and CuO-O_v goes with the ethylene pathway. In addition, CuO-O_v catalysts showed a more intensive CO₂ peak than CuO in the reaction process, highlighting better surface reactivity of CuO-O_v.

  To validate the integral role of LSO in propylene oxidation over CuO-O_v, online mass spectrometry of tail gas was performed to identify reaction intermediates (Fig. S22 in Supporting information). Distinct temporal evolution patterns were observed: The CuO reaction pathway proceeds via acetone oxidation to hydroxyacetone, followed by propylene glycol formation as the key intermediate. Conversely, CuO-O_v exhibits divergent selectivity with predominant propanal formation, undergoing rapid C–C cleavage to ethylene as the key intermediate. These experiments motivated DFT simulations of the LSO-mediated reaction pathway on CuO-O_v (010) in Fig. 4c. Overall, the calculation results declared the C₃H₆ can easily react with LSO species on CuO-O_v with a relative low barrier energy (0.77 eV). Nevertheless, the reaction energy barrier of C₃H₆ catalytic oxidation on CuO is higher (1.81 eV), indicating that the oxidation reaction is detrimental to occur, which is consistent with the light-off curve in Fig. 3a. For CuO, when C₃H₆ is adsorbed on the surface of CuO, the C-H bond will cleave first. The C atom will react with the lattice oxygen on CuO catalyst surface to generate acetone species (step i, blue line). Subsequently, O₂ react with two C atoms, causing the O-O bond to break and forming -OH (step ii). Then, the H atom migrate to O atom to form the configuration of propylene glycol, which is consistent with reaction intermediate shown in situ DRIFTS in Fig. 4a. The DFT results verified that the intermediate is propylene glycol during the oxidation of C₃H₆ on CuO surface. As for CuO-O_v, O₂ could approach to the oxygen vacancy site and bonded with Cu atoms to form LSO species with formation energy of -1.83 eV (step i in rose red line). Subsequently, the C₃H₆ adsorb on the LSO species causing the cleavage of O-O bond and generating propionaldehyde intermediates (step ii). Then, the propionaldehyde intermediates desorb from the surface of CuO-O_v (TS). Finally, the C-O bond collapse, forming ethylene species, which is consistent with in situ DRIFTS observations. All the elementary steps in the calculating process of CuO-O_v are exothermic and the barrier is lower (0.77 eV) than CuO (1.81 eV). Therefore, the LSO species generated on oxygen vacancy of CuO-O_v promoted C-H bond cleavage, which further facilities propene oxidation.

  In summary, the LSO can be generated exclusively on the oxygen vacancy sites of (010) facets in CuO-O_v nanosheets and the uniform electron charge between two oxygen atoms significantly contribute to the high activity of LSO species. In terms of theoretical and experimental results, the LSO species consist of two oxygen atoms in a side-on adsorption form to occupy the position of surface lattice oxygen and with comparable electron (0.22 and 0.20 eV). The electrons accepted by two oxygen atoms in LSO makes the O-O bond more facile to be activated than the O-O bond in ordinary superoxide species. In addition, the LSO possesses moderate activation ability, which is in between superoxide species and lattice oxygen, rendering an adequate oxygen activation process. The activated LSO acts as new adsorption site for C₃H₆ molecule, which can further promote the dissociation of C-H bond. The CuO-O_v catalyst with LSO species went through ethylene pathway instead of propylene glycol pathway, exhibiting much better C₃H₆ oxidation activity than catalyst without LSO (T_90% = 229 ℃ for C₃H₆). The discovery of a novel reactive oxygen species, with a defined activation energy, paves the way for rational catalyst design leveraging ROS in oxidation reactions.

  CRediT authorship contribution statement

  Wei Chen: Writing – original draft, Investigation. Yihang Hua: Methodology. Juan Zheng: Formal analysis. Li Li: Formal analysis. Ruohan Yu: Visualization. Xiaoping Chen: Visualization. Fangfang Xia: Visualization, Supervision. Xiaoli Zhou: Investigation. Guoli Liu: Visualization, Methodology. Yarong Fang: Visualization, Methodology. Chuanqi Pan: Visualization, Methodology. Weihao Li: Visualization, Methodology. Jinsong Wu: Visualization, Software. Zhu Luo: Supervision, Resources, Funding acquisition, Conceptualization. Yanbing Guo: Software, Investigation.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  Acknowledgments

  This project was supported by the National Natural Science Foundation of China (Nos. 22476060, 22376075, 22376074). Along with the Cultivation Program of Wuhan Institute of Photochemistry and Technology (Nos. GHY2023KF002, GHY2023KF010). The Program of Introducing Talents of Discipline to Universities of China (also known as the 111 program, No. B17019), also played a crucial role in funding this research. Furthermore, the authors acknowledge the support from Longyan Major Science and Technology Project (No. 2022LYF1006), Wuhan Knowledge Innovation Special Basic Research Project (No. 2023020201010123), the Central China Normal University project (Nos. CCNU24JCPT017, 30106220497), Wuhan Institute of Photochemical Technology incubation project (No. GHY2023FH03).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2026.112629.

  
  
• 1. [1]

     F. Saaoud, C.D. IV, Y. Shao, et al., Pharmacol. Ther. 220 (2021) 107715. doi: 10.1016/j.pharmthera.2020.107715

  2. [2]

     C. Pan, C. Wang, X. Zhao, et al., J. Am. Chem. Soc. 144 (2022) 4942–4951. doi: 10.1021/jacs.1c12772

  3. [3]

     M.L. Reshi, Y. Su, J.R. Hong, Int. J. Cell Biol. 1 (2014) 467452.

  4. [4]

     Y. Fang, Q. Zhang, H. Zhang, et al., Angew. Chem. Int. Ed. 134 (2022) e202212273. doi: 10.1002/ange.202212273

  5. [5]

     Y. Zhu, J. Wang, S.B. Patel, et al., Proc. Natl. Acad. Sci. U. S. A. 120 (2023) e2215189120. doi: 10.1073/pnas.2215189120

  6. [6]

     D. Zhang, P. Wang, J. Wang, et al., Proc. Natl. Acad. Sci. U. S A. 118 (2021) e2114729118. doi: 10.1073/pnas.2114729118

  7. [7]

     Y. Xiao, H. Li, K. Xie, Angew. Chem. Int. Ed. 60 (2021) 5240–5244. doi: 10.1002/anie.202013633

  8. [8]

     H. Shang, M. Li, H. Li, et al., Environ. Sci. Technol. 53 (2019) 6444–6453. doi: 10.1021/acs.est.8b07322

  9. [9]

     H. Zhang, G.P. Hatzis, C.E. Moore, et al., J. Am. Chem. Soc. 141 (2019) 9516–9520. doi: 10.1021/jacs.9b04215

  10. [10]

      J. Yang, S. Hu, Y. Fang, et al., ACS Catal. 9 (2019) 9751–9763. doi: 10.1021/acscatal.9b02408

  11. [11]

      H. Li, F. Qin, Z. Yang, et al., J. Am. Chem. Soc. 139 (2017) 3513–3521. doi: 10.1021/jacs.6b12850

  12. [12]

      C. Zhang, Y. Yu, Michael E. Grass, et al., J. Am. Chem. Soc. 135 (2013) 11572–11579. doi: 10.1021/ja402604u

  13. [13]

      X. Xiao, Z. Ruan, Q. Li, et al., Adv. Mater. 34 (2022) 2200612. doi: 10.1002/adma.202200612

  14. [14]

      Y. Deng, Z. Zhou, H. Zeng, et al., Appl. Catal. B 320 (2023) 121942. doi: 10.1016/j.apcatb.2022.121942

  15. [15]

      N. Zhang, X. Feng, D. Rao, et al., Nat. Commun. 11 (2020) 4066. doi: 10.1038/s41467-020-17934-7

  16. [16]

      J. Han, H. Wang, Y. Wang, et al., Angew. Chem. Int. Ed. 63 (2024) e202405839. doi: 10.1002/anie.202405839

  17. [17]

      Z. Su, W. Yang, C. Wang, et al., Environ. Sci. Technol. 54 (2020) 12684–12692. doi: 10.1021/acs.est.0c03981

  18. [18]

      X. Chen, J. Chen, H. Chen, et al., Nat. Commun. 14 (2023) 751. doi: 10.1038/s41467-023-36263-z

  19. [19]

      H. Liu, Z. Fan, C. Sun, et al., Appl. Catal. B 244 (2019) 671–683. doi: 10.1016/j.apcatb.2018.12.001

  20. [20]

      Y. Fang, L. Li, J. Yang, et al., Environ. Sci. Technol. 54 (2020) 15476–15488. doi: 10.1021/acs.est.0c05845

  21. [21]

      W. Zhang, M. Jiang, S. Yang, et al., Nano Res. Energy 1 (2022) 9120033. doi: 10.26599/NRE.2022.9120033

  22. [22]

      D.A. Fischer, G.G. Cohen, N.J. Shevchik, J. Phys. F: Metal Phys. 10 (1980) L139–L142. doi: 10.1088/0305-4608/10/5/004

  23. [23]

      A.M. Carrillo, J.G. Carriazo, Appl. Catal. B 164 (2015) 443–452. doi: 10.1016/j.apcatb.2014.09.027

  24. [24]

      Y. Shi, J. Li, C. Mao, et al., Nat. Commun. 12 (2021) 5923. doi: 10.1038/s41467-021-26219-6

  25. [25]

      F. Frati, M. Hunault, F.M.F. de Groot, Chem. Rev. 120 (2020) 4056–4110. doi: 10.1021/acs.chemrev.9b00439

  26. [26]

      M.W. Ruckman, J. Chen, S. Qiu, et al., Phys. Rev. Lett. 67 (1991) 2533–2536. doi: 10.1103/PhysRevLett.67.2533

  27. [27]

      L. Karvonen, M. Valkeapää, R. Liu, et al., Chem. Mater. 22 (2009) 70–76.

  28. [28]

      Y. Zhao, B. Teng, X. Wen, et al., J. Phys. Chem. C 116 (2012) 15986–15991. doi: 10.1021/jp3016326

  29. [29]

      W. Yang, Y. Zhu, F. You, et al., Appl. Catal. B 233 (2018) 184–193. doi: 10.1016/j.apcatb.2018.03.107

  30. [30]

      C. Sun, W. Chen, X. Jia, et al., Chin. J. Catal. 42 (2021) 417–430. doi: 10.1016/S1872-2067(20)63666-X

  31. [31]

      W. Chen, J. Zheng, Y. Fang, et al., Environ. Sci. Technol. 58 (2024) 3041–3053. doi: 10.1021/acs.est.3c08380

  32. [32]

      J. Wang, G. Ni, W. Liao, et al., Angew. Chem. Int. Ed. 62 (2023) e202217026. doi: 10.1002/anie.202217026

  33. [33]

      Y. Jiang, J. Shan, P. Wang, et al., ACS Catal. 13 (2023) 3101–3108. doi: 10.1021/acscatal.3c00123

  34. [34]

      K. Wang, Y. Li, G. Zhang, et al., Appl. Catal. B 240 (2019) 39–49. doi: 10.1016/j.apcatb.2018.08.063

  35. [35]

      T. Luo, H. Zhang, Y. Chen, et al., J. Am. Chem. Soc. 146 (2024) 35057–35063. doi: 10.1021/jacs.4c06154

  36. [36]

      M. Jiang, K. Yang, H. Yu, L. Yao, J. Phys. Chem. Solids 152 (2021) 109957. doi: 10.1016/j.jpcs.2021.109957

  37. [37]

      P. Zhang, L. Liu, X. He, et al., J. Am. Chem. Soc. 141 (2019) 6263–6270. doi: 10.1021/jacs.8b13568

  38. [38]

      J. Ye, B. Zhu, B. Cheng, et al., Adv. Funct. Mater. 32 (2022) 2110423. doi: 10.1002/adfm.202110423

  39. [39]

      Q. Zhang, X.X. Qin, F.P. Duan-Mu, et al., Angew. Chem. Int. Ed. 57 (2018) 9351–9356. doi: 10.1002/anie.201804319

  40. [40]

      H. Gu, X. Liu, X. Liu, et al., Nat. Commun. 12 (2021) 5422. doi: 10.1038/s41467-021-25726-w

  41. [41]

      W. Luo, K. Liu, T. Luo, et al., J. Am. Chem. Soc. 147 (2025) 7391–7399. doi: 10.1021/jacs.4c15280

  42. [42]

      Y. Yan, Y. Dai, H. He, et al., Appl. Catal. B 196 (2016) 108–116. doi: 10.1016/j.apcatb.2016.05.016

  43. [43]

      C. Rodríguez, S. Moreno, R. Molina, Appl. Surf. Sci. 485 (2019) 293–303. doi: 10.1016/j.apsusc.2019.04.183

  44. [44]

      L. Zhang, J. Liang, L. Yue, et al., Nano Res. Energy 1 (2022) 9120028. doi: 10.26599/NRE.2022.9120028

• 

  Figure 1  (a) ABF-STEM image of CuO-O_v. (b, c) Intensity profile corresponding to the green and yellow dashed line in (a), respectively. (d) EPR pattern under vacuum. (e) Structure model of CuO-O_v nanosheet. (f) O₂-TPD profiles of CuO and CuO-O_v. (g) Oxygen K-edge spectra of the CuO and CuO-O_v.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) The configuration of O₂ and the difference charge density of O₂ adsorbed on CuO-O_v surface through theoretical calculation. (b) The configuration of O₂ and the difference charge density of O₂ adsorbed on CuO surface through theoretical calculation. (c) NAP-XPS of O 1s spectra from CuO-O_v with the Ar etching. The arrows represent the order in which the catalyst is treated with O₂ or Ar etching. (d) NAP-XPS of O 1s spectra from CuO with the Ar etching. (e) Crystal orbital hamilton population (COHP) of O-O bond in LSO and O₂^- species. (f) O K-edge XANES of CuO-O_v with different crystal facets. (g) The electronic local function (ELF) of LSO. (h) The electronic local function (ELF) of oxygen adsorbed on CuO. (i) COHP of Cu-O bond (Cu and O_I atom in LSO, Cu and lattice oxygen in CuO). (j) Ab initio molecular dynamics (AIMD) simulation for LSO desorption on CuO at 473 K.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) Light-off curve for as-prepared CuO, CuO-O_v and Commercial CuO (CuO-Com) catalysts for C₃H₆ oxidation. (b) Arrhenius plots for the catalysts for C₃H₆ oxidation. (c) Calculated reaction rate for C₃H₆ oxidation. (d) Upper: C₃H₆ oxidation catalyzed by CuO and CuO-O_v for 100 h stability test. Lower: C₃H₆ oxidation with the presence of 10 vol% H₂O for 24 h stability test. The reaction conditions: [C₃H₆] = 1 vol%, [O₂] = 20 vol%, N₂ balanced. (e) Light-off curve for as-prepared CuO, CuO-O_v and CuO-Com catalysts for toluene oxidation. (f) Comparison of T_50% and T_90% between CuO, CuO-O_v and CuO-Com catalysts. Reaction conditions: [toluene] = 100 ppm, [O₂] = 20 vol%, N₂ balanced.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  In situ DRIFTS spectra of C₃H₆ + O₂ chemisorption on (a) CuO, (b) CuO-O_v. (c) DFT computed reaction mechanism of propene oxidation on CuO and CuO-O_v (010) surface.

  下载: 全尺寸图片 幻灯片

  Table 1.  The calculated change in Bader charge on CuO and CuO-O_v. And the comparison of different ROS in references.

  ROS	Bond length (Å)	Adsorbed energy (eV)	Barder charge (eV)	Ref.
  LSO	1.30	-1.83	OI: 0.22 OII: 0.20	This work
  Peroxide O22-	1.36	-1.99	OI: 0.30 OII: 0.39	This work
  Superoxide O2-	1.31	-1.56	OI: 0.14 OII: 0.39	This work
  Superoxide O2-	1.32	-1.52	-	[37]
  Peroxide O22-	1.43	-2.80	-	[28]
  O-	1.41	-2.71	0.646	[38]
  O2-Pt-H2WO4	1.37	-1.74	0.22	[39]
  Fe(O=O)Fe	-	-2.00	OI: 0.06 OII: 0.37	[40]

  下载: 导出CSV
•